U.S. AI infrastructure buildout may raise systemic financing risks

Traders Union
All News
Financial News
AI infrastructure financing risks

Highlights

AI infrastructure buildout is projected to consume 3.6% of U.S. GDP annually through 2032, surpassing historic infrastructure booms and totaling over $10 trillion.
The industry is moving from tech giants’ cash reserves to complex financing involving banks, private lenders, and real estate, with 183 gigawatts of new data center capacity planned versus 57 gigawatts today.
Van Nieuwerburgh warns that rising leverage, unproven revenue models, and $3.7 trillion in required annual revenue by 2032 create systemic downside risks for the U.S. financial system.

Study flags scale and structure of AI funding

As reported by Reuters, citing Brookings Institution conference materials, a paper by Columbia Business School finance and real estate professor Stijn Van Nieuwerburgh says the AI buildout is on track to consume about 3.6% of U.S. gross domestic product annually through 2032, or more than $10 trillion.

That estimate exceeds the roughly 2.2% of annual GDP absorbed by railroad expansion in the late 1800s, and is also above the levels associated with the rollout of the interstate highway system and the telecommunications buildout tied to the internet era. Van Nieuwerburgh says the push has moved beyond being financed mainly from the cash reserves of large technology groups such as Amazon.com, Meta Platforms and Alphabet's Google.

He says the industry is increasingly relying on more elaborate arrangements involving AI firms, hyperscalers, banks, private credit lenders and real estate companies to support what he conservatively estimates at 183 gigawatts of new data center capacity over the next seven years, compared with about 57 gigawatts currently installed. In comments to reporters, he describes the financing web as highly complicated and says the opacity of special purpose vehicles carries echoes of the subprime mortgage structures that amplified the 2007-2009 financial crisis.

Revenue uncertainty adds to downside risks

Van Nieuwerburgh says the funding shift is increasing leverage, spreading risk more broadly across the economy and making the investment case more dependent on future revenue that has yet to be proven. He writes that these conditions do not mean financial distress is imminent, because strong growth in AI applications, high utilization and continued model improvements could still generate stable cash flows.

Still, he warns that uncertain demand, rapid technological change, execution bottlenecks and high leverage create meaningful downside risk if expectations weaken. As one example, he estimates the AI industry would need to generate about $3.7 trillion in annual revenue by 2032 to deliver the expected return on investment, implying annual growth of roughly 80% from current combined revenue estimates of about $100 billion for OpenAI and Anthropic.

Data center construction and the broader risks tied to AI are already becoming a central issue in U.S. political and economic debates. Some localities are showing greater reluctance to host new facilities because of pressure on local resources, while Federal Reserve officials are weighing whether the building boom is adding to inflation, and some AI executives suggest a slower pace of development could be safer.

In our earlier article on SoftBank’s $11.1 billion bond issuance, we outlined how the group raised record high-yield debt to expand its AI, semiconductor and digital-infrastructure push, including increased capital commitments linked to OpenAI and data centers. We also noted that the transaction highlighted SoftBank’s growing reliance on debt markets to finance large-scale technology bets alongside a widening portfolio of AI-related assets.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.