The AI Investment Race [pdf] (www.bis.org)

A new working paper from the Bank for International Settlements, authored by Phurichai Rungcharoenkitkul, examines what it terms "The AI Investment Race," highlighting the unsustainable nature of the current AI investment boom. With capital expenditures from hyperscalers projected to exceed $700 billion in 2026 and potentially reaching trillions in the coming years, the paper warns that firms are over-committing resources as they compete for limited market dominance. This intense competition breeds financial fragility, suggesting that firms are potentially setting themselves up for a significant downturn as they could be over-investing by a factor of 1.5 to 3 times the optimal level. The work underscores the interconnected financial vulnerabilities prevalent in this boom, illustrating how debt and circular financing arrangements—where firms take equity stakes in labs in exchange for future commitments—create a network of exposure among companies. A downturn in one firm could trigger a cascading effect across the sector. The findings are significant for the AI/ML community as they call into question the sustainability of current investments, emphasizing the need for scrutiny regarding financial practices that may lead to a future bust, which could disrupt not just the AI sector, but broader economic stability as well.