Another bubble is forming around the AI debate. And this one is unlikely to pop anytime soon. (www.businessinsider.com)

The “AI bubble” debate has intensified this week as high-profile skeptics and corporate moves spotlight the risks of massive infrastructure spending. Michael Burry — famed for shorting the housing market — warned that hyperscalers like Meta, Oracle, and Microsoft are misestimating semiconductor depreciation, estimating about $176 billion of understated wear. JPMorgan drew a dot‑com parallel, cautioning that heavy outlays on data centers and networking could replicate the fiber‑optic overbuild that didn’t produce expected returns. Meanwhile SoftBank sold its entire $5.8 billion Nvidia stake, not exiting AI but reallocating capital to double down on its OpenAI exposure. Despite the warnings, the market’s behavior suggests this isn’t unwinding soon: firms are on track to invest trillions into compute, data centers, and chips, and Big Tech executives show no sign of reversing course. The upshot for AI/ML practitioners and investors is a longer, lumpy path to profitability and meaningful risk of stranded assets if hardware lifecycles and capacity utilization don’t align with revenue growth. Still, AI’s strategic centrality to product roadmaps and Wall Street’s dependence on a successful transition mean heavy capital deployment will likely continue — keeping the debate loud and unresolved for the foreseeable future.