Vanderbilt Policy Accelerator warns AI infrastructure investment vastly outpaces revenue; proposes seven congressional reforms to prevent systemic crash

A paper published by the Vanderbilt Policy Accelerator argues that $5 trillion in AI infrastructure investment planned over the next five years, coupled with opaque financial engineering, poses systemic economic risk. OpenAI and Anthropic currently earn annualized revenues of $25 billion and $19 billion respectively, creating a massive mismatch between investment and earnings.

The author identifies seven areas for congressional action: ban circular equity financing (where chip and cloud companies invest in AI firms that then buy their products); require disclosure of off-balance-sheet debt deals currently hiding more than $100 billion; prosecute fraud; build public cloud infrastructure from bankrupt companies' assets; expand worker protections; structurally separate software from hardware ("Glass-Steagall for AI"); and regulate chips, cloud computing, and foundation models as utilities.

AI-related investment accounted for more than 90% of U.S. GDP growth in the first half of 2025. By one macroeconomic measure, the AI bubble is already 17 times larger than the dot-com bubble and four times larger than the 2008 housing bubble. The dot-com bubble led to 200,000 job losses, thousands of company bankruptcies, and an 80% peak-to-trough Nasdaq decline that took 15 years to recover.