Financing historic AI buildout raises systemic risks in US, researcher says
QQQ•A Columbia Business School professor estimates the AI buildout will require more than $10 trillion, or about 3.6% of annual US GDP through 2032. He says complex financing and unproven revenue streams create downside risks, though financial distress is not imminent.
1. Investment scale
The AI buildout is expected to require more than $10 trillion through 2032, equivalent to around 3.6% of annual US GDP, according to a study by Columbia Business School professor Stijn Van Nieuwerburgh. He estimated the spending will fund 183 gigawatts of new data-center capacity over seven years, compared with about 57 gigawatts currently installed.
2. Financing risks
The investment has outgrown what major players can fund from their own cash flows, leading to more outside financing and greater leverage, Van Nieuwerburgh wrote. He said complex arrangements involving AI firms, major technology companies, banks, private credit lenders and real estate firms could spread risks across the economy.
3. Revenue hurdle
The industry would need to generate about $3.7 trillion in annual revenue by 2032 to achieve the expected return on investment, the professor estimated. He noted that combined annual revenues for OpenAI and Anthropic are currently around $100 billion, requiring roughly 80% annual growth; he added that strong AI demand and utilization could support the investment, but uncertain demand and high leverage create meaningful downside risk.
