AI bubble is the only thing keeping the US economy together, Deutsche Bank warns (www.techspot.com)

Deutsche Bank warned in a client note that the current AI spending binge—chiefly Big Tech’s massive build-out of AI data centers and purchases of accelerators like Nvidia GPUs—is artificially propping up the US economy and could be unsustainable. Global Head of FX Research George Saravelos argued the U.S. would be close to recession without this surge in capital expenditure, which is largely driven by construction and hardware investment rather than AI services revenues. The bank and other analysts highlight stark concentration risk: roughly half of recent S&P 500 gains have been tech-driven, Nvidia and OpenAI headline multibillion-dollar commitments (e.g., Nvidia’s reported $100B toward 10 GW of new AI capacity), and investors are heavily exposed to an unproven “AI economy.” The significance for AI/ML is twofold: technically, the sector demands parabolic capex to scale (data centers, accelerators, power, networking), but commercially projected service revenues may not materialize—Bain estimates a 2030 demand requiring $2T in annual AI revenue, leaving an $800B shortfall. That mismatch raises bubble, allocation, and policy risks: if spending decelerates, macro growth and equity markets could reprice sharply; if it persists, capital may be wasted on low-value “AI washing.” Industry leaders including Sam Altman and Baidu’s Robin Li have warned of irrational investment and high failure rates, underscoring large downside for investors and an unsettled path to sustainable AI-driven productivity.