Michael Burry — the investor whose prescient bet against subprime mortgage securities in 2007 was immortalized in The Big Short — is now pointing his lens at a corner of the private credit market he believes carries the same architecture of deferred risk that toppled the financial system nearly two decades ago.
The vehicle this time is not subprime mortgage bonds. It is a chain of structured credit instruments tied to data center leases and chip financing — instruments that Burry argues are accruing inside insurance company balance sheets at the direction of private equity owners, with state guaranty programs providing an implicit public backstop that the market is not adequately pricing.
What Happened
Writing on his Substack, Burry shared and annotated an academic paper examining the trend of private equity firms acquiring insurance companies and repositioning those balance sheets away from conventional fixed income toward riskier, less liquid asset-backed securities (ABS) and structured credit. The practice, Burry noted, has accelerated sharply as the AI infrastructure buildout has generated a new class of financeable assets — data center leases and semiconductor equipment financing — that are being packaged into the structured securities now sitting on insurer books.
“Those asset backed assets and structured securities are increasingly coming off data center and chip leases,” Burry wrote. “This is where the possible contagion takes down the economy — by withdrawing funding for the data center buildout, which is also an increasing part of United States economic growth.”
The warning is specific in its mechanism. Under the private equity playbook Burry describes, PE firms acquire insurance companies, benefit from the float — the premium income that sits on balance sheets before claims are paid — and redeploy that float into higher-yielding but illiquid instruments. As long as yields on those instruments exceed claims obligations and the underlying assets perform, the model works. The danger lies in what happens when it does not.
On the rate environment, Burry was blunt. “Higher rates for longer could prove a catalyst,” he wrote, noting that the 10-year Treasury yield had recently closed at 4.68%. “That is not acceptable to the PE boys, who have been holding their collective breath for a long while now.” The sustained elevation of the long end of the yield curve — a theme that Wall Street’s bond market analysts have been tracking alongside AI hyperscaler debt issuance — compresses the spread dynamics that make the PE-insurance arbitrage viable.
Burry also confirmed that his short positions in certain technology names — including Nvidia and Palantir — remain active, consistent with a broader thesis he has been articulating publicly through 2025 that the AI market has entered bubble territory.
Why It Matters
The core systemic concern Burry and the paper he cites raise is not simply that some PE-owned insurers might suffer losses. It is that the loss socialization mechanism in insurance differs materially from the banking system. When a bank fails, the Federal Deposit Insurance Corporation (FDIC) steps in, backed by federal resources and subject to federal oversight. When an insurance company fails in the United States, policyholders are protected — up to statutory limits — by state guaranty associations, which are funded by assessments on surviving insurers in that state. The authors of the paper Burry shared describe this architecture as one that “socializes losses more sharply than banking’s federal deposit insurance.” In other words, the backstop is thinner, more fragmented, and ultimately reaches ordinary policyholders and, through the assessment mechanism, the broader insurance market.
What makes Burry’s framing particularly pointed is the feedback loop he describes between financial stress and real economic activity. If AI-linked ABS experience elevated defaults or liquidity pressure, the insurers holding them face balance sheet strain. Under stressed conditions, PE sponsors may be unable or unwilling to recapitalize those entities. A forced deleveraging would pull funding from the very data center and chip-lease pipelines that have been financing the AI infrastructure expansion — slowing, or potentially reversing, an investment cycle that has become a meaningful contributor to US GDP growth. The same capital formation story that has driven equity markets higher, in other words, contains within it a credit structure that could become the mechanism of its own unravelling.
This concern does not exist in isolation. The broader shift in corporate AI spending discipline — with enterprises beginning to scrutinize return on AI investment more rigorously — adds an operational layer to the financial risk Burry identifies. If enterprise demand for AI infrastructure softens before the debt underpinning that infrastructure matures, the gap between asset performance and liability obligations inside these insurer balance sheets widens.
Burry himself characterized the current situation with characteristic directness: “This is Private Equity kicking its final can down to the end of that very long road. Taxpayers wait there.” His earlier April commentary, in which he said both private credit and private equity were nearing the “end of the road,” now reads as a precursor to this more specific, mechanistic warning.
What Happens Next
Several developments could accelerate or defuse the risk Burry has identified, and serious market participants should be tracking each independently.
The most immediate variable is the trajectory of long-end Treasury yields. A sustained move higher from current levels — already at multi-year highs — would compress the economics of PE-owned insurance further and potentially force asset sales into illiquid markets. Conversely, a decisive decline in yields would ease the pressure, though it would not resolve the underlying asset-liability mismatch that the academic paper Burry cites treats as a structural concern independent of the rate cycle.
Regulatory scrutiny is a second watch point. State insurance commissioners have begun examining the balance sheet transformations at PE-owned insurers, though the pace and scope of that oversight varies considerably across jurisdictions. A coordinated federal response — whether through the Financial Stability Oversight Council (FSOC) or congressional action — would represent a meaningful escalation. Equity markets have already demonstrated sensitivity to AI credit narratives, and a regulatory signal targeting this specific structure could amplify that sensitivity.
Third, the performance of AI infrastructure assets themselves matters enormously. Data center lease structures and chip financing arrangements are only as durable as the demand supporting them. Any material slowdown in hyperscaler capital expenditure commitments — or a renegotiation of lease terms by large tenants — would flow directly into the cash flow assumptions embedded in the ABS structures Burry is flagging.
Finally, Burry’s short positions in Nvidia and Palantir give his commentary an adversarial dimension that sophisticated readers should hold in mind. His analysis is grounded in a cited academic framework, but he is also a market participant with disclosed positions that would benefit from the scenario he describes. That does not invalidate the thesis — his subprime calls were similarly contested at the time — but it is a material factor in evaluating the weight to assign his public statements. As AI equity valuations come under increasing pressure, the line between analysis and advocacy in public commentary from short sellers deserves scrutiny.
The Implications That Matter
- The taxpayer backstop is the structural story. Unlike bank failures, which trigger federal deposit insurance, the collapse of PE-owned insurers routes losses through state guaranty associations — a thinner, more fragmented safety net that could expose policyholders and the broader insurance market to cascading assessments.
- AI infrastructure financing is a credit story, not just an equity story. The market has priced the AI buildout almost entirely through equity multiples and hyperscaler capital expenditure announcements; the structured credit layer underneath — data center and chip-lease ABS — has received comparatively little public scrutiny and carries distinct risk characteristics.
- Rate policy is the trigger mechanism. Burry’s contagion scenario is rate-conditional. A prolonged high-yield environment narrows the PE-insurance spread arbitrage, and there is limited historical precedent for how these balance sheets perform under sustained pressure at current Treasury yield levels.
- Regulatory fragmentation is a compounding factor. Insurance is state-regulated in the United States, meaning oversight of these transformed balance sheets is uneven and lacks the systemic visibility that federal bank examiners provide; a coordinated regulatory response, if it comes, will likely lag the accumulation of risk.
- The feedback loop to real economic growth is the most consequential variable. If credit stress triggers a withdrawal of funding from data center construction, the macroeconomic consequence extends beyond financial markets — it would directly impair the investment cycle that has been a material contributor to US economic growth in the current expansion.
