VC Paul Kedrosky Warns AI Infrastructure Boom Echoes 2008 Financial Crisis

Paul Kedrosky, a venture capitalist, has warned that the current boom in artificial intelligence infrastructure is structurally similar to the run-up to the 2008 financial crisis, but on a “vastly larger scale.” Speaking on the Between Two COOs podcast, Kedrosky described the current economic environment as a “slop economy,” where companies are producing vast amounts of software, presentations, and content, even when little additional value is created.

Kedrosky pointed to an NBER study involving more than 500,000 GitHub developers, which found that autonomous coding agents increased coding activity by 240%. However, this gain translated to only an 80% increase at the project level and a 30% increase in actual software releases. He noted that junior bankers are using AI to generate hundreds more investment decks and pitch transactions that may never happen, suggesting a lack of rationale for the increased output.

The concern centers on the financing of data centers. Kedrosky estimates that more than 60% of AI financing is now debt-backed, a significant increase from 15% to 20% a year ago. Amazon.com Inc. (NASDAQ: AMZN) is a major example of this spending, with the company expecting capital expenditures of about $220 billion this year to fuel AI and cloud capacity. Prediction traders currently give Amazon a 61% chance of exceeding this capex target.

Kedrosky argues that the data centers being built act as “token factories” producing a “hyper-deflationary industrial commodity.” He stated that higher Treasury yields and credit spreads are pushing required returns on stressed data-center projects toward 10% to 12%. He warned that some of the five-year financing structures behind the AI buildout will hit a refinancing wall around 2029 and 2030, predicting that refinancing will occur at prohibitive terms.

Despite these warnings, Kedrosky acknowledged that AI is a “tremendously important technology,” likely the most consequential of his lifetime. Prediction markets currently place the probability of an AI industry downturn by Dec. 31 at about 10%.