Financing of historic AI buildout raises systemic risks in US, researcher says

A new study warns that the financing of the AI buildout in the United States could pose significant systemic risks. According to professor Stijn Van Nieuwerburgh of Columbia Business School, the AI expansion is projected to consume around 3.6% of the nation's GDP annually through 2032, which is significantly higher than past technology rollouts like railroads (2.2% of GDP) and the interstate highway system (1% annually).

The intricate financing arrangements involved in the AI buildout, which are using ever more complex structures to fund the construction of approximately 183 gigawatts of new data center capacity over the next seven years, may lead to bubbles, busts, and financial distress similar to those seen in the subprime mortgage crisis of 2007-2009. Despite the potential for strong growth in AI applications and stable cash flows, the reliance on uncertain revenue streams and high leverage creates meaningful downside risk.