Carlyle Group’s Jason Thomas warns AI financing mirrors pre-crisis mortgages

 Jason Thomas, Managing Director and Head of Global Research and Investment Strategy at Carlyle Group, is drawing an uncomfortable parallel between today’s AI infrastructure lending boom and the mortgage finance playbook that preceded the 2008 financial crisis. In his report titled “Let Them Eat Compute: Data Centers’ Implications for Mortgage Finance,” Thomas argues that the sheer velocity of capital flowing into AI-related infrastructure is distorting credit markets, pushing interest rates higher, and squeezing out traditional borrowers, particularly in housing. AI infrastructure spending is consuming hundreds of billions of dollars per quarter, compounding at a 40-60% annualized rate.

The capital vacuum

The numbers paint a vivid picture. Hyperscalers have increased their property, plant, and equipment by 50-200% since late 2023, according to Carlyle’s data. Meanwhile, the US 30-year fixed mortgage rate sat at 6.58% as of August 2025. For new homebuyers, the math is punishing. Existing homeowners locked into lower rates years ago are sitting pretty by comparison, creating a two-tier housing market where the haves and have-nots are separated primarily by when they signed their paperwork.

Government spending isn’t helping. Federal deficits are currently 1.5 times larger than the average from 2010 to 2019, adding even more competition for available capital.

Echoes of 2008

Thomas’s comparison to pre-crisis mortgages centers on lender psychology. Data center financing today rests on the assumption that AI demand will keep growing, that the compute capacity being built will find paying customers, and that the revenue streams will justify the debt. The risk is compounded by concentration. A handful of hyperscale companies are driving the majority of this buildout. Carlyle itself has been active in this space. The firm reported a fivefold return from the 2026 sale of its Copia Power platform to EQT.

Carlyle CEO Harvey Schwartz has publicly expressed a positive outlook on the US economy, which suggests the firm sees these risks as manageable rather than existential. But Thomas’s report is clearly designed as a warning shot: the capital markets are being reshaped in real time by AI infrastructure demand, and the lenders enabling that transformation may not be pricing the downside correctly.

https://cryptobriefing.com/carlyle-ai-financing-mirrors-pre-crisis-mortgages/