Is A.I. a Bubble? We explain the fears on Wall Street (www.nytimes.com)

Wall Street’s recent rally has become heavily dependent on AI: by one estimate, 80% of U.S. stock gains this year came from AI-linked companies, and the S&P 500 has hit 30+ record highs. That concentration and sky-high valuations have prompted warnings from figures such as Jeff Bezos, Sam Altman, Jamie Dimon and David Solomon, who fear a market correction that could drag the broader economy down if AI winners falter. The core question investors and technologists are asking is simple but consequential: is AI a sustained boom or an overblown bubble? The bullish case rests on the “scaling hypothesis” — the idea that more chips, data and compute power reliably improve model capabilities — which justifies massive infrastructure bets. Recent examples include Nvidia and AMD deals reportedly worth about $100 billion to populate OpenAI data centers, Amazon’s plan to spend over $100 billion on AI data centers this year, and Meta’s pledge of more than $600 billion over three years. For the AI/ML community this means unprecedented demand for accelerators, datasets, and power, plus rapid commercialization opportunities. The flip side: if scaling yields diminishing returns or monetization lags, valuations could correct sharply, curtailing funding, hiring and large-scale projects. Either outcome will reshape research priorities, hardware supply chains and where capital flows in AI for years to come.