Treasury report warns of AI bubble and growing risks

A draft report from the U.S. Treasury Department has raised concerns about the risks within the AI economy, drawing parallels to the dotcom bubble of the early 2000s. The document highlights a characteristic financial dynamic where key players are investing in each other.

The report, obtained by the nonprofit newsroom NOTUS, signals jitters over AI development amid sagging stock prices for AI industry players, particularly chip makers. Investors are reportedly reassessing the real scale and progress of the AI boom.

At the core of this are a few major U.S. AI labs investing billions to build and rent out large, general-purpose language models. These models, developed by entities like OpenAI and Anthropic, are exceptionally expensive to operate. However, many enterprises are increasingly turning to smaller, open-source or open-weight models, which are more cost-effective and sufficient for a wide range of business tasks.

External experts, such as Professor Howard Yu from IMD Business School, suggest that open-weight models directly challenge the competitive advantage of U.S. AI labs. He notes that cheaper, good-enough alternatives can displace larger, more expensive systems in the market.

Treasury analysts warn that the risks associated with the AI sector are considerable, and potential issues could impact the broader U.S. economy. Some proposals suggest the government should consider taking an equity stake in domestic AI companies.