I forgot who linked to this originally, but Groundbreaker has one of the more finely-worded perspectives about what’s happening with AI in the financial markets.
Essentially, just as the subprime mortgage machine of 2008 relied on ever-accelerating home prices to support refinancing, current AI infrastructure relies on sustained, accelerating growth to service massive take-or-pay debt commitments. These debt commitments end up functioning as a massive, opaque “loan book” rather than standard capital expenditure budgets.
Companies like OpenAI, the author argues, operate as a subprime borrower in this system, relying on constant equity valuation markups to finance its operations. And it’s apparent that OpenAI lacks sufficient profit to cover its massive obligations.
We could argue that another industry saw something similar not too long ago, albeit sans real estate/credit/energy, but with cars, roads, and people (Uber), and that — shockingly — turned out okay, but this is a much more entangled situation:
Three errors, stacked, recreate 2008. The market is pricing AI as a technology cycle when its financing is the machinery of a credit-and-real-estate cycle. It is watching the level and the velocity while the structure breaks on the acceleration.