Stijn Van Nieuwerburgh is a financial economist at Columbia Business School. Brookings identifies him as the Earl Kazis and Benjamin Schore Professor of Real Estate at Columbia University (Source: brookings.edu). He is the author of "Financing the AI Buildout," a paper for the fall 2026 edition of the Brookings Papers on Economic Activity, presented at the BPEA conference on September 25, 2026 (Financing the AI Buildout (Van Nieuwerburgh, BPEA Fall 2026)).
Work on AI infrastructure finance
In the BPEA paper Van Nieuwerburgh estimates that a 200 MW AI training campus costs about $8.2 billion, about two-thirds of it IT equipment, or about $41 billion per GW. Using Cleanview project-level data, his central scenario has 182.7 GW of U.S. data-center capacity completed during 2025–2032, implying $10.3 trillion of investment over the period, an average of 3.63% of GDP, which he compares with the canal, railroad, electrification, highway and telecommunications booms. He calculates that the capacity would require about $3.725 trillion of annual revenue at maturity to earn a 10% unlevered return at a 50% operating cash-flow margin, rising to about $6.0 trillion if IT equipment lasts three years instead of six (Financing the AI Buildout (Van Nieuwerburgh, BPEA Fall 2026)).
His argument is that financing is shifting from hyperscaler balance sheets to leases, joint ventures, project debt, private credit, securitization and special-purpose vehicles, which "transform rather than eliminate risk" by raising leverage at the asset level and making contingent obligations harder to see. He uses Meta's Hyperion data-center financing, in which the Meta–Blue Owl joint venture Beignet raised $27 billion of debt against a roughly $30 billion project, as the worked example. He writes that "it would be premature to conclude that AI infrastructure already poses systemic risk comparable to earlier credit booms," and argues that the priority is better measurement and disclosure of who bears demand, refinancing and residual-value risk (Financing the AI Buildout (Van Nieuwerburgh, BPEA Fall 2026)). The paper cites a follow-on working paper with Z. Sun, "Clouds of credit: Mapping the financial web behind AI compute" (Columbia Business School, dated November 2026), as ongoing work on that measurement problem.
Relationships
- related: Financing the AI Buildout (Van Nieuwerburgh, BPEA Fall 2026) — author
- related: Brookings Institution AI Initiative — BPEA publisher
- related: Private Credit & AI Infrastructure, AI Bubble Debate, AI Infrastructure Capex — debates the paper enters