BIS warns AI boom could amplify future financial instability risks

The BIS is effectively telling central bankers and finance ministries that the AI boom is now macro‑relevant, not just a tech story. When the institution that worries about payment rails and sovereign debt explicitly names “the risk of an abrupt end to the AI investment boom” alongside inflation and fiscal fragility, it is acknowledging that hyperscale model training, data centers and chip capex have become systemically important flows.([bis.org](https://www.bis.org/speeches/sp260628.htm))

For the race to AGI, that matters in two ways. First, it confirms that frontier model development is unlikely to be constrained by capital in the near term; if anything, cheap money chasing AI returns could overbuild compute and infrastructure. Second, it foreshadows a world where macro‑prudential policy and AI policy intersect. If credit cycles in chips, data centers and AI‑heavy equities become destabilizing, regulators may respond with targeted capital rules, stress tests or even sector‑specific cooling measures. That would be a very different regime from the relatively laissez‑faire environment that powered the first GPT and Claude waves.

The speech doesn’t call for clampdowns on AI investment, but it clearly frames AI as a potential source of asset bubbles and mispricing. That framing will inform how prudential regulators think about large banks’ exposures to AI infrastructure and how tolerant they will be of another “dot‑com style” blow‑up centered on AGI‑adjacent bets.