Artificial Intelligence Investments Pose Financial Risks to the US Economy

A new study by Stijn Van Nieuwerburgh, a professor of finance and real estate at Columbia Business School, shows that artificial intelligence infrastructure investments could consume more than 3.6 percent of annual US Gross Domestic Product by 2032, potentially creating systemic risks through complex financial structures.

Trillions of Dollars in Investment Needs

It is stated that the artificial intelligence infrastructure push requires a much larger share of US output than the rollout processes of electricity, railways, interstate highways, or the internet.

The process, financed by companies like Amazon, Meta, and Google using their accumulated cash, has evolved into a structure that will exceed 3.6 percent of annual GDP by 2032 and require over 10 trillion dollars in resources.

Historical Comparisons and GDP Ratios

It was emphasized that this cost estimate is higher than the 2.2 percent share of annual GDP absorbed by railways in the late 1800s.

Additionally, it was noted that this ratio exceeds the 1 percent annual share spent on the construction of the US highway system or telecommunications expansion.

Complex Financing and Subprime Analogy

The study stated that complex financing structures forming around artificial intelligence, unproven revenue streams, and increasing borrowing could trigger potential collapse risks.

Attention was drawn to the fact that the lack of transparency in special purpose vehicles somewhat resembles the past subprime mortgage crisis.

Potential Opportunities and Downside Risks

It is indicated that strong growth, high utilization rates, and continuous improvements in model capabilities could support the infrastructure and generate stable cash flows.

Nevertheless, the combination of uncertain demand, rapid technological change, deployment bottlenecks, and high leverage creates significant downside risk.