<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:googleplay="http://www.google.com/schemas/play-podcasts/1.0"><channel><title><![CDATA[Rovelli Research]]></title><description><![CDATA[Rovelli Research]]></description><link>https://www.rovelliresearch.com</link><image><url>https://www.rovelliresearch.com/img/substack.png</url><title>Rovelli Research</title><link>https://www.rovelliresearch.com</link></image><generator>Substack</generator><lastBuildDate>Tue, 17 Nov 2026 09:30:07 GMT</lastBuildDate><atom:link href="https://www.rovelliresearch.com/feed" rel="self" type="application/rss+xml"/><copyright><![CDATA[Dogycix]]></copyright><language><![CDATA[en]]></language><webMaster><![CDATA[rovelliresearch@substack.com]]></webMaster><itunes:owner><itunes:email><![CDATA[rovelliresearch@substack.com]]></itunes:email><itunes:name><![CDATA[Alessandro Rovelli]]></itunes:name></itunes:owner><itunes:author><![CDATA[Alessandro Rovelli]]></itunes:author><googleplay:owner><![CDATA[rovelliresearch@substack.com]]></googleplay:owner><googleplay:email><![CDATA[rovelliresearch@substack.com]]></googleplay:email><googleplay:author><![CDATA[Alessandro Rovelli]]></googleplay:author><itunes:block><![CDATA[Yes]]></itunes:block><item><title><![CDATA[Cerba. When the French State Controls the Price]]></title><description><![CDATA[&#8364;5 billion of debt, a shrinking price per test and a restructuring in which creditors do not control the most important variable.]]></description><link>https://www.rovelliresearch.com/p/cerba-when-the-french-state-controls</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/cerba-when-the-french-state-controls</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 17 Nov 2026 06:15:30 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/f9f564d9-3de0-4a81-ac27-4c41f2435f06_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>The most important development in Cerba HealthCare&#8217;s restructuring did not happen in a meeting with creditors. It happened on 2 November, when France&#8217;s national health insurance fund, complementary insurers and four laboratory unions signed a three-year agreement on laboratory spending.</p><p>The agreement sets limits on how much can be spent on routine laboratory tests and creates a committee with the power to adjust tariffs to keep spending within those limits. No lenders signed it. Yet it could have a greater impact on their recovery than anything currently being discussed in the debt negotiations.</p><p>Cerba is one of France&#8217;s largest medical laboratory groups. In early September, it entered a court-supervised conciliation process to negotiate with creditors over approximately &#8364;5 billion of debt. Most attention has focused on the restructuring itself. For lenders, however, the November agreement may be just as important because it directly affects the company&#8217;s ability to generate cash and repay its debt.</p><h2>The revenue problem</h2><p>Under the previous agreement, which covered 2024 to 2026, laboratory spending is expected to reach approximately &#8364;3.9 billion this year.</p><p>The new agreement limits annual growth in spending on routine tests to 1.1%. The laboratory unions estimate that the number of tests will naturally increase by 3% to 3.5% a year. The difference will have to be absorbed through lower prices.</p><p>A monitoring committee can adjust both the letter key B and individual test tariffs to keep spending within the agreed budget.</p><p>There are also additional savings of at least &#8364;30 million in 2027 and &#8364;10 million in each following year. A separate &#8364;100 million budget over three years is reserved for new tests and sits outside these restrictions.</p><p>The implications for creditors are significant. Cerba may continue to perform more tests, but that does not mean its revenue will grow at the same rate. The state effectively controls the price of its main product.</p><p>For a lender assessing the business, this changes the assumptions behind any cash-flow forecast. Volume growth alone is not enough. The more important question is how much revenue the company can generate from each test after the next round of tariff reductions.</p><h2>Nearly &#8364;5 billion of debt and limited repayment capacity</h2><p>EQT acquired Cerba from Partners Group in 2021 for approximately $5.3 billion, with PSP Investments taking a minority stake.</p><p>At 30 June, net debt exceeded &#8364;4.9 billion, equivalent to around nine times trailing EBITDA. Annualising second-quarter EBITDA alone would put leverage closer to ten times.</p><p>Second-quarter revenue was approximately &#8364;469 million, up 2% year on year. EBITDA exceeded &#8364;122 million, an increase of 12%.</p><p>The improvement in profitability is encouraging, but it does not solve the debt problem. The company needs to generate substantially more cash than it currently does to reduce a debt burden of this size.</p><p>Consider the repayment arithmetic.</p><p>If trailing EBITDA is approximately &#8364;540 million, consistent with the reported nine-times leverage ratio, and the average interest cost on &#8364;4.9 billion of debt is 7%, annual interest expense would be around &#8364;340 million. This interest assumption is my estimate, not a disclosed figure.</p><p>Assuming capital expenditure of 5% of revenue, or approximately &#8364;94 million a year, the company would have just over &#8364;100 million left before tax and working-capital movements.</p><p>At that rate, repaying &#8364;4.9 billion of debt would take around 45 years, assuming no changes in operating performance, prices or interest costs.</p><p>That is not a credible deleveraging plan. And the new agreement makes the revenue assumptions behind it less certain.</p><p>Cerba had access to approximately &#8364;108 million of liquidity in June. EQT provided &#8364;100 million of additional support earlier this year, but as a loan rather than equity. The period during which interest payments could be deferred was also extended into November.</p><p>In July, S&amp;P described the debt as unsustainable and expected a default within six months.</p><p>For lenders, the question is no longer simply whether Cerba can continue operating. It is how much of the existing debt can realistically be recovered.</p><h2>Three issues lenders should examine</h2><p>There are three particular risks that deserve attention when assessing Cerba&#8217;s debt and the potential recovery for creditors.</p><h3>1. Liquidity buys time, not debt reduction</h3><p>The &#8364;108 million of available liquidity is small compared with &#8364;4.9 billion of debt.</p><p>It gives the company time to negotiate, but does not repair its balance sheet. The &#8364;100 million shareholder loan provides additional support, but it also adds debt rather than reducing leverage through new equity.</p><p>The extension of the interest grace period is another temporary measure. It delays cash payments but does not resolve the underlying repayment problem.</p><p>For a lender, it is important to distinguish between a company that has enough liquidity to survive a restructuring and one that can generate enough cash to repay its obligations.</p><p>Cerba currently has the former, not necessarily the latter.</p><h3>2. Security and recovery are not the same thing</h3><p>A &#8364;28 million portion of the revolving credit facility was sold in September, while restructuring discussions were under way.</p><p>That is worth watching. A lender selling part of its exposure during negotiations may be reducing its risk rather than waiting for the outcome.</p><p>There is also exposure through less visible financial structures. In November 2025, 9fin reported that an Apollo-held leveraged loan significant risk transfer included Cerba among its stressed borrowers.</p><p>This is a reminder that Cerba&#8217;s financial difficulties may affect investors who do not hold its bonds or loans directly.</p><p>For lenders assessing the capital structure, the important questions are who actually holds the credit risk, where the debt ranks and what protection is available if the restructuring fails.</p><p>A diversified portfolio does not necessarily provide much protection when one of its underlying borrowers is already under severe financial pressure.</p><h3>3. The assets supporting the debt are difficult to take control of</h3><p>There is another complication in Cerba&#8217;s ownership structure.</p><p>The debt sits at the holding company level, while the laboratories operate through regulated subsidiaries. French law requires medical laboratories to be majority owned by the biologists who work in them.</p><p>This creates a gap between the location of the debt and the businesses generating the cash.</p><p>Lenders may be able to take control of the holding company through a debt-for-equity restructuring. But that does not automatically give them control of the operating laboratories.</p><p>The biologists remain important shareholders and are essential to the group&#8217;s operations. A restructuring that ignores their position could leave new owners with control of the holding company but limited influence over the people running the underlying business.</p><p>This is a particularly important consideration when assessing security and recovery. The value of the assets on paper is not necessarily the value lenders can realise in practice.</p><p>The issue is not simply whether creditors have security. It is whether that security gives them effective control over the businesses and cash flows supporting their claims.</p><h2>The political dimension</h2><p>Cerba&#8217;s restructuring also has a political dimension.</p><p>A parliamentary commission reported in June on what it described as the acquisition of French productive businesses by speculative investment funds. It recommended tighter rules for funds it believed were operating without sufficient oversight.</p><p>Laboratory employees went on strike in May over pay. Cerballiance, Cerba&#8217;s French laboratory network, also suffered a cyberattack at the end of March.</p><p>The restructuring is therefore taking place against a difficult public backdrop. It involves a heavily indebted healthcare group, private equity ownership, employees concerned about their conditions and a government directly controlling the spending framework.</p><p>This makes a negotiated outcome more likely than a straightforward creditor-led restructuring.</p><p>The government needs the laboratories to continue operating. The biologists need to protect their businesses. Creditors need to recover as much as possible. These interests do not necessarily point in the same direction.</p><p>For lenders, the final recovery will depend not only on the financial negotiations but also on what the government and the operating professionals are prepared to accept.</p><h2>The sector is still financeable</h2><p>I would not draw the conclusion that French medical laboratories are no longer attractive lending opportunities.</p><p>Biogroup refinanced &#8364;2.7 billion of debt in May, although it had to pay a high price to secure the financing.</p><p>The sector can still attract capital. The question is how much debt a business can support, at what interest cost and with what protection for lenders.</p><p>Cerba shows what happens when the original financing assumptions no longer hold. High leverage, limited cash generation, regulated pricing and a complicated ownership structure can leave creditors with few good options.</p><h2>What this means for credit investors</h2><p>Cerba raises three questions that apply well beyond the healthcare sector.</p><p><strong>First, who controls the revenue?</strong> When a company cannot freely set the price of its products or services, lenders need to understand the regulatory framework and how frequently prices can change. Revenue growth based on higher volumes may not translate into higher cash flow.</p><p><strong>Second, where are the assets and cash flows?</strong> A lender needs to understand the relationship between the borrowing entity, the operating companies and the assets supporting the debt. Security at holding company level may offer less protection than expected if the underlying businesses are difficult to control or transfer.</p><p><strong>Third, what is the sponsor actually contributing?</strong> Additional shareholder loans provide liquidity but do not reduce leverage in the same way as new equity. The distinction is important when assessing sponsor support and the likelihood of further capital injections.</p><p>In Cerba&#8217;s case, the answers are uncomfortable. Revenue is subject to government controls, the operating laboratories are partly owned by the professionals running them, and the sponsor&#8217;s recent &#8364;100 million contribution was a loan rather than equity.</p><p>Together, these factors make the recovery difficult to assess and argue against assuming that creditors will be repaid in full.</p><h2>What could change my view?</h2><p>Three developments would improve the recovery outlook.</p><ul><li><p><strong>A better pricing agreement:</strong> An increase in the 1.1% annual spending allowance would support revenue and cash generation. I consider this unlikely given the constraints on the 2027 budget.</p></li><li><p><strong>Further asset sales:</strong> The sale of Lifeanalytics to Eurofins in September shows that disposals are possible. More substantial sales, with proceeds used to reduce debt rather than simply support liquidity, would help.</p></li><li><p><strong>New equity from EQT:</strong> A fresh equity contribution would demonstrate stronger sponsor support and directly reduce the debt burden.</p></li></ul><p>Any one of these would improve the position of creditors. A combination would make a more meaningful difference.</p><h2>My credit view</h2><p>The most likely outcome is a negotiated restructuring rather than a full repayment of existing debt.</p><p>The conciliation process normally runs for four months and can be extended once. The current timetable therefore points towards a resolution around the turn of the year or, at the latest, early February.</p><p>The final recovery will depend on the proceeds from asset sales, the amount of new capital available and the value investors place on a business whose prices are being reduced by government policy.</p><p>I would not assume full recovery anywhere in the capital structure.</p><p>Cerba is a useful reminder that lending risk is not just about leverage and interest coverage. It is also about who controls the revenue, where the assets sit and what lenders can actually recover when a company runs into difficulty.</p><p>For lenders considering French healthcare financing, these questions deserve as much attention as the financial model itself.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Why $1.2 Trillion of Private Credit Depends on Insurance Rules]]></title><description><![CDATA[Private credit's biggest risk may not be borrower defaults, but the insurance rules governing its largest buyers. A regulatory shift could trigger billions in reclassifications, higher capital require]]></description><link>https://www.rovelliresearch.com/p/the-12-trillion-private-credit-buyer</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/the-12-trillion-private-credit-buyer</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Thu, 12 Nov 2026 05:48:27 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/c8889ed2-7ae5-41d3-a789-570bf4b26e92_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Most of the debate about private credit is being conducted by people who don&#8217;t own much of it. The biggest owners are life insurers, and they have a different set of concerns from the rest of the market.</p><p>US insurers held about $1.2 trillion of private credit at the end of 2025, according to the National Association of Insurance Commissioners (NAIC). That&#8217;s roughly 13% of their cash and invested assets and 21% of their bond holdings. Privately placed securities now account for 48.4% of life insurers&#8217; bond portfolios, up from 37.4% five years ago. The NAIC has singled out $544 billion of privately rated bonds, business development companies and private credit funds for particular attention. Moody&#8217;s puts US life insurers&#8217; private and illiquid bond holdings at $807 billion, with the ten largest holders accounting for 44%.</p><p>A fund reports the value of its investments. An insurer also has to report how much capital it needs to support them. Those are two very different exercises. And the rules governing the second could have a bigger effect on private credit prices than the next round of borrower defaults.</p><p>There are three areas to watch: how insurers classify their investments, how those investments are rated, and where the risk ends up after reinsurance.</p><p><strong>First, classification.</strong></p><p>The NAIC&#8217;s new principles-based bond definition took effect in January 2025. It changed the accounting rules used to decide whether an investment qualifies as an issuer credit obligation, an asset-backed security or neither. In other words, it determines whether an asset can be treated as a bond at all.</p><p>The new definition looks at the economic substance of an investment rather than its legal form. Insurers therefore had to check whether their existing holdings still qualified as bonds under the new rules. S&amp;P Global Market Intelligence estimates that nearly $15.4 billion of assets were moved out of the main bond reporting category (Schedule D, Part 1) and into other investment categories. In a more specific group of reclassifications, around 650 bonds worth $1.81 billion were affected.</p><p>These amounts are small compared with the $1.2 trillion total. But the underlying mechanism is important. An investment can receive different regulatory treatment even when its borrower has not missed a single payment. Reclassification does not automatically create an accounting loss, but it can change the capital an insurer needs to hold and how the investment appears on its balance sheet.</p><p><strong>Second, ratings.</strong></p><p>Private credit ratings are not just opinions about the likelihood of repayment. For insurers, they can also determine how much capital an investment requires.</p><p>The NAIC is tightening the information that must accompany private letter ratings and building a framework to assess whether rating agencies&#8217; methodologies and rating mappings remain suitable for regulatory purposes.</p><p>At the same time, capital requirements for collateralized loan obligations (CLOs) are moving away from simple reliance on external ratings towards direct actuarial modelling. Residual interests in structured securities already carry life risk-based capital factors of 45%, while revised CLO capital factors are due to take effect at the end of 2026.</p><p>Regulators are increasingly looking beyond external ratings and seeking capital requirements that better reflect the underlying risk of each investment.</p><p>This creates an interesting problem for private credit. When an agency&#8217;s private letter rating determines the capital treatment of a billion dollars of investments, that rating becomes part of the asset&#8217;s economics. And a private rating is a negotiated document, not an independent market observation. Regulators increasingly want to know how the number was reached.</p><p><strong>Third, reinsurance.</strong></p><p>This is where things get more complicated, because the insurer holding the risk and the entity ultimately exposed to the losses are not always the same.</p><p>Life insurers use reinsurance to transfer some of their obligations to other companies, including offshore affiliates and captive insurers. This can improve capital efficiency, but it also makes it harder for regulators to see where the underlying risks actually sit.</p><p>Actuarial guidelines require regulators to check whether insurers have enough assets and reserves to cover their obligations, including when risks have been transferred to offshore affiliates or other reinsurers. They also give greater attention to complex investments, illiquid assets and valuations that are difficult to verify.</p><p>US life insurers reported $321 billion of investments in affiliated companies, equivalent to about 5% of general account cash and invested assets.</p><p>Regulatory scrutiny is increasing. In May 2026, the US Treasury brought together state insurance commissioners to discuss private credit, offshore reinsurance, private letter ratings and the changing business models of life insurers.</p><p>The message is becoming difficult to miss. Regulators want to understand where the risk actually sits, not simply where the reinsurance agreement says it has gone.</p><p>Put these three developments together and the usual private credit debate starts to look rather different.</p><p>The market has spent the past two years discussing defaults, loan extensions and payment-in-kind interest. These are genuine credit risks. But for the largest holders of private credit, there is another set of questions: Does the investment qualify as a bond? Who provided its rating, and can that rating be defended? And which balance sheet ultimately carries the risk?</p><p><strong>Europe has a similar story, but the numbers and the exposures look different.</strong></p><p>European insurers' private credit holdings reached approximately &#8364;594 billion in the second quarter of 2025, equivalent to 5.8% of total assets, up from &#8364;332 billion, or 3.9%, at the end of 2016.</p><p>But the headline number hides some important differences between countries. German insurers hold roughly &#8364;92 billion of non-listed corporate bonds, compared with about &#8364;28 billion for French insurers. Much of the difference comes from promissory note lending, or <em>Schuldschein</em>. Mortgages and loans account for around half of German private credit exposure, against about a quarter in France.</p><p>That distinction is important. A Schuldschein issued by a municipal utility is not the same investment as a sponsor-backed unitranche loan. Grouping both under the label of private credit may be convenient, but it can produce misleading conclusions about risk.</p><p>European capital requirements are another area to watch, particularly for securitized and illiquid investments. Solvency II is under review, but the available evidence does not yet establish precisely how the 2027 changes will affect capital charges on these assets. I would not build an investment thesis around a specific increase before the final calibration is clear.</p><p>Still, any insurer whose investment allocation depends on favourable regulatory treatment has an additional risk to consider. The borrower may continue to perform, but the economics of holding its debt could change.</p><p>I spent years buying credit inside an insurance group&#8217;s asset manager. One lesson has stayed with me: insurance portfolios rarely sell because someone suddenly changes their mind about credit. They sell because a ratio moves, an auditor asks an uncomfortable question, or a model produces a number the board wasn&#8217;t expecting.</p><p><strong>Rules make sellers. Credit makes headlines.</strong></p><p>The potential losses look manageable against the capital of the largest holders. I am more concerned about the differences that could emerge between insurers holding exactly the same investments.</p><p>One insurer may classify a loan as a bond, while another does not. One may rely on a private letter rating, while another uses its own model. One may have transferred part of the risk to a captive reinsurer, while another retains it on its own balance sheet.</p><p>The underlying credit is identical. The capital treatment is not. That difference will increasingly influence which insurers want to keep holding these assets and at what price.</p><p>The more damaging scenario is narrower, but worth watching. Suppose the 2027 Solvency II review increases capital charges on illiquid and securitized assets just as reinsurance capacity becomes harder to obtain. Insurers may not be forced to sell, but some could decide that holding these investments is no longer worth the capital they consume.</p><p>That could leave private credit facing a wave of willing sellers in a market with few natural buyers. The price discovery that investors have been demanding would finally arrive, but possibly at the worst moment for anyone who needs today&#8217;s valuations to hold.</p><p>If you manage money for an insurer, which of these issues would you least like to be asked about? I read every reply.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[What Covenant-Lite Actually Costs You]]></title><description><![CDATA[The covenant didn&#8217;t disappear. It moved to a moment when you have far less leverage to use it.]]></description><link>https://www.rovelliresearch.com/p/what-covenant-lite-actually-costs</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/what-covenant-lite-actually-costs</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 10 Nov 2026 06:19:39 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/21f83eeb-c9e0-482b-b7fb-dee47c10b0d2_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Consider a family office which buys a floating-rate loan at 97, yielding a little over 6%. The treasurer asks the obvious question: what is the covenant test, and when does it get tested?</p><p>There may be no regular leverage or interest-cover test for the term loan. No quarterly maintenance-covenant certificate. Instead, the borrower must satisfy financial tests when it takes certain actions, such as borrowing more or paying a dividend. Some deals retain a maintenance test on the revolving credit facility, triggered when drawings exceed a specified threshold.</p><p>Without a regular test, the lender may watch the company&#8217;s financial position deteriorate for months without acquiring any contractual right to intervene. The covenant has stopped being an early-warning system and becomes a condition for taking specific actions.</p><p>What does it actually cost the lender, and does the market compensate for the loss?</p><p>A maintenance covenant is tested periodically, whether or not the borrower wants to do anything. It can catch a deterioration in financial performance and, if the test is breached, create a technical default before the company misses a payment.</p><p>That default can give lenders information, negotiating leverage and an opportunity to demand changes. It can also create problems of its own, particularly when a temporary shock pushes leverage above a contractual limit.</p><p>An incurrence covenant works differently. It is tested when the borrower takes a specified action, such as incurring additional debt, making an acquisition or paying a dividend. It restricts what the borrower can do, rather than requiring it to maintain a financial ratio at all times.</p><p>A company can deteriorate for four consecutive quarters without breaching an incurrence covenant, provided it does not take an action that triggers the test.</p><p>Covenant-lite does not mean covenant-free. Reporting obligations, restrictions on debt and asset transfers, mandatory prepayments and other protections may remain. Some facilities also include springing maintenance covenants.</p><p>The important question is what the documentation allows the borrower to do before lenders can intervene.</p><h2>Four costs</h2><p><strong>Information.</strong> The value of a maintenance covenant is not simply the right to declare a default. It is the opportunity to identify deterioration and start a discussion while the company still has options.</p><p>Quarterly testing can reveal a problem before it becomes a payment default, although the warning period depends on the speed of deterioration and the covenant&#8217;s design. There is no universal six-to-nine-month advantage.</p><p>Without a maintenance test, lenders may first confront the problem when the borrower requests an amendment, a rating agency downgrades the debt or a payment is missed. These events do not always arrive in that order, but they can leave lenders with less time to respond.</p><p><strong>Control.</strong> Covenant-lite loans often give borrowers considerable flexibility while leaving individual lenders with limited power to block changes. Amendment rights commonly depend on the percentage of commitments held by lenders, with certain fundamental rights requiring broader consent.</p><p>This creates a coordination problem. A majority may support a transaction that benefits participating lenders while leaving other creditors worse off.</p><p>Two mechanisms deserve particular attention.</p><p>A replacement of lenders provision allows a borrower, subject to the contractual conditions, to replace a lender that refuses to consent to an amendment. The outgoing lender typically receives repayment at par. These provisions are standard in many European leveraged-loan agreements.</p><p><em>Uptiering</em> is different. It allows a transaction to give certain participating creditors priority over other lenders, potentially through new senior debt or changes to collateral arrangements. Whether such a transaction is permitted depends on the documentation and applicable law.</p><p><strong>Recovery.</strong> A borrower may have significant capacity to transfer assets, incur additional debt or make investments without triggering a default. The size and interaction of these permissions depend on the credit agreement.</p><p>Unrestricted subsidiaries are one example. Assets transferred outside the restricted group may no longer benefit from the same guarantees or security package. Other provisions, including investment baskets and allowances based on an <em>available amount</em>, can provide further flexibility. If the calculation incorporates adjusted EBITDA or add-backs, the borrower may have more room to act than its reported earnings alone would suggest.</p><p>The consequences can be severe when value moves beyond the reach of existing creditors.</p><p>The 2016 J.Crew transaction illustrates the risk. The company transferred intellectual property to an unrestricted subsidiary, raising concerns about the extent to which lenders&#8217; collateral could be weakened without their consent.</p><p>British floor manufacturer Victoria offers a different lesson. In 2025, it offered certain holders of its &#8364;500 million 2026 senior secured notes an exchange into new first-priority notes, while seeking to subordinate existing notes through a separate consent solicitation. A subsequent exchange offer involving its 2028 notes was withdrawn in October 2025 following weak investor support.</p><p>The two cases should not be treated as equivalent. J.Crew demonstrated how contractual permissions could facilitate an asset transfer. Victoria demonstrated how a transaction that benefited some lenders could violate the rights of others.</p><p>Both underline the need to read the actual provisions, rather than assume that a covenant-lite label tells you how much protection remains.</p><p><strong>Liquidity.</strong> A maintenance-covenant breach can create a defined event around which lenders negotiate, even if the breach is subsequently waived or cured. It can also force a company to seek relief at an inconvenient time.</p><p>Without that trigger, deterioration may become apparent only when the borrower needs additional financing or a restructuring. By then, the lender may have fewer ways to influence the outcome.</p><p>This does not mean a covenant breach automatically improves recovery, or that a covenant-lite loan is necessarily harder to sell. Secondary-market liquidity depends on the borrower, the instrument and market conditions. The more specific concern is that a lender may lose an opportunity to negotiate before financial stress becomes acute.</p><h2>The counter-case</h2><p>Covenant-lite is not a conspiracy, and it is not synonymous with unprotected. Maintenance covenants have costs too.</p><p>During the COVID-19 shock in 2020, deteriorating earnings pushed some borrowers towards covenant breaches, in part because EBITDA fell sharply while debt remained outstanding. Many lenders responded through waivers, amendments or equity cures. Some companies faced more substantial restructuring.</p><p>A technical breach can therefore trigger a costly negotiation even when the underlying business remains viable. Sponsors value flexibility partly because the alternative can give lenders leverage at the wrong moment.</p><p>The empirical evidence also challenges the assumption that covenant-lite lenders necessarily receive higher spreads or suffer worse recoveries.</p><p>If weaker protections are not consistently rewarded with higher spreads, investors need to understand what they are giving up before accepting the loan&#8217;s yield as adequate compensation.</p><p>Documentation also deserves a case-by-case assessment. Security packages, guarantees and incurrence restrictions can provide meaningful protection, but their strength varies by deal. </p><h2>The closing point</h2><p>A covenant is never a guarantee of repayment. It is a contractual right that could give lenders an earlier opportunity to ask questions, demand concessions and negotiate from a stronger position.</p><p>Covenant-lite documentation reduces that opportunity. Sometimes the flexibility benefits the borrower without materially harming lenders. Sometimes it becomes critical when financial stress exposes weaknesses in the capital structure.</p><p>There is no universal spread premium that compensates investors for accepting this trade-off. The academic evidence suggests that covenant-lite loans have sometimes priced at tighter spreads despite weaker maintenance protections. That makes the quality of the documentation, and the price paid for the loan, more important, not less.</p><p>Owning covenant-lite paper should be a deliberate decision based on the borrower&#8217;s financial position, the protections that remain, the likely recovery in a downside scenario and the price of the instrument. It should not be an automatic consequence of a mandate that says &#8220;investment grade or better, maximum five-year maturity, no equities.&#8221;</p><p>The covenant did not necessarily disappear. But the opportunity to use it may have moved to a moment when the lender has less time, fewer options and less leverage.<br><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[The Great Divide in the CCC Market]]></title><description><![CDATA[The widening gap between CCC and BB yields reflects more than rising credit risk. Forced selling, weaker creditor protections and shifting investor appetite are reshaping Europe's high-yield market.]]></description><link>https://www.rovelliresearch.com/p/the-great-divide-in-the-ccc-market</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/the-great-divide-in-the-ccc-market</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Thu, 05 Nov 2026 06:03:29 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/5c74ed40-4101-4c8f-8ba3-6294352398a2_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>European CCC yields are above 21%, compared with around 18% in the United States. At the start of the year, they stood at roughly 16% and 15%, respectively. In ten months, yields at the bottom of the high-yield market have therefore risen by five percentage points in Europe and three in the United States. The gap between CCC and BB bonds has widened sharply in Europe, while yields on higher-rated BB debt have remained relatively stable.</p><p>Two aspects of this move deserve particular attention. European CCC bonds have underperformed their US counterparts, and the deterioration has been concentrated at the bottom of the credit spectrum rather than spreading across the broader high-yield market.</p><p>The geographical difference provides a useful starting point. Several factors have contributed to the weakness in high-yield credit globally, including higher interest rates, concerns about the impact of artificial intelligence on existing business models and elevated energy costs. Europe faces additional pressure from concerns about France&#8217;s fiscal position, while telecom companies have been particularly exposed to the deterioration in investor sentiment.</p><p>These are primarily cost-of-capital and business-model challenges rather than evidence of a broad economic contraction. They are also particularly difficult for highly leveraged companies with limited financial flexibility. A borrower facing higher refinancing costs, potential disruption to its business and persistently high operating expenses does not necessarily need a recession to run into financial difficulties. It may simply need to refinance its debt at the wrong time.</p><p>The distinction between stronger and weaker CCC issuers is important. Goldman Sachs had already highlighted this risk in June, noting the considerable differences in credit quality within the CCC category. Some issuers continue to perform reasonably well, while others are already under considerable financial pressure. Much of this divergence is not immediately apparent from headline credit spreads.</p><p>Recent market performance supports that assessment. The weakness has been concentrated in a relatively small number of issuers and capital structures, including Altice International, Virgin Media, VodafoneZiggo and Aston Martin, rather than affecting the entire CCC market uniformly. At the same time, BB bonds have held up considerably better.</p><p>The original concern was that a genuine economic slowdown combined with persistent inflation could trigger a broader deterioration in credit quality. That scenario has not materialized, yet the gap between CCC and BB has widened anyway. The explanation lies in several other forces, including interest rates, energy costs, creditor protections and market positioning.</p><p>Documentation is one of them. In recent years, private equity owners have increasingly used the flexibility provided by loosely drafted bond documentation to transfer assets outside the group of companies guaranteeing existing debt or to raise new debt with priority over existing creditors.</p><p>These transactions can materially weaken the position of existing bondholders. The underlying business may continue to generate cash and meet its interest payments, but the assets available to creditors in a restructuring may be worth considerably less. This distinction is important: a transaction of this kind does not necessarily increase the probability of default, but it can reduce the recovery available to lenders if default eventually occurs.</p><p>The incentives are particularly strong when sponsors have lost most or all of their original equity investment. Owners in this position have less to lose from taking additional risks, while loose bond documentation can give them considerable freedom to restructure the debt and move assets within the corporate group. For bondholders, the consequences can be significant even before the company experiences any deterioration in its operating performance.</p><p>Market technicals have also amplified the sell-off. Several large credit funds have suffered substantial losses, while some multi-strategy investors have reduced or closed their credit positions. When funds are forced to unwind positions in an already illiquid market, prices can fall well beyond what the underlying credit fundamentals would justify.</p><p>Forced selling reflects the need for investors to reduce exposure or raise liquidity. However, when a relatively small number of funds hold a large proportion of an issuer&#8217;s bonds, their selling can have a disproportionate impact on market prices.</p><p>A further explanation for the widening gap is the relative attractiveness of higher-rated credit. With BB bonds offering increasingly attractive yields, investors have less incentive to take additional credit risk in exchange for the incremental spread available on CCC securities. This is a portfolio allocation decision rather than necessarily a deterioration in expected default rates. The marginal buyer of CCC debt may simply prefer to earn an attractive return in BB securities without accepting the additional risks associated with weaker borrowers.</p><p>This helps explain why the deterioration has been concentrated at the bottom of the market. If the current weakness were primarily the result of an economic or credit cycle, one would expect BB spreads to widen as well. So far, this has not happened. Higher-quality issuers generally have more manageable leverage, stronger interest coverage and better access to capital markets. They are consequently better positioned to absorb higher interest rates and temporary operating pressures.</p><p>The relative resilience of BB bonds is therefore an important signal. Four forces have contributed to the deterioration in CCC credit: higher financing costs, weaknesses in bond documentation, forced selling and reduced investor appetite for lower-rated debt. Each has affected the weakest borrowers considerably more than higher-rated issuers. This asymmetry suggests that the market is repricing a particularly vulnerable segment of high-yield credit rather than anticipating an imminent wave of defaults across the entire asset class.</p><p>My preference is to focus on generating income from carefully selected bonds rather than relying on a broad recovery in CCC valuations. I favor shorter-dated securities with adequate protection against early redemption, where the contractual terms and expected repayment profile provide greater visibility on returns.</p><p>Within the CCC market, I see more potential opportunities in Europe than in the United States. The European sell-off has been more pronounced, and the concentration of weakness in the telecom sector provides a relatively identifiable group of issuers for further analysis. However, the larger spread widening should not, by itself, be interpreted as evidence of greater value.</p><p>I would avoid broad exposure to CCC indices in both markets. An index combines companies whose bonds have fallen because of temporary selling pressure with those facing genuine structural problems in their capital structures. These are fundamentally different investment propositions, and I would rather select individual issuers where the reasons for the price decline can be clearly established.</p><p>The widening gap between CCC and BB is therefore not, in my view, a straightforward signal to increase exposure to lower-rated credit. It reflects a combination of higher financing costs, technical selling and growing concerns about the rights of existing creditors. There are opportunities in this environment, but they require careful analysis of individual capital structures and, in particular, the documentation governing each bond.</p><p>I would be interested to hear from investors working on European telecom capital structures, particularly where recent asset transfers or debt restructurings have changed the position of existing bondholders. Recovery analysis in these situations can provide valuable insight into where the market may be mispricing risk.<br><br><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Medallia: How a $6.4bn Buyout Became a Restructuring]]></title><description><![CDATA[Medallia&#8217;s restructuring exposes the risks of lending against recurring revenue, compounding deferred interest and peak-cycle software valuations, leaving lenders dependent on cash generation]]></description><link>https://www.rovelliresearch.com/p/medallia-how-a-64bn-buyout-became</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/medallia-how-a-64bn-buyout-became</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 03 Nov 2026 06:18:31 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/4070f284-c270-43de-b47f-0c2c675da23d_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Private credit expected <a href="https://www.medallia.com/">Medallia</a>, a Software as a service company, to grow fast enough to support its debt. When that growth failed to materialize, the lenders ended up taking control of the company. The 2026 restructuring was not a bankruptcy, but it marked the end of the original investment plan.</p><p><a href="https://www.thomabravo.com/">Thoma Bravo</a> took Medallia private in 2021 in a deal valued at $6.4 billion, including debt. The equity was worth about $5.5 billion. The financing included more than $2 billion of loans, reportedly priced at SOFR plus 650 basis points, with 400 basis points of interest that could be added to the debt rather than paid in cash. The loans were due in 2028. The original lenders included Blackstone, Apollo, KKR, Thoma Bravo Credit and Antares Capital.</p><p>By 2026, unpaid interest and additional borrowing had increased the debt burden. A restructuring completed on 3 August transferred ownership to the lenders and wiped out Thoma Bravo&#8217;s equity investment. The lenders also put in $150 million of new capital to support the business. They now own the company they had originally financed.</p><p>The problem was not simply that Thoma Bravo paid too much. The deal assumed that software companies would continue to command high valuations and that Medallia would grow enough to support its debt. The lenders made a similar bet. They were willing to lend heavily against the company&#8217;s recurring subscription revenue, even though that revenue did not necessarily translate into enough cash to pay interest and repay debt. When growth slowed and software valuations fell, the assumptions behind the deal no longer held.</p><p>Recurring revenue is useful because it shows how much subscription income a company expects to receive each year. But does not account for the cost of running the business, winning new customers or keeping existing ones. Nor does it tell lenders how much cash is left to pay interest and repay debt. A company can have substantial recurring revenue and still struggle to service its loans.</p><p>The same problem arises when the value of the business depends on expectations of future growth. If customers demand lower prices at renewal, growth slows or artificial intelligence changes how customers use the product, the business may be worth much less than lenders expected. Recurring revenue can help lenders assess a software company, but it cannot replace an analysis of its earnings and cash flow.</p><p>The loans paid a margin of 650 basis points above the floating interest rate. That looked attractive, but the spread did not fully compensate lenders for the risks they were taking. They were not just exposed to higher interest rates or a temporary decline in earnings. Their investment also depended on Medallia continuing to grow and retaining a high valuation. If the value of the business falls below its debt, a generous interest margin offers little protection against losses.</p><p>The option to add some interest to the debt rather than pay it in cash gave Medallia more breathing room. But it also increased the amount the company owed. Each time interest was added to the balance, the debt grew without bringing in any new cash. This can help a struggling company survive while it improves its business, but it can also delay a reckoning. If earnings do not recover, the growing debt makes the eventual restructuring more difficult.</p><p>Medallia&#8217;s problems reflect a wider trend in private credit. Houlihan Lokey reported that more than 20% of private-debt transactions included an option to add interest to the loan balance in late 2025, up from 6.9% three years earlier. Software was among the sectors most reliant on these arrangements. I have not found reliable public evidence to support the more specific claim that more than a third of software loan deals included this feature. The broader figures still show that lenders have become more willing to let borrowers defer interest as financing costs rise and refinancing becomes harder.</p><p>The restructuring changed who owns Medallia and reduced its debt, while providing $150 million of new capital. But the exact amount of debt left after the deal is not clear from the company&#8217;s public disclosures. The often-cited $2.8 billion figure refers to the debt before the restructuring, not a confirmed post-restructuring balance. It would therefore be misleading to treat that figure as the amount the company still owes. To assess Medallia today, lenders need to look at the actual debt remaining, the interest bill and when repayments fall due.</p><p>The deal also shows that having several lenders does not necessarily reduce risk. Blackstone, Apollo and KKR participated in the original 2021 financing. HPS, FS KKR, Monroe and Onex were among the lenders holding debt before the restructuring, rather than all being original lenders. These institutions may have entered at different times and prices, but they ultimately faced the same problem: the business was no longer worth enough to support the original financing.</p><p>Medallia offers three important lessons. First, recurring revenue can support substantial borrowing only if it generates enough cash to pay the bills and service the debt. Second, deferred interest is not income. It increases the amount a borrower owes and can make a bad situation worse. Third, a long list of sophisticated lenders does not make a deal safe. Several institutions can make the same mistake when they rely on the same assumptions about growth and valuations.</p><p>There is still a case for recovery. Medallia has an established customer base and sells software that helps companies understand and improve customer experience. Under lender ownership, it may be able to cut costs, improve its products and return to growth. The new capital gives it some support while it works through these changes.</p><p>But the company must improve its business while dealing with its remaining debt. The fact that customer-experience software has a future does not guarantee that Medallia will earn enough to support its financing. What matters is whether it can turn its customer base and revenue into cash.</p><p>I remain cautious. The restructuring has given Medallia new owners and fresh capital, but that does not guarantee a successful recovery. I would want to see the company cover its cash interest payments from operating cash flow, retain its customers without having to cut prices too aggressively, and have enough time before its next major debt repayment to deliver on its recovery plan. Without those improvements, the lenders may simply have exchanged troubled loans for ownership of a business that is still struggling.</p><p>The next few years should make the outcome clearer. Can Medallia generate enough cash to service its debt before repayments fall due? Are customers renewing at the prices assumed in the recovery plan? If the business underperforms, will the lenders put in more equity, or will they simply extend the debt and delay the problem? The answers will show whether this restructuring has given Medallia a viable future or merely postponed another round of negotiations.</p><p>Would you lend against recurring revenue today, and how much debt would you be willing to provide?</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p><p></p>]]></content:encoded></item><item><title><![CDATA[Europe’s Maturity Wall: Smaller, but Not Solved]]></title><description><![CDATA[Europe has refinanced much of its looming debt wall, but the borrowers left behind are weaker, harder to fund, and dependent on issuer-specific credit analysis.]]></description><link>https://www.rovelliresearch.com/p/europes-maturity-wall-smaller-but</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/europes-maturity-wall-smaller-but</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Thu, 29 Oct 2026 06:20:42 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/0f8b2a49-3f81-4aa0-b2de-9fa88f92f30a_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>In late September I argued that European credit had completed much of its near-term refinancing. Stronger borrowers had extended their maturities, reducing the amount of debt due in 2027&#8211;28. The latest data supports that view, but with an important qualification: the remaining borrowers face greater financing challenges. The market has reduced its aggregate refinancing risk, but the weaker credits still need to demonstrate that they can service their debt and secure funding at maturity.</p><p>Between July 2025 and July 2026, the amount of European leveraged loans maturing in 2028 fell from &#8364;91 billion to &#8364;38 billion, a reduction of 58%. The combined amount of European loans maturing in 2026 and 2027 also fell from &#8364;9 billion at the end of 2025 to &#8364;4 billion at June 2026. Refinancing and extensions accounted for much of the activity. Over the 12 months to September, they represented approximately 67% of European leveraged-loan volume, or &#8364;111 billion. The peak of the maturity profile has shifted to 2031, when maturities are projected to reach &#8364;99 billion.</p><p>This represents a substantial reduction in scheduled maturities. It does not necessarily indicate an equivalent improvement in borrowers&#8217; credit quality. An extension changes the repayment date but does not increase a company&#8217;s capacity to repay. Companies that refinanced early secured more time, while those unable to access the market on acceptable terms face fewer options. The 2027&#8211;28 maturity wall therefore poses less of a broad market threat than previously feared, although refinancing risk remains concentrated among weaker issuers.</p><p>The secondary market offers a further perspective on the remaining maturity wall. By the end of July, nearly 40% of European leveraged loans maturing in 2028 traded below &#8364;0.90 per &#8364;1 of principal. These discounts can reflect both contractual spreads that lag current market levels and concerns about borrowers&#8217; credit quality, refinancing prospects and potential recovery rates. The price data alone cannot distinguish between these factors or establish that every affected borrower faces distress. It does, however, show that a substantial share of the remaining debt trades at a discount, despite the reduction in aggregate maturities. The next step is to separate loans that offer inadequate spreads from those whose borrowers face genuine refinancing or solvency risks.</p><p>Software deserves particular scrutiny, although the sector is not uniformly distressed. Cegid illustrates both the availability and the cost of private financing. In June, the group secured a &#8364;1.1 billion private-credit facility to finance its acquisition of Shine after its syndicated financing process had stalled. The facility reportedly carries a margin of 550 basis points over EURIBOR and was issued at 98, compared with a 325-basis-point margin on Cegid&#8217;s existing &#8364;1.04 billion syndicated loan due in 2028. The new lenders therefore received substantially higher pricing, although differences in maturity and purpose limit the comparability of the two facilities. The transaction shows that private credit can fund large, complex software deals when syndicated markets hesitate, but at a higher financing cost for the borrower. It also leaves Cegid&#8217;s existing 2028 maturity unresolved.</p><p>Software&#8217;s exposure to changing growth assumptions also needs to be assessed at the issuer level. Second-quarter private-credit research found that software had among the lowest default rates of any industry in the dataset. Median software EBITDA was also 20% higher than when those loans were initially originated. These findings do not rule out refinancing pressure among individual borrowers, particularly those with high leverage or business models vulnerable to artificial intelligence. They do, however, argue against assuming that AI disruption has already produced sector-wide credit deterioration. The central credit question is whether each borrower&#8217;s operating performance and cash generation can support its capital structure.</p><p>Refinancing pressure is not confined to software. Chemicals and building materials also feature among the sectors facing unresolved financing challenges. Rather than identify one industry as the worst-performing, investors should distinguish between borrowers with viable businesses and manageable debt, and those whose repayment depends on a favorable refinancing market, a recovery in demand or further support from their sponsors.</p><p>Ratings and capital structure add another dimension. A downgrade into the CCC category can push a loan above the permitted exposure limits of a collateralized loan obligation (CLO) and create pressure to sell. It does not automatically force every CLO holding the loan to dispose of it. The outcome depends on each vehicle&#8217;s existing portfolio, applicable limits, the loan&#8217;s valuation and the availability of other buyers. A smaller maturity wall can therefore coexist with selling pressure and further price declines among individual issuers.</p><p>Private credit provides a complementary view of this concentration of risk. In the second quarter of 2026, 7% of loans in a private-credit dataset were valued below 90% of par, more than twice the historical average. Among borrowers with $10&#8211;20 million of EBITDA, the share was 12%, compared with approximately 1% in 2023. For private-credit borrowers with less than $100 million of EBITDA, the default rate reached 4% by borrower count and 3% on a size-weighted basis. These rates include technical defaults as well as payment defaults.</p><p>The broader private-credit figures provide an important counterweight. Across the full market, defaults represented 1% of outstanding loan principal, or 3% of borrowers by count. Median revenue rose 7% year over year and median EBITDA increased 7%, with more than two-thirds of borrowers growing both. The data therefore points to concentrated stress rather than a general collapse. Smaller and weaker borrowers face greater pressure, while many larger businesses continue to grow.</p><p>For investors, the focus should shift from the size of the aggregate maturity wall to the cash generation and refinancing capacity of individual borrowers. In software, this requires testing whether recurring revenue translates into cash earnings, how much adjusted EBITDA depends on addbacks, whether borrowers actually capitalize interest through payment-in-kind arrangements, and whether cash flow comfortably covers interest payments. Artificial intelligence presents a potential threat to growth and pricing power, but its credit implications depend on each company&#8217;s competitive position, financial structure and ability to adapt.</p><p>Investors should prioritize short-dated, higher-quality bonds with call protection. These securities can provide contractual income without requiring a substantial decline in credit spreads or depending on a distant refinancing. By contrast, long-dated, low-coupon bonds may offer limited compensation for the risk that their issuers will need to refinance in a market demanding higher yields for weaker credit. Their performance could suffer both from wider spreads and from doubts about refinancing capacity.</p><p>Europe has pushed out much of its near-term debt burden, but the remaining borrowers still face significant financing constraints. Their ability to generate cash, cover interest and refinance on sustainable terms will increasingly determine credit performance. The investment case consequently depends less on the size of the market-wide maturity wall and more on the financial strength of each issuer.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[The Yield Is the Easy Part]]></title><description><![CDATA[A 7% yield can look attractive until taxes, hedging costs, refinancing risk, and early redemption eat into your returns.]]></description><link>https://www.rovelliresearch.com/p/the-yield-is-the-easy-part</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/the-yield-is-the-easy-part</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 27 Oct 2026 18:15:36 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/1287c9ac-7122-4ecc-8f30-f62db5cd7826_1080x1350.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Choosing a bond is only one part of investing in European high yield. The return you actually earn also depends on how you hold it, the tax you pay, the cost of hedging currency risk, and the terms under which the issuer can repay the debt early. These factors can make a meaningful difference to the income left in your portfolio, even when two investments offer similar yields.</p><p>For investors allocating capital across countries and currencies, the order of decisions matters. Start with the investment structure and the currency in which you measure your wealth. Then examine tax treatment, call protection, and the issuer&#8217;s refinancing needs. Only after that should you compare credit spreads, the additional yield a corporate bond offers over a government bond.</p><h2>Start with what you own</h2><p>Direct bonds, bond funds, and collateralized loan obligations (CLOs) are different investments, even when they appear under the same credit allocation.</p><p>A direct bond gives you a contractual claim on the issuer. You receive interest and, provided the issuer meets its obligations, repayment of principal at maturity. You also bear the risk that the bond loses value or becomes difficult to sell.</p><p>A bond fund pools investors&#8217; money and holds a portfolio of bonds. Your return depends on the fund&#8217;s holdings, management fees, tax treatment, and redemption terms. You own units in the fund rather than the underlying bonds directly.</p><p>A CLO holds a portfolio of corporate loans and finances those loans by issuing different classes of securities. These classes absorb losses in a defined order. Senior securities receive priority, while the equity tranche absorbs losses first and can benefit disproportionately if the loan portfolio performs well. An investment in a CLO is therefore not equivalent to holding a conventional corporate bond. The risks depend on the structure, the underlying loans, and the tranche you own.</p><p>None of these structures is inherently unsuitable. The important point is to understand what you own, what fees and restrictions apply, and how the investment will be valued and sold when you need the money.</p><h2>Tax can change the return you keep</h2><p>Tax treatment depends on the instrument, the investor&#8217;s tax residence, and the legal structure through which the investment is held. A different investment vehicle can produce a different after-tax return even when the underlying credit exposure is similar.</p><p>For example, interest paid on qualifying US registered corporate bonds can generally be exempt from US withholding tax under the portfolio interest exemption. Conditions apply, including restrictions relating to certain shareholders and the form of the debt.</p><p>Holding the same exposure through a US fund can produce a different result. Fund distributions may include interest-related dividends that qualify for an exemption from US withholding tax, but this depends on the fund&#8217;s eligibility and reporting. Investors should not assume that every distribution qualifies or that the treatment is identical to holding the bond directly.</p><p>European bond interest is also often paid without withholding tax to eligible nonresident investors, but the rules vary by country and investor type. Before investing, check the issuer&#8217;s country, the bond documentation, the paying arrangements, the custody structure, and the tax residence of the investment vehicle.</p><p>Bond documentation also sometimes contains a <em>gross-up clause</em>. This provision may require an issuer to increase payments if certain withholding taxes apply, so that the investor receives the contracted amount. The protection is subject to the wording and exclusions in the bond documents. Do not assume it exists or covers every tax charge.</p><p>The relevant comparison is the net return after tax and fees, not simply the yield quoted on a trading screen.</p><h2>Currency hedging is not free money</h2><p>An investor whose wealth is measured in US dollars may want to buy euro-denominated bonds without taking an open position on the euro. A currency hedge uses a forward contract or another financial instrument to reduce the effect of exchange-rate movements on the investment&#8217;s value.</p><p>The cost or benefit of that hedge depends partly on the difference between interest rates in the two currencies. This is reflected in <em>forward points</em>, the adjustment to today&#8217;s exchange rate used to determine the agreed future exchange rate.</p><p>When US interest rates are above euro interest rates, a dollar-based investor hedging euros into dollars will generally receive a favorable interest-rate differential through the forward exchange rate. When US rates are below euro rates, the hedge will generally have a cost instead. The precise result depends on the currencies, maturity, market pricing, and transaction costs.</p><p>At mid-October 2026 levels, the three-year euro-dollar interest-rate differential was estimated to provide a dollar-based investor with a hedge benefit of roughly 150 to 160 basis points a year. That estimate should be checked against executable forward prices before making an investment decision. It is not a guaranteed return, and the benefit will change as interest rates and market pricing move.</p><p>For a euro-based investor buying euro bonds, no currency hedge is required. For an investor whose base currency is different, the decision is whether to accept the exchange-rate risk or pay the cost, or receive the benefit, of hedging it.</p><p>A favorable hedge does not create free investment returns. It largely reflects the interest-rate differential between currencies. The credit spread earned on the bond remains compensation for taking credit risk. An investor should compare the bond&#8217;s yield after hedging costs with the yield available on comparable investments in their base currency, rather than treating a favorable hedge as additional credit alpha.</p><h2>Know which government bond is your benchmark</h2><p>European corporate bonds are commonly priced relative to government bonds or swap rates. The benchmark helps separate the general level of interest rates from the additional compensation investors demand for taking credit risk.</p><p>For euro-denominated bonds, the German government bond, or Bund, is a widely used sovereign benchmark. French government bonds can trade at a higher yield than German Bunds because investors demand additional compensation for differences in perceived sovereign risk, liquidity, and supply.</p><p>The French-German government bond spread has widened substantially over recent years, although the precise change depends on the maturity selected. This reflects concerns about French public finances and political uncertainty, alongside market supply and demand.</p><p>The distinction is important for corporate credit analysis. A higher corporate bond yield does not necessarily mean that the issuer&#8217;s credit risk has increased by the same amount. Some of the movement may reflect changes in government bond yields or the benchmark used for comparison. Investors should establish whether a quoted spread is measured against a government bond or a swap rate before comparing securities or managers.</p><h2>Low coupons and call protection</h2><p>Many European high-yield bonds issued in 2020 and 2021 carried coupons of 2% to 4%, reflecting the exceptionally low interest rates of that period. A bond&#8217;s <em>coupon</em> is the contractual interest rate applied to its principal. Its <em>yield to maturity</em> is the annualized return implied by its market price, assuming the contractual payments are made and the bond is held to maturity.</p><p>When market yields rise, the price of an existing fixed-rate bond generally falls. A bond paying a 3% coupon becomes less attractive when investors can buy comparable new bonds yielding 6.5%. Its price must decline to compensate a new buyer for the lower contractual interest payments.</p><p>This explains why some low-coupon bonds trade well below their original issue price. A bond with a face value of &#8364;100 might trade at &#8364;85, for example, because its coupon is below the return currently demanded by the market. The precise price depends on the remaining maturity, credit quality, liquidity, and contractual terms.</p><p>Investors must also check <em>call protection</em>, the period during which an issuer cannot redeem a bond early, or can do so only under specified conditions. Some bonds allow early redemption at a premium to face value. A first call price of 101.5, for example, means the issuer can redeem &#8364;100 of principal for &#8364;101.50, subject to the terms of the bond.</p><p>Call protection has two implications. If a bond trades below its call price, early redemption is generally less attractive to the issuer than leaving the debt outstanding, although other contractual provisions may apply. If interest rates fall and the bond&#8217;s price rises above the redemption price, the issuer may be able to refinance at a lower rate and repay investors early. Investors then have to reinvest the proceeds, potentially at a lower yield.</p><p>The practical lesson is to examine both the current yield and the bond&#8217;s redemption schedule. A high quoted yield is not necessarily the return you will earn if the bond is repaid before maturity.</p><h2>The refinancing risk is more important than the coupon</h2><p>The larger concern for many low-coupon bonds issued in 2020 and 2021 is what happens when they mature in 2027 and 2028.</p><p>An issuer that borrowed at 3% may have to refinance at a much higher rate. If it replaces &#8364;1 billion of debt at 6.5% rather than 3%, its annual interest expense increases by approximately &#8364;35 million, assuming the full amount is refinanced at the new rate and other terms remain unchanged.</p><p>That is a material increase for a company with limited cash generation. Higher interest costs reduce the cash available for investment, debt repayment, and shareholder distributions. Some issuers may be able to absorb the increase through earnings growth or cost savings. Others may need to cut spending, sell assets, or negotiate with lenders.</p><p>Refinancing risk is not the same for every company. An issuer with low leverage, stable cash flow, and access to bank facilities is better placed than one with weak earnings, substantial near-term maturities, and limited liquidity.</p><p>For investors, the first step is to map the maturity schedule of the bonds they own. The next is to examine the issuer&#8217;s ability to repay or refinance them. A bond&#8217;s maturity should also fit the investor&#8217;s own cash requirements. Selling before maturity may be necessary, and the market price at that point could be below the purchase price.</p><p>Low-coupon bonds deserve particular attention because a larger share of their total return may depend on changes in market price rather than on the interest received. That makes the timing of the investment and the issuer&#8217;s credit trajectory especially important.</p><h2>Test cash interest coverage, not just reported earnings</h2><p>Another useful measure is <em>cash interest coverage</em>: the cash generated by a company relative to the interest it must pay in cash. It helps assess whether the borrower can service its debt without relying on additional borrowing or noncash accounting items.</p><p>Reported coverage can look stronger than the underlying cash position if the calculation includes payment-in-kind (PIK) interest, which is added to the debt rather than paid in cash, or other adjustments that reduce the reported interest burden. Capitalized fees and changes in contractual interest rates can also affect the comparison.</p><p>Consider a borrower reporting interest coverage of 1.4 times, meaning the earnings measure used in the calculation is 1.4 times interest expense. If its coverage falls to 1.05 times when only cash earnings and cash interest payments are considered, the margin of safety is much thinner. A modest fall in earnings or increase in interest costs could leave it struggling to meet its obligations.</p><p>The exact calculation matters. Investors should establish whether the numerator is EBITDA, operating cash flow, or another measure, and whether the denominator includes all cash interest, fees, and financing costs. EBITDA is earnings before interest, taxes, depreciation, and amortization; it is not the same as cash flow because it excludes capital expenditures, working-capital movements, and other cash requirements.</p><p>Covenants provide another source of protection, but their value depends on how they are written and tested. A <em>maintenance covenant</em> requires a borrower to meet a financial test at regular intervals, such as keeping leverage below a specified level. Other covenants may apply only when the company takes a particular action, such as raising more debt.</p><p>For loans backed by private assets, ask how the financial figures are calculated, who values the assets, and how frequently the values are updated. A covenant based on stale or optimistic valuations may provide less protection than the headline terms suggest.</p><h2>Liquidity can look stronger than it is</h2><p>Private credit and other less liquid investments introduce a separate problem: the price reported by a fund may not be the price at which an investor could sell.</p><p><em>Net asset value (NAV)</em> is the reported value of a fund&#8217;s assets minus its liabilities. For frequently traded securities, market prices can provide a reasonably current indication of value. For privately negotiated loans, valuations may depend on models, comparable transactions, and assumptions about the borrower&#8217;s prospects.</p><p>A quarterly NAV is therefore an estimate as of a particular date, not necessarily an executable sale price. The difference becomes important when credit conditions deteriorate or several investors try to sell at once.</p><p>Liquidity also depends on the investment vehicle. A fund that offers frequent redemptions while holding loans that take months to sell may face pressure if investors withdraw their money. The fund may need to restrict redemptions, sell assets at a discount, or use other measures permitted by its documents.</p><p>Investors should ask what proportion of a portfolio could realistically be sold within a week, at what discount, and under what conditions. They should also distinguish between cash held by the fund, securities that can be sold quickly, and assets that are only valued periodically.</p><p>A portfolio described as liquid may still contain a meaningful allocation to investments that cannot be sold quickly at their reported value.</p><h2>What could make this view wrong?</h2><p>The refinancing squeeze is a risk, not a certainty. If interest rates fall, credit spreads tighten, and lenders remain willing to refinance companies on reasonable terms, many issuers could manage their 2027 and 2028 maturities without serious difficulty. In that scenario, investors who favor shorter-dated, higher-quality bonds may underperform those who hold longer-dated securities offering more spread.</p><p>The opposite outcome is also possible. If refinancing costs remain high while earnings weaken, interest expense could rise just as companies become less able to absorb it. Defaults and distressed exchanges could increase, particularly among issuers with high leverage and limited access to funding.</p><p>Currency hedging is equally dependent on the rate outlook. A dollar-based investor benefiting from today&#8217;s euro-dollar interest-rate differential should not assume that the same benefit will be available when the hedge is renewed. If the rate differential narrows, the annual hedge benefit will decline.</p><p>The appropriate response is not to avoid all risk. It is to identify which risks are being taken, what return compensates for them, and whether the portfolio can withstand an adverse outcome.</p><h2>Three questions before your next allocation</h2><p>Before investing in European credit, ask your manager for three figures.</p><p>First, what is the portfolio&#8217;s cash interest coverage, calculated using a clearly defined earnings measure and cash interest expense?</p><p>Second, how much of the portfolio matures or can be redeemed early over the next 12 months, and how much debt will the underlying issuers need to refinance?</p><p>Third, what is the annual cost or benefit of hedging the portfolio into your base currency, after transaction costs?</p><p>The answers will not replace issuer-level credit analysis. They will, however, show whether the portfolio&#8217;s headline yield adequately reflects its refinancing, liquidity, and currency risks. A high yield is only attractive when the risks behind it are understood and the investor is adequately compensated for taking them.</p><p></p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent Credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[When a Technology Failure Becomes a Credit Event]]></title><description><![CDATA[A future liquidity shock could originate not in a deterioration of credit quality, but in the simultaneous failure of technology used by otherwise independent financial institutions.]]></description><link>https://www.rovelliresearch.com/p/when-a-technology-failure-becomes</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/when-a-technology-failure-becomes</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Thu, 22 Oct 2026 23:24:09 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/b5d521de-5e0f-46be-aa33-abb0a93a2f41_1080x1920.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p><em>This article presents a thought experiment, not a forecast. It does not predict when or whether such an event will occur. Instead, it examines the potential credit consequences of a large-scale technology failure and considers whether current risk analysis and stress testing adequately capture the resulting exposures.</em></p><p>The central hypothesis is that a future liquidity shock could originate not in a deterioration of credit quality, but in the simultaneous failure of technology used by otherwise independent financial institutions.</p><p>The relevant historical parallel is the withdrawal of funding during 2007 and 2008. At that time, financing markets closed by sector and then by borrower. The availability of funding deteriorated before prices fully reflected the underlying risks. A comparable dynamic today could originate in a different part of the financial system: the technology used by banks, custodians, payment agents and asset managers to process transactions, maintain records and verify ownership.</p><p>A failure in a widely used software component could interrupt payments, compromise the reliability of financial records and create immediate liquidity demands across several solvent institutions. The initial event would be technological, but its consequences could quickly become a credit problem.</p><p>The concern is not that artificial intelligence will act independently of human control on a global scale. It is that numerous outsourcing and automation decisions, each reasonable in isolation, may have created common dependencies across institutions that otherwise appear diversified. If those dependencies were disrupted simultaneously, the resulting risks could become correlated.</p><h2>How the risk could develop</h2><p>Financial institutions have increasingly outsourced functions such as cloud hosting, identity verification, payment routing, reconciliation, fraud screening and endpoint management. The rationale is straightforward: lower costs, faster software development and, in many cases, more resilient infrastructure.</p><p>The collective result, however, may be a concentration of operational risk. Institutions that appear independent can rely on the same technology provider, software library, artificial intelligence model or system configuration. A defective software update, compromised credential repository or unreliable data feed could disrupt multiple institutions at once.</p><p>The duration of the disruption would be less important than the extent of the shared exposure. Even a relatively short interruption could create significant liquidity demands if several institutions were unable to establish ownership, reconcile transactions or process payments at the same time.</p><p>Automated execution could intensify the disruption. Systems capable of transferring funds, modifying code, accessing production environments or changing user permissions may continue operating while an incident is being investigated. The distinction between systems that make recommendations and those authorized to execute actions is therefore important.</p><p>Under normal conditions, automation improves speed and consistency. If the underlying instructions, data or identity controls were compromised, however, automated processes could propagate an error before operators identified the source and intervened. The speed that ordinarily makes these systems valuable could then accelerate the disruption.</p><h2>What financial authorities have identified</h2><p>Financial authorities have identified operational resilience, cyber disruption and reliance on external technology providers as areas of concern. The potential systemic risk arises when a disruption affects multiple institutions through the same technology dependencies rather than remaining confined to an individual firm.</p><p>These concerns do not require a scenario in which machines take control of financial institutions. They relate to a more conventional problem: failures in operational infrastructure that create funding pressures.</p><p>Analysis of financial structures also offers a useful perspective. Off-balance-sheet vehicles, outsourced servicing and leased infrastructure can relocate financial exposures without eliminating them. Some elements of the current investment cycle in artificial intelligence could have similar characteristics. Joint ventures, special-purpose vehicles and vendor-operated infrastructure may distribute obligations across several entities, making the resulting exposures difficult to assess in isolation.</p><p>The potential vulnerability lies in the interaction between these structures and a common operational disruption. Obligations that appear manageable under normal conditions could generate simultaneous cash requirements if several counterparties or service providers became unable to perform.</p><h2>From operational disruption to credit deterioration</h2><p>The transmission mechanism would probably begin with funding pressure rather than immediate insolvency.</p><p>If payment and settlement systems were interrupted, contractual obligations would remain due even when the underlying transactions could not be verified or completed. Institutions might need to post additional collateral against positions they could not reconcile. Customers could withdraw deposits or assets while correspondent banks restricted activity pending confirmation of balances and instructions.</p><p>Demand for high-quality liquid assets could consequently rise sharply before any credit downgrade or confirmed financial loss. Market prices would then reflect the uncertainty.</p><p>Short-dated bank funding could become more expensive. Credit default swap spreads could widen before losses were established. Additional Tier 1 bank securities could weaken as investors reassessed the potential for coupon suspension, maturity extension or loss absorption, even where reported capital ratios remained above regulatory thresholds.</p><p>Three areas deserve particular attention.</p><p>First, funding desks that depend on same-day settlement and the rapid transfer of collateral could face immediate liquidity pressure. Second, custodians and payment agents could become critical points of failure where their records are required to establish ownership within a narrow time window. Third, issuers that outsource identity verification and screening could face disruption precisely when reliable checks are most important.</p><p>Concentration in enterprise software could also create correlated credit risk among investment-grade companies. A single vendor incident might force several issuers to reduce earnings guidance in the same week, not because demand had weakened, but because billing, provisioning or authentication systems had been interrupted.</p><p>Banks could face additional fraud losses and higher verification costs. If voice and remote verification systems were compromised during an increase in artificial intelligence-enabled fraud, operating expenses could rise while customers became less willing or able to transact. The resulting pressure on earnings could precede any deterioration in regulatory capital.</p><h2>Market conditions could amplify the shock</h2><p>The effect on credit markets would depend partly on prevailing valuations and liquidity conditions. Neither current credit spreads nor the default-rate forecast indicates an imminent crisis. </p><p>Under stress, risk models that appear independent in normal markets may respond to the same signals. Institutions could reduce market-making activity, issue collateral calls and sell assets to raise cash. The resulting demand for liquidity could affect government bonds and corporate credit at the same time, even though investors ordinarily regard these markets as serving different portfolio functions.</p><p>Such a move would represent an amplification of the initial disruption rather than an entirely separate financial shock. The concern is that a common operational failure could trigger defensive actions that reinforce one another.</p><h2>The Anthropic Mythos episode</h2><p>The April 2026 episode involving Anthropic&#8217;s Claude Mythos Preview provides a relevant example of the attention that advanced artificial intelligence capabilities have received from financial policymakers.</p><p>In April 2026, United States Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting with the chief executives of Bank of America, Citigroup, Goldman Sachs, Morgan Stanley and Wells Fargo. The meeting followed the release of Mythos Preview, which Anthropic described as particularly capable of identifying software vulnerabilities.</p><p>Anthropic restricted access to a small group of partners, including Amazon, Apple and Microsoft, to give them an opportunity to prepare defensive measures. The episode attracted attention from US financial policymakers concerned about the potential use of advanced artificial intelligence to exploit weaknesses in financial infrastructure.</p><p>The episode illustrates the operational concerns that can arise when powerful capabilities interact with widely shared technology dependencies. It also shows why coordination between public authorities and financial institutions may be necessary before a vulnerability results in actual losses.</p><p>The episode does not establish that a systemic financial disruption is imminent. It does, however, provide a concrete example of policymakers treating the interaction between advanced technology and financial infrastructure as a potential operational risk.</p><h2>Lessons from previous banking and market crises</h2><p>The past two decades provide several useful comparisons. Each illustrates one part of the scenario described here, rather than establishing that the same sequence of events will recur.</p><p><strong>2007 to 2008: Funding availability deteriorated before credit losses were fully recognised.</strong></p><p>Asset-backed commercial paper programmes and structured investment vehicles that had previously relied on regular refinancing were unable to roll over their funding. Interbank term lending contracted as lenders reassessed counterparties and markets. Institutions that might otherwise have remained solvent faced liquidity difficulties when common exposures to United States mortgage assets were repriced and collateral could no longer be financed on acceptable terms.</p><p>The relevant parallel is the possibility of simultaneous liquidity stress without simultaneous insolvency. Funding can disappear before spreads fully reflect the scale of the underlying problem.</p><p><strong>2011 to 2012: Common exposures linked banks and sovereigns.</strong></p><p>During the euro-area sovereign debt crisis, banks holding substantial amounts of their domestic government&#8217;s debt faced correlated funding pressure. The relationship between sovereign credit quality and bank balance sheets became a common driver of risk across institutions.</p><p>Long-term central bank refinancing operations and, subsequently, the European Central Bank&#8217;s announcement of its Outright Monetary Transactions framework helped stabilise conditions.</p><p>The parallel is that one common exposure can influence the funding costs of multiple institutions. An effective response is easier when authorities can rely on reliable records and direct support towards identifiable sources of stress.</p><p><strong>March 2020: The dash for cash affected assets normally considered liquid.</strong></p><p>Investors sold United States Treasury securities, investment-grade credit and money-market instruments simultaneously to raise cash. Dealer intermediation capacity proved insufficient to absorb the one-way flow.</p><p>Official facilities that supported dealers and purchased assets helped restore two-way trading.</p><p>The lesson is that the demand for cash can override conventional distinctions between asset classes. Liquidity pressure can cause investors to sell assets that they would otherwise retain, without any corresponding deterioration in the credit quality of those assets.</p><p><strong>September to October 2022: Collateral calls created a self-reinforcing cycle in United Kingdom government bonds.</strong></p><p>Liability-driven investment strategies came under pressure as long-dated gilt yields rose sharply. Collateral calls forced funds to sell gilts, putting further upward pressure on yields and generating additional calls. The Bank of England temporarily purchased government bonds to interrupt the cycle.</p><p>This episode offers a particularly relevant comparison because it combined leveraged exposures, collateral demands and forced selling. The feedback loop was contained through intervention by an external buyer.</p><p>The records needed to establish positions and collateral remained available throughout the episode. In the technology scenario considered here, the reliability of those records could itself be in question, potentially complicating the response.</p><p><strong>March 2023: Digital banking accelerated the movement of deposits.</strong></p><p>Silicon Valley Bank, Signature Bank and First Republic experienced deposit outflows at a pace facilitated by mobile banking and rapid communication. Credit Suisse also faced acute liquidity pressure as confidence deteriorated, despite reported capital ratios remaining above minimum requirements. Its Additional Tier 1 securities were written down as part of the arranged takeover.</p><p>The comparison has two elements. Deposits can move rapidly when customers lose confidence, and subordinated bank capital can reprice or absorb losses before a conventional solvency threshold is breached.</p><p>Taken together, these episodes show how common exposures can interact with funding mechanisms that assume more time than is available. Official intervention has generally been more effective when authorities could identify exposures, assess reliable records and direct support at a visible source of market stress.</p><p>The scenario considered here adds a further complication: the common exposure would be operational, automated processes could continue during the initial disruption, and the records needed to assess positions might temporarily be unreliable.</p><h2>What regulation already covers, and what may remain untested</h2><p>Regulatory frameworks address several components of this scenario, although the combined effects of common dependencies, continued automated execution and simultaneous liquidity demands deserve further examination.</p><p>In Europe, the Digital Operational Resilience Act has applied since January 2025. It requires financial entities to identify information technology dependencies, assess the potential effects of disruption and demonstrate that important services can remain within defined tolerances. In 2024, the European Central Bank conducted a cyber-resilience stress test focused on firms&#8217; response and recovery capabilities.</p><p>In the United Kingdom, the Bank of England, Prudential Regulation Authority and Financial Conduct Authority operate an operational resilience framework centred on important business services and their impact tolerances. A regime for designating critical third-party service providers has supplemented these arrangements since 2025.</p><p>In the United States, federal banking agencies updated their third-party risk management guidance in 2023, addressing concentrated reliance on external providers as a safety and soundness concern. The International Monetary Fund and the Financial Stability Board have also identified third-party concentration and cyber disruption as potential channels of systemic risk.</p><p>These frameworks have strengthened the way financial institutions identify and manage operational dependencies. The potential gap lies in how risks interact across the system.</p><p>Many resilience exercises focus on whether an individual institution can maintain critical services within its tolerance limits. Fewer appear to test a situation in which several institutions activate the same contingency procedures simultaneously, collateral cannot be verified, and automated systems continue to operate before the source of the disruption has been isolated.</p><p>The distinction is important for credit analysis. A system may be resilient at the level of an individual institution but less resilient when several institutions depend on the same provider and respond to the same incident in similar ways.</p><h2>What would weaken the hypothesis?</h2><p>Evidence that operational controls are keeping pace with technological capability would weaken the scenario.</p><p>Relevant indicators would include sustained reductions in software patching and recovery times, supported by independent verification; genuine diversification of critical systems, including tested manual alternatives; and limits on the ability of automated agents to execute sensitive actions without human authorization. Records should also be independently reconstructable when a primary system is unavailable.</p><p>Two consecutive quarters of such evidence would provide a useful starting point for assessing whether resilience is improving, although the period alone would not establish that the risks had been eliminated.</p><p>Until the evidence is stronger, credit investors could incorporate operational concentration into spread and maturity decisions. Issuer selection should also consider seniority and the extent to which recovery depends on reliable records, functioning settlement systems and the continued availability of third-party services.</p><p>Three indicators would suggest that the risk is increasing rather than receding: expanding autonomous execution without equivalent containment controls, growing common dependencies across institutions and technology capabilities advancing faster than patching, oversight and recovery arrangements.</p><p>For credit investors, the practical question is straightforward: <strong>which issuers could withstand three days without functioning payment and settlement systems?</strong></p><p>The answer would depend not only on cash reserves and committed liquidity facilities, but also on the issuer&#8217;s ability to establish its positions, access its collateral, maintain essential operations and meet contractual obligations while the disruption is resolved.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent Credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[The Credit Market Is Repricing the AI Boom]]></title><description><![CDATA[The sell-off in AI infrastructure debt is exposing a gap between contracted demand, the cash required to build capacity and the returns needed to justify the investment.]]></description><link>https://www.rovelliresearch.com/p/coreweave-bonds-price-real-risk</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/coreweave-bonds-price-real-risk</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Mon, 19 Oct 2026 05:47:32 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/55a1bcfe-d054-4e78-b66f-f918646b0a8a_1080x1920.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Last week, one series of CoreWeave&#8217;s unsecured notes traded at a yield above 15%. Market participants were already treating the level as distressed. The shares fell roughly 16%, to the mid-$60s, while credit default swaps traded above 800 basis points.  </p><p>At a 15% yield, the market is no longer questioning CoreWeave&#8217;s ability to grow. It is questioning whether the company can turn its backlog into cash before the cost of financing that growth overwhelms the returns.</p><p>This is no longer an isolated move in one company&#8217;s debt. Bonds financing several of the largest US data-centre projects have also come under pressure. The market is starting to distinguish between the demand for AI computing capacity and the ability of the infrastructure built to supply it to generate adequate returns.</p><h2>The wider repricing of AI infrastructure</h2><p>The bond market has begun to challenge the assumption that the revenues generated by the AI boom will be sufficient to support the enormous amounts of debt being accumulated to finance it.</p><p>The debt financing the Hyperion data-centre project, backed by Blue Owl and Meta, fell from above par in early July to below 92 cents on the dollar. Bonds linked to two projects developed by Hut 8 traded at around 94 and 92 cents, respectively. Debt financing Blackstone-owned QTS&#8217;s Project Excalibur, leased to Microsoft, fell below 88 cents.</p><p>Oracle-related projects have also experienced pressure. Approximately $18bn of loans associated with Project Jupiter in New Mexico reportedly entered stressed territory following delays involving the site&#8217;s power supply and local opposition. Bonds financing Oracle&#8217;s Michigan AI campus fell to around 87 cents on the dollar.</p><p>These projects are not interchangeable credit exposures. Their tenants, financing arrangements, collateral packages and remaining construction risks differ. But the direction of travel is difficult to dismiss.</p><p>The repricing has several potential explanations. Investors are questioning the timing of AI monetization, the scale of the capital expenditure required and the execution risks attached to power, construction and permitting. OpenAI&#8217;s decision to defer its IPO until next year and reports of slower quarterly sales growth at Anthropic have added to those concerns.  </p><h2>How much capital can the industry commit before the revenues generated by that infrastructure catch up?</h2><p>Gartner has estimated that data-centre owners would need to generate approximately $8.5 trillion in revenue by the end of 2030 to cover their capital and financing costs while achieving healthy returns. Its forecast for consumer and corporate spending on AI services that would flow through to data-center providers was no more than $3.6 trillion. Gartner has warned that around 80% of neo-cloud providers could cease independent operations by 2030 and that more than $4 trillion in asset value could be exposed to write-downs through 2032 if infrastructure investment continues on its current trajectory.</p><p>These are forecasts, not certainties. But the scale of the potential gap explains why the market is reassessing the assumption that almost any available AI capacity will eventually generate an adequate return.</p><p>The risks extend beyond demand. Political opposition to data centres is growing in the US, with concerns about electricity consumption, water, noise and local pollution delaying or complicating projects. Permitting and power constraints can postpone revenue while financing costs continue to accrue. The Bank for International Settlements has also highlighted the interconnected nature of some AI financing and commercial relationships, in which companies invest in or finance counterparties that are also their customers or suppliers.</p><p>Taken together, these developments point to a more discriminating credit market, questioning which projects can be completed on time, which operators can fund them without continually returning to the capital markets, and which assets will retain their value if demand or pricing disappoints.</p><h2>CoreWeave brings these questions into particularly sharp focus.</h2><p>CoreWeave began in September 2017 as Atlantic Crypto, a small Ethereum-mining operation in New Jersey. It pivoted towards renting Nvidia graphics processors by the hour in 2019 and 2020 and eventually adopted the CoreWeave name. The business had little non-crypto revenue before 2022, but it was well positioned to benefit from the subsequent shortage of accelerated computing capacity.</p><p>The expansion has been rapid. By August 2026, CoreWeave was operating around 1.5 gigawatts across 51 data centers, with contracted power capacity of approximately 4.2 gigawatts.</p><p>The company listed in March 2025 at $40 a share, raising approximately $1.4 billion. Nvidia, already an investor, subsequently made a further $2 billion equity investment in January 2026. Nvidia has a commercial interest in the expansion of AI computing capacity, and its involvement may help CoreWeave secure equipment and maintain access to capital.</p><p>But a strong strategic backer does not remove the underlying financing risk.</p><p>The model is straightforward to describe and demanding to finance. CoreWeave signs multiyear, take-or-pay contracts for computing capacity, acquires processors and builds the infrastructure needed to deliver that capacity. It then uses a combination of corporate debt, asset-backed facilities and project financing to fund further expansion.</p><p>The contracted revenue provides visibility. The challenge is that much of the cash needed to deliver it must be spent before the revenue arrives.</p><p>The scale of the commitments is substantial. CoreWeave reported remaining performance obligations of $60.7 billion in December 2025. Its reported revenue backlog reached $105 billion in June 2026. The company added more than $25 billion of net new commitments early in the third quarter, which were not included in that June figure.</p><p>But backlog is not cash. It must be converted into installed capacity, recognized revenue and ultimately cash available for debt service. Each step depends on execution, equipment availability, electricity supply, customer requirements and the economics of the contracts themselves.</p><p>Customer concentration adds another layer of risk. The largest customer&#8217;s share of revenue reportedly declined from around 85% towards 35% during 2025 as Meta and OpenAI expanded their relationships with CoreWeave. That diversification is positive, but the business remains dependent on a relatively small group of very large customers.</p><h2>Impressive growth, but the cash-flow burden is substantial</h2><p>CoreWeave&#8217;s operating performance has been exceptional by conventional infrastructure standards.</p><p>Revenue increased from $229 million in 2023 to $1.9 billion in 2024 and $5.13 billion in 2025. In the second quarter of 2026, revenue reached $2.575 billion, up 112% year on year. Adjusted EBITDA was $1.51billion, equivalent to a margin of approximately 59%.</p><p>Those are powerful growth and margin figures. However, they need to be considered alongside interest expense, capital expenditure and the cash required to keep building capacity.</p><p>The second-quarter operating loss was $49 million. Net interest expense was $640 million, contributing to a net loss of $626 million. Interest expense for the first half reached $1.176 billion.</p><p>Capital expenditure is the other major pressure point. CoreWeave spent approximately $9.4 billion in the second quarter and has guided to full-year capital expenditure of $35 billion to $39 billion.</p><p>The result is a business whose adjusted EBITDA margin looks attractive while its funding requirement remains enormous, generating free cash burn at roughly $10 billion for the first half, depending on the treatment of working capital and other cash-flow items.</p><p>CoreWeave is growing quickly, and its contracted capacity offers a route to much greater future earnings. But operating cash generation currently covers only a fraction of the investment program.</p><p>The company must therefore rely on access to debt and equity markets, asset-backed financing, customer commitments and strategic support to fund the difference.</p><p>That arrangement can work while financing remains available at acceptable prices. It becomes more difficult when the cost of capital rises at the same time that the company needs to refinance maturing obligations and continue investing in new capacity.</p><h2>The balance sheet explains the 15% yield</h2><p>At June 30, 2026, CoreWeave reported approximately $31 billion of recourse debt and $3.6 billion of non-recourse project debt. Operating leases add a further debt-like obligation measured in the mid-teens of billions of dollars.</p><p>Cash and restricted cash provide some liquidity, but the company&#8217;s funding requirements must be considered against its capital expenditure program and scheduled repayments. Debt maturities also create refinancing requirements over the next several years, with multibillion-dollar repayments due annually through 2028.</p><p>Annualizing second-quarter adjusted EBITDA produces gross leverage of approximately 5.9 times and net leverage of around 5 times, using the reported debt figures and adjusting for cash. On the same adjusted earnings basis, interest coverage is approximately 2.4 times.  </p><p>In short, the balance sheet depends on future growth arriving as planned.</p><p>The latest trading levels show what investors are demanding in return for bearing that risk. A yield around 15% on unsecured debt reflects deep concern about the company&#8217;s ability to fund an ambitious investment program, service its debt and preserve financial flexibility under less favorable conditions.</p><h2>Growth versus funding risk</h2><p>The downside does not require AI demand to collapse. If hyperscalers pause to digest recent investment, contract renewals bring lower prices and older processors lose competitiveness, revenue could stall at $11bn&#8211;$13bn in 2027 and $12bn&#8211;$15bn in 2028. With margins falling into the high 30s, EBITDA could settle around $5bn&#8211;$6bn and leverage remain at 6&#8211;7 times. That would make refinancing more difficult even if the business continued to grow.</p><p>I will focus on the conversion of contracted capacity into energized megawatts, recognized revenue and cash, not headline backlog alone. CoreWeave reported $104 billion of backlog at June 30, 2026, but delivery depends on power, construction, customer demand and contract economics. Customer concentration, upcoming maturities and the terms of new asset-backed loans will also show how much financing the expansion still requires and at what cost.</p><p>The wider data-center question is whether expected AI revenues can support the capital committed to build capacity. CoreWeave shows that a large backlog and rapid revenue growth can coexist with heavy cash burn, high leverage and dependence on external financing.</p><p>The recent repricing does not, by itself, prove that the AI infrastructure boom is ending. It does show that investors are distinguishing between demand for computing power and the ability of individual projects and operators to finance, deliver and earn an adequate return on that capacity. </p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Private Credit Stress Is Higher Than Default Rates Suggest]]></title><description><![CDATA[Why the Official Numbers May Understate the Problem]]></description><link>https://www.rovelliresearch.com/p/what-does-senior-secured-actually</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/what-does-senior-secured-actually</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Fri, 16 Oct 2026 05:19:32 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/59e7448d-aa13-4439-9d73-d53694629fbe_1080x1920.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Private credit is showing signs of strain, but the extent of the problem depends heavily on how stress is defined. Several widely followed measures put troubled loans at around 1% to 3% of the relevant portfolios. An analysis by J.P. Morgan&#8217;s Michael Cembalest found that the proportion of positions experiencing some form of stress had doubled from 8% in 2022 to 16%.</p><p>The analysis, published in October 2026, uses a broader definition of stress than conventional default measures.</p><p>These figures are not contradictory. They measure different things. The distinction is important for investors trying to assess whether reported default rates adequately capture the risks building up in private credit.</p><h2>The problem with measuring stress</h2><p>Private credit refers here to loans originated or held by investment funds and other non-bank lenders, often to medium-sized and sponsor-owned companies. These loans are generally not traded in the same way as broadly syndicated loans, making market-based price discovery more difficult.</p><p>Five to seven years ago, underwriting standards in parts of the private credit market were often stricter than those in the broadly syndicated loan market. Direct negotiations between lenders and borrowers could allow lenders to negotiate contractual protections and earn additional returns for providing capital through a less liquid market.</p><p>The subsequent inflow of capital has raised questions about whether underwriting discipline has kept pace with competition to deploy money. The extent to which lending terms have weakened varies across the market, and establishing the trend requires evidence on covenant terms, payment-in-kind provisions and restrictions on asset transfers. These changes should not be assumed to have occurred uniformly across private credit.</p><p>The concern is not simply that more borrowers may default. It is that the contractual and accounting measures used to identify deterioration may not capture the full extent of the problem.</p><p>A borrower that cannot meet its original repayment schedule might negotiate an extension, accept a higher interest rate or capitalise interest instead of paying it in cash. These arrangements can keep a loan performing according to its revised terms, even though the borrower is no longer meeting the original agreement.</p><p>Restructuring is not necessarily evidence of an eventual loss. It can preserve value for both borrower and lender. But repeated amendments can also defer the recognition of financial difficulties and delay the point at which a loan&#8217;s underlying value becomes clear.</p><h2>Why the reported figures differ</h2><p>Several established providers track indicators of private credit stress, including Lincoln International, Houlihan Lokey, Proskauer, S&amp;P and KBRA. Their figures cover different borrower populations and use different definitions. Some measures include covenant breaches; others focus on defaults, non-accrual loans or other specific indicators.</p><p>Consequently, a reported stress or default rate of 1% to 3% should not be interpreted as a comprehensive estimate of the proportion of private credit exposures experiencing financial difficulties.</p><p>J.P. Morgan addressed this comparability problem by applying a common set of criteria to identify stressed positions. The framework covers five categories:</p><ul><li><p><strong>Non-accrual loans:</strong> loans on which lenders have stopped recognising interest income because collection is no longer sufficiently certain.</p></li><li><p><strong>Distressed payment-in-kind interest:</strong> loans where borrowers capitalise interest instead of paying it in cash.</p></li><li><p><strong>Certain maturity extensions:</strong> loans extended by more than two years where the loan is priced below 95 cents on the dollar.</p></li><li><p><strong>Debt converted into equity following foreclosure:</strong> situations in which lenders have taken control of collateral or converted debt into an equity position.</p></li><li><p><strong>Deeply discounted or otherwise distressed positions:</strong> loans valued below 80 cents on the dollar for reasons not already captured by the other categories.</p></li></ul><p>Using this framework, J.P. Morgan found that the proportion of positions experiencing stress in the defined private credit universe had increased from 8% in 2022 to 16%.</p><p>This is a broader measure than a conventional default rate. It captures certain forms of deterioration and restructuring before they necessarily result in a recognised default or realised loss.</p><h2>What the individual indicators tell us</h2><p>The published data remain useful, provided that their limitations are recognised.</p><p>Business development companies, or BDCs, are publicly traded or non-traded investment vehicles that lend primarily to private companies. Their disclosures provide one of the more accessible windows into private credit performance. Non-accrual rates are particularly useful because they indicate when lenders have stopped recognising contractual interest as income. However, this measure can lag behind deterioration in a borrower&#8217;s finances, especially when the loan has been amended before being classified as non-accrual.</p><p>Lincoln International tracks covenant breaches and other indicators of loan performance. Its measures can identify pressure that has not yet resulted in a formal default. Payment-in-kind interest is another relevant indicator: when borrowers increasingly pay interest by adding it to the outstanding principal, the lender may continue to record income even though the borrower is not generating enough cash to meet its interest obligations.</p><p>Houlihan Lokey&#8217;s loan database provides another perspective, including differences between borrower groups and the distribution of loan valuations. A low aggregate default rate can conceal weaker performance among smaller borrowers or a concentration of positions trading at substantial discounts to face value.</p><p>Proskauer tracks defaults among loans involving its lender clients, while S&amp;P and KBRA provide additional evidence from rated borrowers and reported loan portfolios. Each dataset has value, but none should be treated as interchangeable with the others.</p><p>The available figures should not simply be averaged. Differences in coverage, definitions and reporting periods make such a calculation potentially misleading. The more useful exercise is to understand what each measure captures, identify what it omits and assess whether the indicators collectively point to a deterioration in credit quality.</p><h2>How credit deterioration unfolds</h2><p>A borrower in difficulty does not necessarily default immediately. The sequence can instead involve weaker cash generation, a breach of financial covenants, an amendment to the loan agreement, capitalised interest and a maturity extension. Only later might the loan be classified as non-accrual, written down or defaulted.</p><p>Not every restructuring follows this sequence, and an amendment can be the economically sensible solution. Nevertheless, the distinction between contractual performance and underlying financial health is important.</p><p>Private credit also presents a valuation challenge. Loans are generally less liquid than publicly traded debt, and lenders may have limited opportunities to test reported valuations through actual sales. If a loan is extended rather than sold, the market price at which an independent buyer would be willing to acquire it may remain unknown.</p><p>The risk is therefore twofold: borrowers may be deteriorating, while the extent of any eventual loss remains uncertain. A loan that continues to generate accounting income is not necessarily worth its reported value, particularly if interest is being capitalised or the maturity has been repeatedly extended.</p><h2>Three indicators deserve particular attention</h2><p>First, investors should distinguish cash interest from accounting income. Interest coverage calculated using reported earnings or accrued interest can overstate a borrower&#8217;s ability to service its debt. A company covering interest expense 1.4 times on an accounting basis but barely covering cash interest payments has substantially less financial flexibility than the headline ratio suggests.</p><p>Second, investors should examine the frequency and substance of loan amendments. A single extension may reflect a temporary liquidity problem. Repeated extensions, particularly when accompanied by higher interest rates or weaker contractual protections, may indicate that the original financing assumptions no longer hold. The relevant question is not simply how many amendments have occurred, but why they were necessary and whether the borrower has returned to sustainable cash generation.</p><p>Third, the sponsor&#8217;s willingness to contribute additional equity can provide useful information. Fresh capital from the owner can reduce leverage and demonstrate a commitment to the business. Conversely, a sponsor that declines to inject capital while seeking further payment deferrals may leave lenders bearing a greater share of the downside. Neither response is conclusive on its own: the sponsor&#8217;s resources, the business&#8217;s prospects and the terms of any proposed restructuring also need to be considered.</p><p>These indicators are most informative when assessed together. A loan with weak cash interest coverage, repeated amendments and no additional equity support warrants closer scrutiny than one showing only an isolated covenant breach.</p><h2>The implications for investors</h2><p>J.P. Morgan&#8217;s analysis offers a useful illustration of how the definition of stress changes the result. The increase from 8% to 16% indicates that a materially larger share of positions in the defined universe are experiencing some form of stress than conventional default measures alone might suggest. It does not mean that 16% of loans will default or that investors will lose 16% of their capital.</p><p>For investors, the practical task is to look beyond headline default rates. Cash interest payments, repeated loan amendments, sponsor support and the reliability of reported valuations provide complementary evidence about the quality of a portfolio.</p><p>The market may ultimately absorb the current pressure without large losses. But that conclusion should rest on improving borrower fundamentals and credible evidence of recoverability, rather than on low reported default rates alone.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Aston Martin. What Does Senior Secured Actually Mean?]]></title><description><![CDATA[The July 2026 financing improved near-term liquidity but may have weakened the position of existing bondholders relative to new lenders. The dispute over the brand&#8217;s intellectual property is central t]]></description><link>https://www.rovelliresearch.com/p/spv-checklist-the-cut-material</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/spv-checklist-the-cut-material</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 13 Oct 2026 05:45:50 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/4a0fdd94-436c-4286-8c50-30f7b91c6b49_1080x1920.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Aston Martin raised &#163;550 million in July 2026 through a new financing arranged by funds managed by HPS, which has been owned by BlackRock since July 2025. The facility comprises &#163;450 million of funded debt and a &#163;100 million delayed-draw commitment, priced at SONIA plus 6.75% and maturing in July 2031.  </p><p>The proceeds repaid and cancelled the group&#8217;s &#163;170 million super senior revolving credit facility, with the balance available for transaction costs and general corporate purposes. The transaction also leaves room for a further &#163;100 million of junior debt.</p><p>The financing addresses an immediate liquidity need, but it raises a separate question: how much protection do the group&#8217;s existing bondholders retain against the assets that could support a recovery if Aston Martin encounters financial difficulties?</p><p>The issue centers on the group&#8217;s brand intellectual property. According to a memorandum prepared by existing creditors challenging the transaction, 195 UK trademarks, 69 EU trademarks, 474 UK designs and 558 EU designs were transferred from Aston Martin Lagonda Limited to SilverCo, a newly established Cayman Islands subsidiary. The existing fixed charge over these assets was released on 22 July. SilverCo did not become a guarantor under the existing bond indenture, and the memorandum states that existing note holders received no replacement security over the transferred intellectual property.</p><p>Aston Martin disputes the characterization of the transaction, arguing that no brand intellectual property left the restricted group, that existing note holders retain their share pledge over Aston Martin Lagonda Limited and that the financing was permitted under the existing documentation.</p><p>A transaction can comply with contractual permissions while still changing the relative economic position of creditors. The relevant question is not simply whether the bonds remain labelled senior secured, but which assets secure each claim, which legal entities own those assets and how competing claims would rank in an insolvency.</p><h2>The dispute over the collateral</h2><p>The dispute has moved beyond correspondence. On 24th August 2026, Arini and Tresidor, investment firms holding Aston Martin bonds, sought permission to gather evidence in the United States against HPS, the private credit lender that arranged the new financing; Authentic Brands, a brand management company; and Moelis and Lazard, investment banks involved in the transaction. The evidence is intended to support a potential claim in England under section 423 of the Insolvency Act, which can be used to challenge transactions alleged to have put assets beyond creditors&#8217; reach. Arini and Tresidor had sent a formal letter before claim on 2nd August.</p><p>The US proceedings are intended to gather evidence, not decide whether the transaction was unlawful. At the time of writing, the English claim had not been filed and no court had ruled on the allegations. The concern for existing bondholders is that moving the brand&#8217;s intellectual property and releasing the previous security could reduce the assets available to support their claims if Aston Martin defaults. The outcome depends on the legal effect of the transfer and the security arrangements. It is therefore too early to conclude that existing bondholders have lost their rights over the brand.</p><h2>How the debt ranks</h2><p>The capital structure is easier to understand when the claims are considered separately.</p><p>The first change is the cancellation of the &#163;170 million super senior revolving credit facility. Its repayment removes an existing senior claim and provides immediate liquidity relief by replacing it with new financing.</p><p>The new facility comprises &#163;450 million of funded senior secured term debt and a further &#163;100 million that can be drawn later, subject to the relevant conditions. Both mature in July 2031. The creditors&#8217; memorandum states that the delayed draw requires a transfer of control over part of the brand intellectual property. This feature warrants attention because it could affect the collateral available to existing creditors if the additional commitment is drawn.</p><p>The existing notes sit below the new financing in the potential economic recovery analysis, although the exact legal ranking depends on the contractual security arrangements. They comprise $1,050 million and &#163;565 million of senior secured notes issued by Aston Martin Capital Holdings Limited and due in March 2029. The dollar notes carry a 10.000% coupon and the sterling notes a 10.375% coupon. Both were issued under an indenture dated March 2024.</p><p>The notes retain their senior secured designation and the share pledge over Aston Martin Lagonda Limited, according to the company&#8217;s position. However, the creditors&#8217; memorandum argues that they no longer benefit from the direct fixed charge over the transferred brand intellectual property. That potential loss of direct collateral protection is the central issue in the dispute.</p><h2>Why the brand matters to recoveries</h2><p>Aston Martin&#8217;s factories, tooling and other physical assets provide some support for creditors, but their value in a stressed sale may be limited by depreciation, specialized use and the condition of the automotive market. The brand is potentially more important because it supports pricing power, customer demand and the commercial value of the business. The value of the trademarks and designs depends on the continued ability to manufacture, market and sell vehicles, as well as on the costs and legal arrangements associated with exploiting those assets.</p><p>Nevertheless, the allocation of intellectual property between legal entities can influence which creditors have access to the associated value. If the new financing benefits from stronger rights over the brand while existing bondholders rely principally on a share pledge, the relative recovery position may differ materially from what the senior secured label alone suggests.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent Credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Anthropic’s Debt Problems]]></title><description><![CDATA[When a frontier AI lab becomes a credit story]]></description><link>https://www.rovelliresearch.com/p/anthropic-goes-public-when-a-lab</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/anthropic-goes-public-when-a-lab</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Fri, 09 Oct 2026 06:41:32 GMT</pubDate><enclosure url="https://substack-post-media.s3.amazonaws.com/public/images/27680a80-44e3-4589-8b42-e8bdcaaf57fd_800x1000.jpeg" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Anthropic is approaching the stock market at a moment when its need for money is becoming almost as important as its growth. The company filed a confidential draft S-1, the document used to prepare for an IPO, on June 1, 2026. A reported leak of the later prospectus now gives investors a first look at how the business works financially. I think the central question is not whether customers want Anthropic&#8217;s products. They clearly do. The question is whether Anthropic can earn enough cash to pay for all the computers, data centers, employees and other costs needed to keep growing. I therefore look at Anthropic first as a credit story, meaning I focus on how it will pay what it owes, and only then as an equity story.</p><p>The growth numbers are remarkable, but I need to compare that growth with the amount of revenue the company is actually producing. Anthropic raised $30 billion in February 2026 at a $380 billion post-money valuation. At that time, it said its annualized revenue run-rate was $14 billion. In May, it raised another $65 billion at a $965 billion valuation, after the annualized revenue run-rate had passed $47 billion. In April, the company had already said the run-rate was above $30 billion. It also reported more than 1,000 customers generating at least $1 million of annualized revenue. Later reports put the annualized pace at $65 billion by the end of July. An annualized pace is simply the latest sales rate multiplied by twelve months. It tells me what revenue would look like over a full year if that recent pace continued. It is not the same as revenue actually earned during the year. Q2 preliminary revenue was $11.5 billion, compared with $4.73 billion in 1Q26 and $4.6 billion for all of 2025. I have little doubt that demand for Anthropic&#8217;s products is growing very quickly. What concerns me is that the cost of serving those customers is also extremely high. Reported 2025 operating loss was about $8.1 billion on operating costs of roughly $13 billion. The reported net loss of about $42 billion included roughly $34 billion of non-cash charges linked to funding. Using the operating figures, Anthropic spent about $2.76 for every $1 of revenue in 2025. In simple terms, the company spent almost three dollars to generate one dollar of sales. I think growth gives Anthropic time, but it has not yet shown that the business can fund itself.</p><p>The next issue is how Anthropic plans to pay for that growth. Reuters reports about $518 billion of planned spending and commitments over roughly ten years. A large part of this appears to be tied to contracts that cannot easily be canceled. Anthropic also reportedly signed a five-year lease worth $125.2 billion for specialized AI chips called TPUs. TPU stands for Tensor Processing Unit. These are computer chips designed specifically for the huge amount of processing needed to train and run AI systems. The lease works out to an average of about $25 billion a year. A Broadcom loan of up to $42 billion in convertible notes would help finance roughly one-third of that lease. Anthropic has also been reported to be close to a $15 billion revolving credit facility led by Morgan Stanley. A revolving credit facility is essentially a large bank line that the company can draw when it needs cash and repay later. I would look at these obligations in order of priority. The bank facility comes first. The convertible notes come next. Then come leases and other contracts that require Anthropic to keep paying for computing capacity. There is also a reported $35 billion financing arrangement involving Apollo and Blackstone to support additional computing capacity. What makes the structure unusual is that some of Anthropic&#8217;s biggest suppliers are also its biggest customers and investors. Amazon and Google accounted for 47% of 2025 sales, or about $2.16 billion. At the same time, Anthropic committed more than $100 billion to Amazon&#8217;s AWS cloud business over ten years. Amazon agreed to invest $5 billion initially and potentially another $20 billion. Google and Broadcom are also involved in providing the computing infrastructure Anthropic needs. I think these relationships make the business more connected and therefore more vulnerable if one part of the system comes under pressure.</p><p>Cash is the next question, and this is where I still do not have enough information. The reported prospectus showed about $20.3 billion of cash. That is a large amount for most software companies. It is much less impressive against the commitments Anthropic has reportedly made. The five-year $125.2 billion TPU lease alone averages about $25 billion a year. That is before paying Amazon for cloud services, buying other computing power, training new models, paying employees and covering financing costs. At a $47 billion annualized revenue pace, I can illustrate the problem with a simple assumption. If Anthropic earned a 35% gross margin, it would produce about $16.5 billion of gross profit. That is my estimate, not company guidance. Even under that generous assumption, the annual TPU payment would be larger than the gross profit. The reported $518 billion of commitments would also average about $51.8 billion a year over ten years, although the actual payments could be spread unevenly. I therefore do not want to treat the entire $518 billion as an immediate bill. The full S-1 should tell me when these payments are due, which contracts can be canceled or delayed, how the bank facility works, and what conditions Anthropic must meet to keep borrowing. The fact-check also flags the reported $65 billion July revenue figure because the exact way that number was calculated has not yet been confirmed. These details are important because they determine how much financial pressure Anthropic will face and when.</p><p>I think the IPO is therefore a test of more than Anthropic&#8217;s growth. The optimistic case is easy to understand. Customers continue to spend heavily on AI, the cost of running each AI query falls, Anthropic uses its computing capacity more efficiently and profit margins improve quickly. I give that scenario real weight because the company&#8217;s sales growth is exceptional. But my base case still requires several things to happen at the same time. Revenue must keep rising very rapidly, margins must improve, the computers Anthropic has committed to must be used efficiently, and investors and banks must remain willing to provide more money. A weaker period could create problems very quickly. Lower sales would leave more unused computing capacity. That would put pressure on cash generation and make new financing more expensive. I also want to know how much revenue comes directly from customers and how much comes through Amazon and Google, because those companies are also major suppliers to Anthropic. Anthropic may build one of the most valuable technology businesses of the decade. I think investors can justify a very high equity valuation if growth and margins remain exceptional. But I would not assume that a great product automatically creates a safe investment. The numbers currently show a company with $4.6 billion of 2025 revenue, an $8.1 billion operating loss, a $125.2 billion TPU lease and $518 billion of reported long-term commitments. <strong>My recommendation is to understand the financing burden before chasing the IPO valuation.</strong></p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent Credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Banks vs Private Credit]]></title><description><![CDATA[Banks are becoming more selective on credit. Good borrowers can still refinance, but weaker sectors are increasingly turning to private credit as bank financing becomes harder to secure.]]></description><link>https://www.rovelliresearch.com/p/how-a-small-team-watches-twelve-borrowers</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/how-a-small-team-watches-twelve-borrowers</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 06 Oct 2026 05:36:34 GMT</pubDate><content:encoded><![CDATA[<p>The latest ECB Bank Lending Survey points to continued selectivity among European banks. That makes the question behind this article more relevant: when a bank declines a refinancing, who steps in, and on what terms? I have watched this process for 25 years. In 2007 and 2008, I saw financing markets close for companies that had looked fine only a quarter earlier. The lesson stayed with me: when banks pull back, access to funding changes before pricing fully reflects the shift. Private credit can fill that gap, but it does not provide the same economics or the same protection to existing bondholders.</p><p>The bank market still works for borrowers that fit its risk criteria. A net 7% of euro area banks tightened lending standards for companies in 2Q26, down from 10% in 1Q26, while banks expected another 5% tightening in 3Q26. Autos and energy-intensive manufacturing saw the sharpest tightening. Investment-grade revolvers still clear, while single-B cyclicals often need shorter maturities or wider pricing. The maturity wall points to the same split. Europe reduced its 2028 loan wall by 57.7%, from &#8364;90.9 billion to &#8364;38.4 billion by July 2026. Refinancings and extensions represented 67% of &#8364;110.6 billion of year-to-date loan volume, excluding repricings. Loans due in 2027 fell to &#8364;3.7 billion from &#8364;9.0 billion at end-2025. I see this as evidence that stronger borrowers are still refinancing through banks and syndicated markets, while weaker credits increasingly need another source of capital.</p><p>That is where private credit differs from bank financing. Banks can offer lower-cost funding to borrowers that fit their underwriting framework, but they also face balance-sheet constraints, syndication requirements and limits on sector risk. Private lenders can structure around more complex situations and provide financing when banks are unwilling to extend enough capital, but they demand a higher return and stronger protection. The borrowers left on the maturity wall increasingly fall into that category. The share trading below 90 cents rose from 33.4% in April 2026 to 39.4% in July. Software&#8217;s share of the wall rose from 12.5% to 16.4%, despite a lower euro amount. Builders&#8217; merchants also screen weak. Huws Gray&#8217;s 2028 term loan traded around 70-76 in March, while STARK Group&#8217;s 2028 loan traded around 72-82. BME&#8217;s 2029 term loan also remained under pressure. Cegid shows how the two markets can coexist: the French software group still had &#8364;1.04 billion of syndicated debt due in 2028 when it secured a new &#8364;1.1 billion private credit facility in June 2026 to finance its acquisition of Shine. Aston Martin shows the implications for existing creditors. HPS Investment Partners led a &#163;550 million financing in July 2026 secured against assets in a newly incorporated subsidiary. Existing creditors subsequently challenged the transfer of intellectual property. The new lender obtained stronger security, while the recovery position of existing creditors came under pressure.</p><p>Private credit therefore does not simply replace a bank loan. It changes the balance between price, flexibility and creditor protection. More than 90% of private credit loans are floating rate, so higher base rates feed directly into interest expense. PIK, or payment-in-kind interest, allows a borrower to add interest to the loan balance instead of paying cash. That preserves liquidity temporarily, but increases debt and the eventual refinancing burden. For bondholders, I therefore look beyond the coupon to the full capital structure. Revolvers and first-lien loans rank ahead of senior unsecured bonds, while junior and shareholder loans rank behind them. Operating leases also absorb cash before equity. In a weak auto supplier or chemical company, I want to know exactly who controls the collateral, how much secured debt sits ahead of the bond and what recovery value remains for unsecured creditors. A 7% coupon does not compensate me for poor priority if most of the enterprise value is already pledged to lenders above it. The private lender may have solved the borrower&#8217;s funding problem, but the bondholder can still end up with a weaker position.</p><p>That is the real difference between the two funding markets. Banks remain the lower-cost option for borrowers that meet their risk criteria. Private credit becomes relevant when the borrower needs more flexibility, a bespoke structure or capital that banks are unwilling to provide. The trade-off is higher pricing and stronger lender protections. European high-yield spreads remain tight relative to their long-term range, so I do not want to take that additional structural risk for a modest incremental yield. I stay senior and short in cyclical sectors such as autos, chemicals and weaker software names. I want no hard maturity inside 18 months unless cash or a tested revolver clearly covers it. Beyond five years, I need enough coupon and covenant protection to compensate for refinancing risk. When I assess a bond, I therefore ask two questions: would a bank still fund this borrower today, and if not, what would a private lender demand in return?<br><br><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Five Things to Check Before Investing in a CLO]]></title><description><![CDATA[Last week I explained how to read the key signals in a CLO trustee report. This week I look at what those signals mean for an allocator deciding whether to invest.]]></description><link>https://www.rovelliresearch.com/p/when-the-valuation-arrives-late-what</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/when-the-valuation-arrives-late-what</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Fri, 02 Oct 2026 12:45:33 GMT</pubDate><content:encoded><![CDATA[<p>First of all, a quick reminder of the basics, because you do not need to be a CLO specialist to follow the analysis.</p><p>A leveraged loan is a senior secured loan to a company in the high-yield universe. It normally pays a floating interest rate. A CLO buys a large portfolio of these loans, typically 150 to 250, and finances the portfolio by issuing several layers of debt with different levels of risk and return.</p><p>The CLO manager can generally reinvest principal for four to five years. During that period, the quality of the underlying loans can change materially, even if the CLO itself continues to carry the same ratings.</p><p>When I look at a CLO, I am essentially trying to answer five questions.</p><p><strong>First, is the structure still providing enough protection?</strong> I look at the cushion above the overcollateralization triggers and how quickly it is changing.</p><p><strong>Second, is the loan portfolio deteriorating?</strong> The percentage of loans rated CCC or below is one of the first places I look.</p><p><strong>Third, is the portfolio generating enough cash?</strong> Interest coverage can tell us whether there is enough interest income to support the CLO&#8217;s payment obligations, but the quality of that income also matters.</p><p><strong>Fourth, what is the manager doing about weaker loans?</strong> Trading activity can tell us whether the manager is actively managing deterioration or allowing weaker credits to remain in the portfolio.</p><p><strong>Fifth, how should an allocator get exposure?</strong> The choice is not necessarily between one CLO and another. It may be between owning a few CLOs directly and investing through a CLO fund.</p><p>You do not need a full CLO model to answer these questions initially. The trustee report usually gives you enough information to identify whether the structure, the collateral or the cash generation is beginning to deteriorate, and whether a deeper analysis is warranted.</p><h2>1. Do I need a full CLO model to know if something is wrong?</h2><p>No. A full CLO model recreates the entire transaction, including the loan portfolio, the different debt tranches and the rules governing how cash moves between them. It is useful for testing what happens under different default and recovery scenarios. But you can identify early signs of deterioration from the trustee report without rebuilding all of this.</p><p>I start with the trustee report and look at five things: the outstanding tranche balances, the overcollateralization tests, the interest coverage tests, the size and direction of the lowest-rated loan bucket, and the manager&#8217;s trading activity during the period.</p><p>The change over time is often more informative than the level itself.</p><p>Suppose the senior overcollateralization cushion is eight percentage points above its trigger and falls to six, then four, over two reporting periods. The transaction has not necessarily breached a test. But the protection available to the senior tranche is clearly declining.</p><p>By contrast, a cushion that stays around five points for several quarters tells me something different. The level may not be generous, but the structure is stable.</p><p>My rule is simple. One weak number can be noise. Two numbers deteriorating together deserve investigation. Three consecutive periods of deterioration usually justify a much deeper review.</p><p>The trustee report is therefore a triage tool. It tells me whether I need the full model, not whether I need it before I can form a view.</p><h2>2. What does the overcollateralization cushion actually tell me?</h2><p>Overcollateralization is one of the main protections built into a CLO.</p><p>In simple terms, it compares the value of the loan collateral with the debt that the CLO has issued. If the collateral loses value through defaults, losses, or other adjustments, the ratio falls.</p><p>Each tranche has a relevant test level. The difference between the actual ratio and that level is the cushion.</p><p>For example, if the relevant test level is 120 percent and the current ratio is 125 percent, there are five percentage points of protection above the trigger. If the ratio falls to 122 percent, only two points remain.</p><p>Why does is this important?</p><p>If a test is breached, the cash waterfall can change. Cash that would normally have flowed to junior investors can instead be redirected to repay more senior debt.</p><p>That protects senior investors, but it can materially reduce cash flows to mezzanine and equity investors.</p><p>So I do not look at the ratio in isolation. I look at the distance to the trigger and the speed at which that distance is changing.</p><p>A stable five-point cushion can be manageable. A cushion falling from eight points to four points in two reporting periods deserves much more attention.</p><p>The next question is what caused the decline. Defaults are an obvious possibility. But downgrades, loans sold below par and other changes in the collateral can also reduce the protection.</p><h2>3. What does a 7% CCC bucket really mean?</h2><p>The 7% figure refers to the <strong>s</strong>hare of the CLO's collateral pool that can be rated CCC+ or below before the structure starts applying a penalty in its overcollateralization tests. It is not a 7% default rate, and it does not mean that 7% of the loans have already defaulted.It means the quality of the collateral has deteriorated enough that the structure may start to feel the effect. Loans rated CCC are generally the weakest part of the performing collateral pool. They have a much higher probability of default than higher-rated loans.</p><p>The precise treatment of these loans is transaction-specific. In many CLOs, there is a threshold above which excess CCC exposure receives less than full credit in certain collateral tests. That means an increase in CCC exposure can affect the reported protection even before a loan actually defaults.</p><p>Consider a simple example. If the CCC bucket rises from 4% to 6%, the portfolio is clearly weaker, but the structural impact may still be limited. If it rises from 4% to 8% in one quarter, I want to know why. Was the increase caused by a broad deterioration across the portfolio? Did several loans get downgraded at the same time? Is the exposure concentrated in one sector?</p><p>I also look at what the manager did. If the manager sold a deteriorating loan at 98 before it was downgraded, that may indicate good trading discipline and an effort to limit downside risk. If the manager retained the loan and it subsequently moved into CCC, it may indicate that the manager was slower to react to the deterioration. The same increase in the CCC bucket therefore does not necessarily tell the same story about the manager's credit selection and risk management.</p><h2>4. How do I read interest coverage without building a model?</h2><p>Interest coverage asks a simple question: does the loan portfolio generate enough interest income to meet the payments required by the CLO structure? The reported ratio can therefore look healthy while the underlying cash generation is becoming less comfortable.</p><p>I start with the reported interest coverage cushion and then ask three questions.</p><p>First, how much of the reported income is actually cash interest? Some loans can pay part of their interest in kind. That increases reported interest income without producing the same amount of cash.</p><p>Second, how durable is the spread income? A CLO may currently own loans with attractive spreads, but those loans can refinance at tighter spreads. The income available to the structure can then decline.</p><p>Third, has the manager introduced fixed-rate assets into a predominantly floating-rate liability structure, and if so, is the interest-rate risk properly hedged?</p><p>This is why I distinguish between reported coverage and cash-pay coverage. If reported interest coverage looks comfortable but the amount of cash interest available to the structure is much lower, I would treat the reported cushion as less robust.</p><p>That does not mean the CLO is about to fail. It means there is less cash available to absorb deterioration in the loan portfolio.</p><h2>5. Should I buy individual CLOs or a CLO fund?</h2><p>For a wealth manager or family office, this may be the most important question of all. The choice is not simply between two ways of getting CLO exposure. It is a choice between control and diversification, but also between doing the credit work yourself and delegating it to a fund manager.</p><p>Buying individual CLO tranches gives you much more control. You can choose the manager, transaction, tranche, maturity and price. You can also decide how much credit risk and structural protection you are willing to accept. But that control comes with a cost: you have to analyze each transaction yourself, monitor it over time and accept the concentration that comes with owning a relatively small number of positions.</p><p>A CLO fund solves part of that problem. It can diversify across dozens or even hundreds of CLOs, managers, sectors and tranches, while allowing investors to enter with a much smaller ticket. The trade-off is that you give up some control and pay another layer of fees.</p><p>The fund can therefore be the better vehicle without necessarily being the better investment. The important question is what you are actually getting for the additional fee and delegation.</p><p>I would focus on four questions.</p><p>First, how diversified is the portfolio?</p><p>Do not stop at the number of CLOs in the portfolio. Look at how much exposure the fund has to its largest managers, individual transactions and underlying sectors. A fund holding 50 CLOs may still have significant concentration if several positions are managed by the same large managers or have similar exposure to the same industries.</p><p>Diversification should reduce the impact of an individual credit problem. If a single manager, transaction or sector can materially affect the fund&#8217;s performance, the apparent diversification may be less useful than it looks.</p><p>Second, what are you actually buying?</p><p>The headline rating of the fund tells you very little about the risks inside the portfolio. You need to understand which CLO tranches the fund owns and where they sit in the capital structure.</p><p>A fund concentrated in AAA and AA CLO tranches has a very different risk profile from one investing in BBB or BB tranches, even if both funds have attractive headline yields. I would also look at the underlying CLOs&#8217; overcollateralization cushions, CCC exposure, portfolio quality and remaining reinvestment period.</p><p>Third, where is the return coming from?</p><p>A high distribution yield does not automatically mean that the underlying credit is attractive. I would want to separate the return generated by the underlying CLO tranches from the effects of leverage, fund fees, trading gains and changes in market prices. A fund can produce a strong recent return because it benefited from falling spreads or successful trading, rather than because the underlying credit risk is particularly well compensated.</p><p>Fourth, how much can I actually see?</p><p>This is particularly important for a private wealth investor. A fund may provide diversification, but you are relying on the manager to select and monitor the underlying CLOs. How frequently does the fund disclose its holdings? Can you see the individual CLO managers and tranches? Does it report changes in CCC exposure, overcollateralization cushions and deteriorating positions? Does it explain why it sold a position or increased exposure to a particular manager?</p><p>The more limited the transparency, the more difficult it becomes to determine whether the fund is actually doing something you could not replicate yourself.</p><p>This is where independent credit analysis can add value. The question is not simply whether the fund manager has produced a good track record. It is whether the portfolio is taking risks that you understand, whether you are being adequately compensated for those risks, and whether the additional fees paid to the fund manager are justified by the diversification and credit selection you receive in return.</p><p>For an investor with the expertise and resources to analyze CLOs directly, individual tranches may offer greater control and potentially better value. For an investor who cannot realistically monitor a large CLO portfolio, a fund may be the more practical solution.</p><h2>When should I ask for a second opinion?</h2><p>There are three situations when I think an independent review can be particularly useful. The first is before adding to a CLO position that you already own. The question is not simply whether the tranche remains investment grade and offers an attractive spread. You want to know whether the protection supporting that rating is improving or deteriorating.</p><p>The second is when a tranche goes on negative watch or receives a downgrade. The rating action is important, but the next question is why it happened and whether the underlying collateral is still deteriorating.</p><p>The third is when two or more indicators move in the wrong direction together.</p><p>For example, suppose the CCC bucket increases sharply while the overcollateralization cushion falls. Those two numbers are not independent. The same deterioration in the loan portfolio can affect both collateral quality and structural protection.</p><p>That is when I want to understand what is happening at loan level.</p><p>My output in this situation would not normally be a 50-page model. It would be much shorter: where the transaction stands today, which indicators have changed, what would have to happen for a structural test to become a problem, and whether the current price compensates you for that risk.</p><p>If the answer is unclear, that is when I would go deeper into the trustee report, offering circular and historical payment dates.</p><p>Most of the time, public information is enough to determine whether deeper work is justified.</p><h2>What would change my view?</h2><p>The risk I would watch most closely is a broadening of loan deterioration.</p><p>If downgrades remain concentrated in weaker sectors, the problem may remain manageable at individual transaction level. If deterioration spreads into more defensive sectors, falling cushions could become a much broader structural issue.</p><p>I would also watch how managers deal with weak credits. If loans are repeatedly amended, extended or otherwise kept alive without a corresponding improvement in fundamentals, the reported CCC bucket may look relatively stable while other measures of credit quality deteriorate.</p><p>That is why I do not treat any single trustee-report metric as decisive.</p><p>The value is in connecting the numbers.</p><p>My previous article explained where to look. This one is about how to interpret what you find, and how that analysis can inform the actual investment decision.</p><p>Whether you choose individual CLOs or a CLO fund, the underlying question is the same: are you being paid adequately for the credit and structural risk you are taking?</p><p>That is also where I think independent credit research can be useful. Not as a substitute for the manager, but as a way of challenging the assumptions behind the investment.</p><p>If you own a CLO fund and there is one position you keep coming back to, send me the name and what concerns you about it. I read every reply.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p><p></p>]]></content:encoded></item><item><title><![CDATA[European Credit is Priced for Perfection]]></title><description><![CDATA[Spreads are tight, the easy refinancings are done, and banks are still tightening. Carry still works, but only if you get paid for tenor.]]></description><link>https://www.rovelliresearch.com/p/european-credit-is-priced-for-perfection</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/european-credit-is-priced-for-perfection</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 29 Sep 2026 11:43:39 GMT</pubDate><content:encoded><![CDATA[<h1>Carry, Not Compression</h1><p>European credit is entering late September with a less comfortable message than the headline yield suggests. Euro high yield cash closed around 269bps over government bonds in mid September 2026, with an all-in yield of 6.1%. The iTraxx Xover 5 year CDS index closed at 290bps on 24 September 2026. That is above the 1 year average of 264bps and well above the 239bps low on 6 July, but below the 362bps high on 27 March. We read that as a market that still supports credit risk, but not being paid for a major improvement in fundamentals.</p><p>The cash market still leaves little room for another big spread rally. We think that is the key point for income investors: the 6.1% yield looks appealing, but only compensates for credit risk before losses and fees. Further tightening would add a one-off gain, but the starting spread is already low. The risk is not that the market is about to collapse. It is that investors have become used to earning a good headline yield while being paid relatively little for deterioration in credit quality. BB and single-B spreads have also moved closer together in 2026. We would not expect lower-quality borrowers to deliver much extra return simply because they carry more risk. This is a market where security selection has to do more of the work.</p><p>The refinancing picture is better than it was six months ago, but the easy work is largely done. Europe cut its 2028 loan maturity wall by 57.7% to &#8364;38.4 billion by end-July 2026, while the 2029 wall fell by 30.1% over the same period. Refinancings and extensions accounted for 67% of &#8364;110.6 billion of loan volume year to date in 2026, excluding repricings. We think this is genuine de-risking, but it is concentrated in borrowers who could access markets. The remaining debt is weaker: the share of loans below 90 cents rose to 39.4% at the end of July 2026 from 33.4% in April 2026. Smaller private-credit borrowers are also showing greater stress, with defaults of 3.6% by count in 2Q26 versus 2.5% across the full market. Higher coupons are adding pressure because more than 90% of these loans were floating-rate in 2Q26. Some stressed borrowers are now paying interest with more debt rather than cash.  </p><p>Banks are not shutting the door. ECB data show corporate lending standards still tightening, but at a slower pace: 7% of banks reported tighter standards in Q2, down from 10% in Q1, with 5% expecting further tightening in Q3. Private credit is filling part of that gap, but at a price. For bond investors, the order of repayment becomes more important when refinancing gets harder. Revolvers and first-lien loans sit ahead of unsecured bonds, while junior and shareholder instruments sit behind them but can still drain cash before deleveraging begins. We also see a slower building of risk around AI infrastructure in Europe. Utilities, telecoms, and data-center projects can absorb large amounts of capital before revenue starts. A delayed power connection or tenant can leave debt service running while cash generation waits. We think that is a more credible medium-term credit risk than a sudden jump in defaults.</p><p><strong>Our recommendation towards year-end is to favor carry with a clear margin for error, rather than a strategy built around much tighter spreads.</strong> We prefer shorter-dated, higher-quality credit and liquidity that can withstand a tougher refinancing window. We are less comfortable with long-dated, low-coupon bonds that depend on a benign market in 2028-2031. The refinancing wall has become smaller because stronger borrowers acted early; that does not mean weaker borrowers have fixed their balance sheets. We would watch three things into year-end: the new issue pricing, weaker credits that need to refinance, and free cash flow after capex and dividends. At 290bps, Xover has moved wider from the summer lows but is still far from stressed. Carry should work into year-end, but investors need to be paid for taking longer-dated risk.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p><p></p>]]></content:encoded></item><item><title><![CDATA[Looking Inside a CLO]]></title><description><![CDATA[A CLO can carry an investment-grade rating and pay an appealing spread, yet the real risks often sit inside the portfolio. Trustee reports contain the clues.]]></description><link>https://www.rovelliresearch.com/p/looking-inside-a-clo</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/looking-inside-a-clo</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Fri, 25 Sep 2026 12:39:46 GMT</pubDate><content:encoded><![CDATA[<p>CLO tranches have become a familiar part of diversified credit portfolios. They often sit under labels such as structured credit or alternative fixed income, usually in a senior tranche with an investment-grade rating and a yield of 150 to 250 basis points over base rates.</p><p>On paper, that can look like a fairly conservative way to pick up additional income. The complication is that the rating and the headline yield tell you relatively little about what is happening inside the collateral pool. We think this is one of the less visible analytical gaps in many wealth portfolios.</p><p>CLOs are now one of the largest sources of funding for leveraged companies. They are also an asset class that requires a different type of credit analysis from conventional corporate bonds. An investor can own a highly rated CLO tranche without having a clear view of the underlying loans, the structural protection available to the tranche, or how the waterfall would behave if credit conditions deteriorated.</p><p>CLO analysis is difficult for fairly specific reasons. A CLO is a structured vehicle backed by a portfolio of leveraged loans. The cash flow waterfall determines how interest and principal move through the different tranches, while the transaction documents contain a detailed set of tests, limitations and provisions. The documentation can run to hundreds of pages.</p><p>The underlying loans create another layer of complexity. Unlike public corporate bonds, they often do not have the same level of publicly available financial information or trading liquidity. Many are middle-market credits. Financial reporting can be limited, and loan prices can be less informative than the prices of actively traded bonds.</p><p>There is also no equivalent of a simple Bloomberg screen that brings together the fundamentals, spread, rating history and capital structure of the entire portfolio. Instead, much of the relevant information is contained in a trustee report. These reports are generally published monthly or quarterly and contain loan-level information, tranche balances, collateral quality metrics and the results of the structural tests.</p><p>The problem is that the data is often presented in dense PDFs rather than in a format that makes portfolio monitoring straightforward. Most wealth platforms do not have direct access to the specialist data feeds that turn this information into something easily searchable and comparable.</p><p>The cash flow modelling presents a further barrier. Understanding how each tranche performs under different assumptions for defaults, recoveries, prepayments and interest rates requires specialist software. Intex and Clarity are widely used across the structured-credit market, but both require expertise as well as a financial commitment.</p><p>CLO analysis therefore sits somewhere between traditional credit analysis, structured finance and legal interpretation. The specialists who do this work every day tend to sit at CLO managers, rating agencies and dedicated credit funds rather than within wealth platforms serving private clients.</p><p>That does not mean the investor is left without useful information.</p><p>You do not need a full CLO modelling platform to identify whether a transaction deserves closer attention. The trustee report contains much of the information required. What matters is knowing which numbers to look at and, more importantly, how they interact.</p><p>We would focus on six areas.</p><ol><li><p><strong>Tranche ratings and watch lists.</strong> Moody&#8217;s and S&amp;P publish rating actions on CLO tranches. A downgrade or a move to negative watch does not automatically mean the tranche is impaired, but it is a clear signal that the perceived risk has changed. Rating actions should be checked directly rather than relying solely on the rating displayed by a wealth platform.</p></li><li><p><strong>Overcollateralization tests.</strong> This is one of the most important structural protections in a CLO. The overcollateralization ratio measures the collateral supporting the outstanding debt. If the ratio falls below the relevant trigger, cash that would otherwise flow to junior tranches can be redirected towards senior tranches. The current ratio, trigger level and available headroom are disclosed in the trustee report. We would pay particular attention when the ratio begins to approach its trigger rather than waiting for an actual breach.</p></li><li><p><strong>Weighted average life tests.</strong> CLOs also have tests designed to prevent the portfolio from extending too far. The weighted average life can increase when loans are not repaid as expected or when the manager adds longer-dated assets. A deterioration here can become relevant because a longer portfolio life changes the refinancing and credit-risk profile of the transaction.</p></li><li><p><strong>Collateral quality metrics.</strong> Trustee reports normally provide information on the average rating of the collateral, the percentage of CCC-rated loans, industry concentrations and defaulted or recently downgraded credits. These figures are more useful when viewed as a series rather than as isolated observations. A single weak quarter may not change our assessment. A deterioration that persists for three or four quarters is more difficult to dismiss.</p></li><li><p><strong>Interest coverage tests.</strong> The interest coverage test provides another view of the health of the structure. It measures whether the collateral is generating sufficient interest income to meet the relevant obligations higher up the waterfall. A declining ratio can indicate that the income generated by the pool is becoming less comfortable relative to the payments and expenses that need to be covered.</p></li><li><p><strong>Default and downgrade rates.</strong> Defaults and downgrades are normal features of leveraged loan portfolios. The question is whether the trend is deteriorating. A pool where the share of credits moving into CCC territory increases sharply over one quarter presents a different risk profile from a pool where the same level has remained broadly stable for a year. The trustee report provides the historical data needed to make that comparison.</p></li></ol><p>There are also five areas where we would look more closely.</p><p><strong>First, the CCC bucket.</strong> A rising proportion of CCC-rated loans can put increasing pressure on the structure, although the relevant threshold varies by transaction. The important point is not to rely on a generic rule. Read the indenture and understand how the transaction treats CCC-rated collateral when calculating its tests.</p><p><strong>Second, WARF deterioration.</strong> The weighted average rating factor provides a single measure of the credit quality of the collateral. A rising WARF indicates that the portfolio is becoming weaker. Again, the absolute number needs to be considered in the context of the specific transaction. The direction of travel is often the more useful signal.</p><p><strong>Third, spread compression in the underlying loans.</strong> This is easy to overlook. If a CLO is buying new loans at progressively tighter spreads because the leveraged loan market is strong, the economics of the transaction can deteriorate even when headline credit metrics remain stable. Equity and mezzanine investors are effectively taking credit risk for less compensation. Changes in the average spread of the collateral pool can reveal this over time.</p><p><strong>Fourth, refinancing risk.</strong> The maturity profile of the underlying loans deserves close attention. A pool with a large proportion of loans maturing within the next 18 to 24 months will depend more heavily on the refinancing market. If credit conditions weaken, some borrowers may refinance at materially higher costs, while others may struggle to refinance at all. That can eventually translate into defaults or restructurings.</p><p><strong>Fifth, manager quality.</strong> CLO managers differ in their investment style, portfolio construction and willingness to trade around deteriorating credits. Track record matters, but so does the way that track record has been generated. A manager producing higher returns by accepting weaker collateral is not necessarily adding more value than a manager taking a more conservative position.</p><p>None of this requires a wealth platform to build its own CLO modelling desk. It requires someone who can read the trustee report, identify the relevant changes and interpret them consistently from one reporting period to the next.</p><p>That is where an independent research partner can be useful.</p><p>The first task is triage. Someone needs to review the latest trustee report and determine whether anything has changed that warrants attention. Most reports contain a large amount of data. The value lies in separating the information that matters from everything else.</p><p>The second is stress estimation. What happens if defaults rise to 5%? What if the CCC bucket reaches 8%? What if interest rates remain elevated for another year? A full Intex model will provide greater precision, but the basic direction of risk can often be assessed using the structure of the transaction, the relevant triggers and reasonable assumptions for defaults and recoveries.</p><p>The third is manager assessment. The question is not simply whether a CLO has produced a high return. We want to know whether the manager is generating that return through security selection and trading discipline, or by taking risks that will only become visible when the credit cycle turns.</p><p>The fourth is a second opinion. That can be useful before increasing an allocation, after a rating action, or simply when a quarterly report shows a change that is difficult to interpret. Independent credit analysis does not replace the investor&#8217;s judgement. It gives that judgement a stronger analytical base.</p><p>CLOs are not inherently bad investments. Senior tranches of well-managed transactions can be well protected by the structure and can provide useful income relative to conventional fixed income. The issue is not whether CLOs belong in portfolios. It is whether investors understand the risks they are taking.</p><p>A rating is only one part of the analysis. The collateral pool, structural tests, manager behavior and refinancing profile are equally important.</p><p>For wealth platforms with meaningful CLO exposure, we think that level of analysis should be part of the normal credit process. The information is available. The challenge is turning it into a consistent assessment of risk.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent European credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[The 5 Questions I Ask Before Lending to a Data Centre SPV]]></title><description><![CDATA[How I assess a private placement in 20 minutes, and why much of the US data-centre pipeline may never get built.]]></description><link>https://www.rovelliresearch.com/p/the-5-questions-i-ask-before-lending</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/the-5-questions-i-ask-before-lending</guid><pubDate>Tue, 22 Sep 2026 12:36:40 GMT</pubDate><content:encoded><![CDATA[<p>Two weeks ago, I wrote about the hidden debt layer behind the AI infrastructure boom. Since then, readers have asked how I distinguish a sound private-credit structure from one that may become a refinancing problem later. I have spent 25 years analyzing credit structures where the investment story looks convincing until you examine the contracts and cash flows. Data centers have some of the same characteristics I saw during the fiber and shale booms. I think five questions provide a useful first filter. With the right documents, I can make that initial assessment in about 20 minutes. Missing documents are useful information too.  </p><p>First, I want to know whether the power is actually secured. A data center without power is not a viable data-center project. I draw a sharp distinction between power being announced and power being contracted. I want an executed interconnection agreement, a credible position in the grid queue, and a realistic connection date. I also want to know who bears the cost if that date slips. A statement that power is &#8220;allocated&#8221; tells me little without contractual protection. Local permitting matters as well. A financed project with a site and prospective customer can still fail if rezoning, permits, or community opposition delay construction. I therefore ask for the interconnection agreement and the financial consequences of a two-year delay. I think a signed agreement, a credible connection date, and defined protections provide useful evidence. A letter of intent and a list of outstanding approvals do not.</p><p>Second, I examine whether the customer commitment is actually binding. When I lend to a data-center SPV, I am not lending directly to Microsoft, Google, or Amazon. I am lending to a vehicle whose cash flow depends on a contractual relationship with one of them, or sometimes on another intermediary that itself relies on a hyperscaler. That distinction matters. A 15-year lease with limited termination rights and a parent guarantee is very different from a capacity reservation that allows the customer to reduce its commitment as AI demand changes. I therefore focus on termination, change-of-control, assignment, and payment provisions. I want to know whether the obligation is genuinely &#8220;hell or high water.&#8221; Oracle illustrates why looking through the structure matters. Its growing role in AI infrastructure does not make it an equivalent counterparty to the largest hyperscalers. I therefore assess where the economic risk ultimately sits, rather than relying on the logo on the presentation.</p><p>Third, I stress the cash-flow model and the capital structure. I typically ask what happens at 60% utilization, with electricity costs 30% above plan, or under another project-specific downside case. I want sufficient debt-service coverage before assumptions become heroic. A 1.4x DSCR at 60% utilization gives me more comfort than 1.1x at 95%, even if the latter looks better in the base case. I also examine technology risk. GPUs may sit outside the SPV, with separate financing, ownership, and useful lives. The building can therefore outlast the equipment, but adapting it for a new generation of chips may require additional capital. I think lenders must understand that mismatch before relying on long-dated contracted cash flow.</p><p>Fourth, I focus on refinancing, liquidity, and visibility. Many private placements mature in five to seven years while the underlying lease runs for 15 years or more. That maturity mismatch leaves the project dependent on a future lender. Refinancing becomes harder when credit spreads widen, or private-credit funds become more selective, even if the project continues to perform. I therefore want to see amortization, cash sweeps, interest-rate protection, and sufficient headroom to absorb higher refinancing costs. I also want regular construction reporting, independent oversight, and credible valuation evidence. A project that requires refinancing in 2027 or 2028 needs a realistic source of capital, not an assumption that the next lender will accept the same terms. For now, however, my biggest concern is still whether the asset can actually be built and whether the contracted cash flow can support the debt.</p><p>I think the simplest filter is whether both power and the customer commitment are sufficiently secured. The broader industry pipeline may eventually prove much smaller than announced capacity, because projects can fail to obtain electricity, permits, financing, or customers. A proposed 7-gigawatt project without a signed tenant is, in my view, a development opportunity rather than a financed data-center asset. I remain constructive on the need for computing infrastructure, but I am cautious about debt structures built on high leverage and optimistic assumptions. I would rather establish the quality of the power contract, customer obligation, downside cash flow, refinancing plan, and valuation evidence in the first 20 minutes than discover the weakness when the loan comes due.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[The 2027 Refinancing Squeeze. When Data-Center Debt Comes Due]]></title><description><![CDATA[Private placements that funded data centers in 2023&#8211;25 were written for 5&#8211;7 years. The assets they funded need 10&#8211;15 years to prove their economics. That ga hidden debt layer has a maturity date.]]></description><link>https://www.rovelliresearch.com/p/the-2027-refinancing-squeeze-when-bea</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/the-2027-refinancing-squeeze-when-bea</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Fri, 18 Sep 2026 11:04:04 GMT</pubDate><content:encoded><![CDATA[<p>I have mapped the supply of AI debt and the structures that keep it outside hyperscaler balance sheets. The next question is when it needs to be refinanced. Most off-balance-sheet data center debt was not raised as 10-year corporate bonds. It was raised as 5-7 year private placements and project-level loans held by private credit funds and insurers. The 2023-25 vintage, the first large wave of AI data center SPVs and JVs, starts to mature in 2027. The 2025-26 vintage follows in 2030-32, but the pipeline is already building. That creates a wall that is invisible in hyperscaler filings but real at the vehicle level.</p><p>In 2024, that debt priced on the assumption of on-time construction, cheap power, and a hyperscaler lease that would be renewed without negotiation. In 2027, it will reprice on what actually happened. I do not expect a Microsoft or Amazon default to trigger this. I expect a refinancing conversation at a vehicle that needs an extension, a higher coupon, or a guarantee draw, and a private credit fund that has to explain a mark to its investors.</p><p>The BIS paper in March flagged the stock. Industry data fills in the timing. The 2023 vintage financed early hyperscale JVs where land and power were secured before the AI surge. It used 5-year private placements, first maturity in 2028, priced around 5-6% with IG-like lease coverage. The 2024 vintage financed the first large AI-optimized campuses, GPU-dense and liquid-cooled. It used 5-7 year debt, maturing 2029-31, priced around 6-7% on compressed spreads for the hyperscaler name. The 2025-26 wave is different. It combines SPVs plus project finance plus ABS, with 5-7 years plus a 3-year construction loan, maturing 2028-33 with extensions, priced at 6.5-8% with completion risk still open.</p><p>Two things matter here. A data center is a 15-year asset, lease plus renewal. The debt funding it is half that. The structure assumes refinancing succeeds. I have seen this mismatch before in 2007-08 conduits: short debt funding long assets works until it needs to be refinanced. Then terms change. The wall is also concentrated. Unlike corporate bonds that are staggered across issuers, data center private placements cluster around the same construction cycle. When power delivery slips for one campus, it often slips for the neighboring one. Same grid, same contractor, same equipment lead times.</p><p>Investors who lent against a Microsoft lease in 2024 were not really lending to Microsoft. They were lending to a single-asset vehicle whose ability to repay depends on that lease still being wanted at the same rent in 2029. That is not Microsoft risk. It is single-building risk. A 10-year Microsoft bond, MSFT 4.5% 2033 for example, is a claim on Microsoft global cash flow. If one data center underperforms, the other businesses cover it. A data center SPV private placement is a claim on one building cash flow, plus whatever guarantee was negotiated. Construction risk lives at the vehicle. A 12-18 month delay does not change Microsoft rating. It changes whether the vehicle generates cash in time to service debt before maturity. The BIS notes this explicitly: risk is transferred from corporate balance sheet to vehicle, then to non-bank holder.</p><p>Power is the bottleneck, not concrete. Grid interconnection queues in Virginia, Arizona, and parts of Europe run 2-4 years. A completed shell without power is not an operating asset. I am tracking several 2024-vintage campuses where shell completion is on time but substation delivery is 9-15 months late. The lease often does not commence until power is live. Rent does not start until power is live. Coverage is lease-dependent. Underwriting assumed 1.4-1.5x DSCR on a 15-year lease. If power costs rise 15-20% or utilization ramps slower than 70% in year 2, coverage falls below 1.2x. That is the typical threshold for refinancing without additional equity.</p><p>For 2027-28 I am watching refinancing at the vehicle first. The private credit fund that bought a 5-year note in 2023 expected to be refinanced by a similar fund in 2028, or by an ABS takeout once the asset was stabilized. That takeout requires a stabilized asset and a receptive market. If either is missing, the choice is extension at a higher spread, additional equity from the sponsor, or a guarantee draw. My base case for weak 2023 vintages is extension plus 100-200bp. I will watch for extensions described as amend and extend and for new-issue spreads versus 2024 levels.</p><p>Then guarantee activation, where off-balance-sheet becomes on-balance-sheet. Many JVs include completion guarantees, cost-overrun guarantees, or a put on capacity. These sit in footnotes, not debt tables. Hyperscalers disclose them as commitments and contingencies, often without a notional. If a 2024 campus needs EUR 200-400m of overrun funding to reach power-on, the hyperscaler can either inject equity and keep the structure off-balance-sheet or let the vehicle draw the guarantee and bring the obligation home. The first preserves optics. The second preserves cash. Management teams with leverage targets will hesitate until they have to. I will watch for changes in unconsolidated JV commitments in 10-K and 10-Q, and any shift from operating lease language to variable interest entity consolidation.</p><p>Then marks, not defaults. Private placements do not trade daily. But insurers and private credit funds must mark them for solvency and investor reporting. A vehicle that was par in 2024 at 6% with 1.5x coverage may be 92-95 in 2027 if coverage is now 1.15x and refinancing will be at 8%. That is not a default. It is a loss that has to be recognized, and it reduces capacity for the next wave. The first sign will not be a missed payment. It will be a fund reporting a selective markdown in infrastructure debt and a slower deployment pace. The second sign will be a hyperscaler quietly consolidating a JV it previously called unconsolidated.</p><p>My probabilities are straightforward. I assign 50% to an orderly roll. Assets power on within 6 months of plan, leases commence, DSCR stays above 1.3x. Refinancings in 2028 clear 50-100bp wider than 2024. The ABS market absorbs stabilized deals. Guarantees stay untouched. I assign 35% to extended and repriced. That is my central risk. 30-40% of 2023-24 vintages need a 12-18 month extension, plus 150-250bp and partial sponsor equity. Private credit marks fall 5 to 8 points. Hyperscalers inject EUR 1-3bn in aggregate completion equity. No rating action, but footnote obligations rise. I assign 15% to structure fails. AI capex slows more broadly, power constraints leave campuses underutilized, several vehicles cannot refinance without material guarantee activation. Hyperscalers consolidate EUR 5-10bn of previously off-balance-sheet debt.</p><p>I keep orderly roll at 50% because hyperscalers have the cash to bridge. Microsoft, Amazon, and Alphabet each generated USD 60-100bn plus operating cash flow in 2024-25. The question is willingness, not ability. The 35% case does not require AI to fail, only to be late.</p><p>Into September 2027 I will track a short list. At the vehicle level: completion versus plan, power-on date versus shell, any amend and extend. At the hyperscaler level: JV footnotes, guarantee language, lease reclassification. A capex cut also matters. It means fewer new vehicles will be needed. At the market level: private credit marks and deployment pace, insurance allocation to infrastructure debt, hyperscaler 10-year spreads versus Treasuries. They widened 5-10bp in 2025. A further 10-20bp would signal crowding out from my September 8 thesis. I expect the first evidence to appear in refinancing terms and private-market marks, not in public ratings.</p><p>The hyperscaler is unlikely to default. The vehicle may need additional equity, a guarantee, or an extension. That is where the squeeze becomes visible. The AI infrastructure trade has been priced as if debt is permanent and off-balance-sheet means risk-free. Neither is true. Debt that looked elegant in 2024 has to be refinanced in a world where power is scarce, construction is late, and private credit is more selective. In 2027, I expect to find out whether 2024 underwriting was right about utilization. The hyperscaler will not default. The vehicle might need to be rescued. The holder will have to mark it. That is the refinancing squeeze. It will show up first in a footnote, then in a spread, then in a fund NAV, long before it shows up in a rating action.</p><p>What do you think, does off-balance-sheet still mean off-risk when the debt comes due? Join the discussion in the comments.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[The Hidden Debt Layer in AI Infrastructure]]></title><description><![CDATA[Hyperscalers are finding ways to finance billions of dollars of data-centre construction without putting all of the debt directly on their own balance sheets.]]></description><link>https://www.rovelliresearch.com/p/the-hidden-debt-layer-in-ai-infrastructure</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/the-hidden-debt-layer-in-ai-infrastructure</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Tue, 15 Sep 2026 08:26:31 GMT</pubDate><content:encoded><![CDATA[<p></p><p><strong>The AI infrastructure boom is creating a hidden layer of debt, and credit investors may be among the first to feel the consequences if the cycle turns.</strong></p><p>The Bank for International Settlements recently <a href="https://www.bis.org/publications/financing-ai-infrastructure-boom-on-and-off-balance-sheet-borrowing">published</a> an analysis of how the major technology companies are financing their enormous expansion in data-centre capacity. The headline numbers are already familiar: corporate bond issuance has surged, credit spreads have started to widen for some lower-rated companies, and the largest technology companies are committing hundreds of billions of dollars to infrastructure.</p><p>I wrote about the more visible, on-balance-sheet debt in my previous article. What caught my attention in the BIS analysis, however, was the other side of the equation: the financing that does not necessarily appear as conventional corporate debt at the hyperscaler level.</p><p>The basic structure is relatively simple. A technology company can create or participate in a joint venture that builds and owns a data center. The joint venture then raises debt from private investors, while the technology company takes an equity stake and signs a long-term lease or capacity agreement.</p><p>From an economic perspective, this can look remarkably similar to the technology company borrowing money itself to build the data center.</p><p>From an accounting perspective, however, it can look very different. The debt sits at the project or joint-venture level, while the technology company records payments under its contractual arrangements. In other words, some of the financial risk has moved outside the corporate balance sheet.</p><p><strong>There are three reasons why I think this deserves more attention.</strong></p><p>The first is the sheer scale of the investment cycle. Microsoft, Amazon, Alphabet, Meta and Oracle are collectively spending hundreds of billions of dollars on data centers, chips, networking equipment and other AI infrastructure. Their enormous operating cash flows cover a significant part of this spending, but they cannot cover everything indefinitely as capital expenditure continues to rise. The remaining funding has to come from somewhere, and increasingly it is coming from a combination of corporate debt, private credit and project-level financing.</p><p>The second issue is who is providing that capital. Private credit funds, insurance companies and other institutional investors have become increasingly important sources of financing for infrastructure projects. That is not necessarily a problem, but the risks involved in financing a data center are different from those involved in lending to a conventional company. A traditional corporate lender can spend most of its time analyzing revenues, margins, cash flow and leverage. A data-center lender has to make assumptions about construction schedules, electricity availability, equipment deliveries, future demand and the utilization rate of the finished facility. Those assumptions can be difficult to get right, particularly in an industry that is expanding so quickly that there is relatively little historical data on which to base them.</p><p>The third issue is the way these structures can make risk appear safer than it really is. A data-center SPV backed by a 15-year agreement with Microsoft or Amazon looks extremely attractive. You have a physical asset, a long-term contract and a counter-party with a very strong balance sheet. But the most important risk may not be that Microsoft or Amazon defaults. The more realistic question is what happens to the project if construction takes significantly longer than expected, the electricity connection is delayed, costs increase sharply or the technology company eventually discovers that it needs less capacity than it originally expected.</p><p><strong>How the structure works</strong></p><p>The structure varies from transaction to transaction, but the basic mechanics are relatively straightforward.</p><p>A group of investors establishes a joint venture to develop a data center, with the vehicle acquiring the land, securing an electricity connection and arranging the financing required to construct the facility.</p><p>The joint venture then raises debt, often through private placements rather than through the public bond market. A small number of institutional investors may provide the financing, with the creditworthiness of the project depending on both the completed data center and the contractual commitment from the hyperscaler.</p><p>The technology company may own part of the project, but its more important role is often its long-term agreement to use the facility.</p><p>That agreement is what creates the expected cash flow needed to repay the debt. For the technology company, this arrangement can be attractive because it provides access to additional capacity without requiring the company to finance the entire project directly. It can also allow some of the borrowing to remain at the project level rather than appearing as conventional corporate debt.</p><p>For the investor, the proposition is equally attractive. Instead of lending to a highly leveraged start-up, the investor is financing a physical asset that is expected to generate cash flow under a long-term agreement with one of the world&#8217;s largest technology companies.</p><p>The problem is that this apparent simplicity can hide a concentration of risks that are difficult to observe from outside the transaction.</p><p><strong>Three risks I am watching</strong></p><p><strong>1. Refinancing risk</strong></p><p>The first risk is refinancing. Data centers can take several years to develop and construct, while the debt used to finance them may have a much shorter maturity. That creates a potential mismatch between the life of the asset and the maturity of the financing. Imagine a project that takes four years to build but has six-year debt. The facility may have only two years of operating history before the debt needs to be refinanced.</p><p>If credit markets remain favorable, that may not matter. But if the refinancing takes place during a period when private credit investors have become more cautious, the economics can change very quickly. The new lenders may demand a substantially higher interest rate, additional collateral or tighter financial conditions.</p><p>A project that looked highly attractive when the original financing was arranged can therefore become much less attractive when it has to return to the market. This is one reason I am particularly interested in the maturity profile of the debt being raised today.</p><p><strong>2. Guarantees can become real liabilities</strong></p><p>The second risk is the use of guarantees and other contractual commitments from the hyperscaler. Some structures may include completion guarantees, minimum-payment commitments, capacity agreements or other forms of support designed to give investors additional comfort.</p><p>These commitments may not appear as conventional debt on the hyperscaler&#8217;s balance sheet, but they still represent economic obligations. That distinction becomes particularly important when something goes wrong.</p><p>If a project runs significantly over budget and cannot complete construction without additional funding, a guarantee that looked like a remote contingency when the deal was signed can suddenly become a very real liability for the technology company. At that point, the distinction between on-balance-sheet and off-balance-sheet financing becomes much less meaningful. The obligation still has to be met by someone.</p><p><strong>3. Private credit valuations can lag reality</strong></p><p>The third risk is more subtle and relates to the way private credit is valued. Unlike public bonds, private loans do not trade continuously in a transparent market. An investor therefore does not receive a new market price every day telling them that the value of their investment has fallen.</p><p>That can make private credit appear more stable than it actually is. If the economics of a data-centre project begin to deteriorate, the value of the loan may not immediately reflect that change. The adjustment could come later, when the project needs refinancing, when the assumptions underlying the original valuation are reassessed, or when an investor is forced to sell.</p><p>A loan that was valued at par when it was originated could therefore be worth materially less two years later if construction costs have risen, electricity supply has been delayed or expected utilization has fallen. The risk has not necessarily appeared suddenly. The market may simply have taken longer to recognize it.</p><p><strong>Scenarios</strong></p><p><strong>50%</strong></p><p>Projects are completed broadly on schedule, demand continues to grow and contractual cash flows are sufficient to service the debt.</p><p><strong>Delayed completion</strong></p><p><strong>30%</strong></p><p>Construction and power delays push projects beyond their original timelines, forcing some vehicles to seek additional financing or refinance on less favorable terms.</p><p><strong>Structural repricing</strong></p><p><strong>20%</strong></p><p>AI infrastructure spending slows materially, demand for new capacity falls and some projects struggle to refinance their debt at anything close to the original terms.</p><p>My base case remains an orderly digestion of the current investment cycle. The major hyperscalers have enormous financial resources, strong access to capital and considerable experience managing large infrastructure projects. It would therefore be wrong to assume that every project financed through an SPV is going to run into trouble.</p><p>My greater concern is the middle scenario, where nothing actually breaks but the assumptions made when the projects were financed prove to be too optimistic. Construction delays and electricity constraints are already real issues in parts of the data-center market. A project does not need to fail completely to create problems for its lenders. It simply needs to take longer and cost more than originally expected.</p><p>The third scenario is the tail risk. For that to happen, we would need more than a few delayed projects. The growth in AI infrastructure spending itself would have to slow significantly, leaving the industry with more data-center capacity than it needs or can profitably use. I think that is less likely over the next two years, but the probability would increase if capital expenditure continues to grow much faster than the revenues generated by AI services.</p><p><strong>What I will be watching</strong></p><p>The most useful warning signs may not come from the headline numbers published by the hyperscalers. They may instead appear in the details of individual projects and financing transactions.</p><p>At the project level, I will be watching whether data centers are being completed on schedule, whether electricity connections arrive when expected, whether construction costs are increasing and how much new private credit is flowing into data-center financing.</p><p>I will also be watching the pricing of new transactions. If investors suddenly demand significantly higher yields to finance similar projects, that would suggest that the market is becoming less comfortable with the underlying risks.</p><p>At the hyperscaler level, changes in long-term capacity commitments could be particularly revealing. If companies start reducing, delaying or restructuring their commitments to new data centers, that would be an important signal that the assumptions behind some of these projects are changing.</p><p>I would also pay close attention to the footnotes of corporate financial statements. Guarantees and other contractual commitments are often buried there rather than highlighted in the headline financial figures.</p><p>Finally, the private credit market itself deserves attention. If private credit funds begin experiencing significant withdrawals, or if institutional investors become less willing to allocate capital to infrastructure debt, refinancing conditions could deteriorate even if the underlying data centres are still operating normally. That is why the financing market may provide an earlier warning than the operating companies themselves.</p><p><strong>The credit implication</strong></p><p>The on-balance-sheet debt problem I wrote about two weeks ago is relatively easy to see. You can track corporate bond issuance, look at credit spreads and examine the amount of long-term debt being added to hyperscaler balance sheets.</p><p>The off-balance-sheet problem is much harder to measure. That does not automatically make it more dangerous, and I would not argue that using SPVs is inherently problematic. In many cases, project financing is simply a sensible way of matching a long-lived physical asset with the investors who are best equipped to finance it.</p><p>The problem arises when investors treat the structure itself as a source of safety. Moving debt from a hyperscaler&#8217;s balance sheet to a project company does not eliminate construction risk. It does not eliminate electricity risk. It does not eliminate demand risk, and it does not eliminate refinancing risk.</p><p>This distinction is particularly important because private credit has grown so rapidly in recent years. The market has attracted enormous amounts of capital, and investors have been willing to accept increasingly complex structures in exchange for relatively attractive yields.</p><p><strong>The first warning sign may not be a tech default</strong></p><p>A plausible first warning sign would a private credit fund might mark down a data-center investment; or a project might need additional financing because construction costs have increased. A refinancing might take place at a much higher interest rate than the original loan. Or a hyperscaler might have to honor a guarantee that everyone expected to remain a remote contingency.</p><p>None of these events would necessarily constitute a crisis. But they would tell us that the economics of the AI infrastructure boom are beginning to change. And because a significant amount of this financing sits outside the traditional banking system and public bond markets, the first losses could appear among private credit funds, insurance companies and other institutional investors rather than among the banks or the technology companies themselves.</p><p><strong>Bottom line</strong></p><p>The AI infrastructure boom is creating a new layer of financing that is easy to overlook. The major technology companies are increasingly using joint ventures, project companies and other structures to finance data centres without putting all of the borrowing directly onto their own balance sheets. For investors, these structures can offer an attractive combination of yield, physical assets and long-term contracts with some of the world&#8217;s strongest companies. But the apparent safety ultimately depends on a series of assumptions. Data centres have to be completed on time. Power has to be available. Construction costs have to remain under control. Demand for computing capacity has to continue growing, and the projects have to be able to refinance their debt when it matures. If those assumptions hold, the structure works very well. If they don&#8217;t, the losses may appear first in private credit funds, insurance companies and other non-bank investors.</p><p>That is why I think the off-balance-sheet financing of AI infrastructure deserves considerably more attention than it is currently receiving. The most important question is not how much debt the hyperscalers have on their balance sheets. It is how much economic leverage exists around them.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item><item><title><![CDATA[Nvidia’s $12.9 Billion Bet on Hugging Face Is About Ecosystem, Not Models]]></title><description><![CDATA[The chipmaker is buying something potentially more valuable than a portfolio of AI models: a direct influence in the community that decides which models get built, tested and used.]]></description><link>https://www.rovelliresearch.com/p/nvidias-ai-power-move-why-129-billion</link><guid isPermaLink="false">https://www.rovelliresearch.com/p/nvidias-ai-power-move-why-129-billion</guid><dc:creator><![CDATA[Alessandro Rovelli]]></dc:creator><pubDate>Thu, 10 Sep 2026 14:17:35 GMT</pubDate><content:encoded><![CDATA[<p>Nvidia announced an agreement to acquire Hugging Face for $12.9 billion. The transaction comprises approximately $11.9 billion payable to Hugging Face shareholders and up to $1 billion in equity-based retention for employees who join Nvidia. The deal is expected to close in the first half of 2027, subject to regulatory approval. The headline number is large, but not that large for a company of Nvidia&#8217;s scale. </p><p>The strategic rationale is interesting. Hugging Face has become one of the central distribution points for open AI development. More than 18 million developers, researchers and creators use the platform to access more than three million models, 500,000 datasets and one million applications, while more than 200,000 companies use it to discover, evaluate, customize and deploy AI.</p><p>Nvidia is therefore not primarily buying models. It is buying a position at the point where developers decide which models to use, how to modify them and, increasingly, where to run them.</p><h2>The deal is easier to understand if you think about Hugging Face as infrastructure</h2><p>Hugging Face is often described as a repository for AI models. This description understates what the company has become.</p><p>A developer can discover a model, inspect how it performs, download it, fine-tune it on proprietary data and move it into an application without leaving the platform&#8217;s broader ecosystem. The same infrastructure connects models with datasets, libraries, deployment tools and a large developer community.</p><p>That makes Hugging Face a piece of digital infrastructure for the open-model economy. The value to Nvidia is not simply that it now owns three million models. Nvidia already has its own model and software portfolio, including NIM and NeMo, and has relationships with virtually every major AI laboratory.</p><p>The more important asset is the developer relationship. When a developer chooses a model, that choice eventually has consequences for the hardware on which the model is trained and deployed. Nvidia does not need to prevent developers from using AMD, Intel or custom accelerators. It simply needs its own hardware and software to remain the easiest and best-performing option.</p><h2>The most valuable feature of Hugging Face may be its neutrality</h2><p>There is an important paradox at the centre of this transaction. Hugging Face is valuable because it has historically been neutral. Developers can use models and tools across different hardware platforms, clouds and software frameworks. Nvidia&#8217;s announcement explicitly says that this will continue: Hugging Face will remain open, support multiple clouds and accelerators, and developers will not be required to use Nvidia computing to build or deploy through the platform.</p><p>For Nvidia, preserving that neutrality is necessary to preserve the economic value of the acquisition. If Hugging Face becomes perceived as an Nvidia storefront, its competitive position changes. Developers who currently use the platform precisely because it is hardware-agnostic have an obvious alternative: they can move to other repositories, cloud platforms or open-source projects.</p><p>Open-source communities are unusually sensitive to this kind of change because switching costs are often low. Code can be forked. Models can be mirrored. Communities can migrate. Nvidia is therefore buying an asset whose value depends partly on remaining independent enough to be trusted by developers.</p><h2>Why Nvidia needs this now</h2><p>The timing is particularly interesting because Nvidia&#8217;s most important customers are trying to reduce their dependence on Nvidia.</p><p>Amazon has Trainium. Google has TPUs. Microsoft has Maia. Meta has MTIA. OpenAI is also developing custom silicon with Broadcom. The objective is broadly the same: reduce the proportion of AI infrastructure spending that ultimately flows through Nvidia GPUs.</p><p>At the same time, open-weight models are becoming increasingly important. Enterprises can download models, fine-tune them on proprietary data and deploy them within their own environments rather than relying entirely on APIs from companies such as OpenAI or Anthropic.</p><p>If AI increasingly moves from a handful of proprietary models toward a much larger ecosystem of open-weight models, Nvidia needs to remain influential at the software and developer layers surrounding those models, rather than relying solely on its position as the dominant GPU supplier. Hugging Face gives it a direct relationship with that ecosystem.</p><h2>The information advantage may be as important as the distribution advantage</h2><p>Hugging Face provides Nvidia with visibility into what developers are actually experimenting with.</p><p>Which models are being downloaded?</p><p>Which ones are being fine-tuned?</p><p>Which datasets are attracting attention?</p><p>Which applications are gaining users?</p><p>Which models are moving from experimentation toward deployment?</p><p>That information could be extremely valuable to a semiconductor company.</p><p>Chip demand ultimately depends on software adoption. If Nvidia can see which models and applications are gaining traction before those trends appear in conventional corporate spending data, it potentially gets an earlier signal about where future computing demand is heading. In that sense, Hugging Face could become an intelligence asset as much as a distribution platform.</p><h2>There is another piece of the deal: robotics</h2><p>I also think the robotics angle deserves more attention. Hugging Face acquired Pollen Robotics in 2025 and has been developing an open robotics ecosystem around projects such as LeRobot. Nvidia, meanwhile, has made physical AI and robotics a major part of its longer-term strategy.</p><p>If AI increasingly moves from software applications into machines, autonomous systems and industrial environments, Nvidia will want to control as many of the enabling layers as possible: simulation, training, inference, robotics software and ultimately the compute underneath it.</p><p>Hugging Face gives Nvidia a stronger position within the developer community that could build those applications. The opportunity potentially extends into the next generation of physical AI.</p><h2>What this means for enterprises</h2><p>For companies using AI, the immediate effect should be relatively limited. The attraction of open models remains the same: greater control over data, potentially lower inference costs, the ability to fine-tune models and less dependence on a single proprietary API provider.</p><p>What could change is the path from experimentation to deployment. I would expect Nvidia to make the journey from a Hugging Face model to Nvidia-optimized inference increasingly seamless. The important point is that Nvidia does not necessarily need to make alternative hardware unavailable.</p><p>It simply needs to make its own infrastructure the path of least resistance. If a developer can deploy a model on AMD, Intel or a custom accelerator, but gets better performance, easier optimization and better tooling on Nvidia, the economics will marginally favor Nvidia without the company ever having to explicitly close the ecosystem.</p><h2>For credit investors, the financial impact is secondary</h2><p>From a credit perspective, the transaction itself is unlikely to matter much to Nvidia&#8217;s balance sheet. Even a $12.93 billion transaction is small relative to the company&#8217;s financial resources and does not fundamentally change its leverage profile. The signal is more important than the financing.</p><p>Nvidia is effectively paying to extend its position further up the AI value chain at precisely the moment when some of its largest customers are attempting to move in the opposite direction.</p><p>That suggests management is thinking beyond the current GPU cycle. For me, the question is therefore not whether Nvidia paid too much for Hugging Face on today&#8217;s revenue. It almost certainly did if the business is valued purely as a conventional software company. Hugging Face has an annualized revenue at roughly $150 million, which would imply a purchase price of around 86 times revenue.</p><p>But that is the wrong valuation framework. Nvidia is paying for strategic position, developer access, information and ecosystem influence. The relevant question is whether those assets help protect the much larger economics of Nvidia&#8217;s core business.</p><h2>The central risk is neutrality</h2><p>The most important risk is also the simplest: Nvidia could destroy part of what it has just paid for.</p><p>Developers will be watching whether non-Nvidia hardware continues to receive serious support. They will notice if AMD, Intel or custom accelerators consistently receive optimizations later than Nvidia hardware. They will notice if cloud deployment increasingly nudges users toward Nvidia&#8217;s own infrastructure. And they will notice if frameworks outside Nvidia&#8217;s ecosystem receive less attention.</p><p>Nvidia does not have to explicitly prohibit competitors for this to become a problem. Perceived favoritism could be enough. That is why the company&#8217;s commitment to keeping Hugging Face open is so important. Nvidia has explicitly said that developers will retain the ability to choose their models, frameworks, clouds, inference providers and computing platforms.</p><h2>The regulatory question  </h2><p>The antitrust issue is similarly interesting. This is not simply a semiconductor acquisition. Nvidia is acquiring a major platform sitting between developers, models and computing infrastructure. Regulators therefore have to consider whether control of that platform could eventually be used to disadvantage competing hardware or cloud providers.</p><p>Nvidia will have a straightforward argument: Hugging Face is open-source infrastructure with multiple competing alternatives, and the company has committed to maintaining interoperability.</p><p>Regulators can reasonably ask a different question: does ownership of an important distribution platform give Nvidia the ability to influence developer behavior even without formally restricting access?</p><p>I believe that the the acquisition will receive unconditional or conditional clearance in H1 2027. If regulators impose remedies, I would expect them to focus primarily on interoperability, access and support for competing hardware rather than requiring a structural remedy.</p><h2>What I am watching</h2><p>The first indicator will be technical rather than financial: how quickly new models receive optimized support across different hardware platforms. If Nvidia&#8217;s hardware consistently receives first-class treatment while competitors fall behind, the company&#8217;s neutrality pledge will become increasingly difficult to take at face value.</p><p>I will also watch contributor and maintainer activity around Hugging Face&#8217;s key open-source projects. A noticeable increase in developer departures, forks or alternative repositories would be an early warning that the community perceives a loss of independence.</p><p>On the regulatory side, I will be watching the European Commission and FTC processes, particularly any requests for commitments around interoperability or non-discrimination.</p><p>And finally, I will watch the commercial integration. How quickly does Hugging Face become embedded in Nvidia&#8217;s enterprise and cloud products? The more tightly the two businesses become integrated, the greater the potential benefit, but also the greater the risk that developers stop viewing Hugging Face as neutral infrastructure.</p><h2>My conclusion</h2><p>I don&#8217;t think Nvidia paid $12.9 billion to acquire a collection of AI models. It paid to put itself closer to the developers who decide which models matter and how those models ultimately get deployed.</p><p>Nvidia&#8217;s existing moat has been built around hardware, and the software ecosystem surrounding its GPUs. Hugging Face potentially extends that moat into a different part of the AI stack: model discovery, experimentation, deployment and developer distribution.</p><p>The irony is that the value of the acquisition depends on Nvidia not behaving like the owner of the platform.</p><p>If Nvidia can preserve Hugging Face&#8217;s neutrality while making Nvidia the most efficient and best-supported route from model development to production, the acquisition would give Nvidia a way to influence the direction of AI infrastructure without having to control every layer of the stack.</p><p>If developers instead conclude that Hugging Face has become an Nvidia distribution channel, the economics change. The community can migrate, competitors can emerge and the very neutrality that made the platform strategically valuable begins to disappear.</p><p>The next phase of the AI competition may not be decided only by who builds the fastest chip or the most capable model. It may be decided by who controls the infrastructure through which developers discover, test and deploy those models.</p><p><em>Alessandro Rovelli</em><br><em>Rovelli Research</em><br><em>Independent credit investment research</em></p>]]></content:encoded></item></channel></rss>