When Jessica Wachter, a finance professor at the University of Pennsylvania’s Wharton School, assessed AI’s economic impact, she started with a "remarkable fact": hyperscalers are investing huge sums in AI data centers. Instead of predicting AI usefulness, she calculated how fast their earnings must grow to justify spending through 2027, when expenditures will reach nearly $1.1 trillion. The analysis shows AI companies must increase productivity by a factor of 2.7 to break even by 2030, accounting for cost of capital and 15% return.

Wachter, formerly the SEC’s chief economist, notes this would require growth comparable to the 1990s US IT boom but compressed into a few years. If unmet, hyperscalers risk falling behind on interest payments and bankruptcy. She concludes that a failed productivity boom would make the current buildout "the largest misallocation of capital in history." Hyperscalers—Alphabet, Microsoft, Amazon, Meta, and Oracle (partnering with OpenAI)—will spend about $750 billion this year on data centers, with total AI capital investments potentially exceeding $5 trillion over four years.

Yet current AI revenues are only $150–200 billion annually, per Gary Gensler, former SEC chair and MIT Sloan professor, who calls the spending-revenue mismatch "a fact." The investments could reach 3% of US GDP, raising risks to corporate financial health and the broader economy as free cash flow turns negative across the group.