When Barron’s asked in March 2000, “When will the internet bubble burst?” the answer came swiftly, as tech stocks began to collapse and trillions in market value evaporated. Today, more than two decades later, some investors are wondering whether the same fate awaits the artificial intelligence boom.
This week, those concerns were reignited when researchers at the Massachusetts Institute of Technology (MIT) released a report claiming that most corporate investments in generative AI are producing “zero return.”
According to MIT, despite $30–40 billion spent on enterprise AI, 95 percent of projects have failed to deliver measurable profits. The study found that half of corporate AI initiatives collapse entirely, while only a small fraction ever progress from pilot programs to full production.
The findings shook investor confidence. Shares of Nvidia, the $4 trillion chipmaker at the heart of the AI surge, slipped 3.5 percent, while Palantir lost 9 percent.
The MIT team stressed that while employees are often eager to use AI, many prefer consumer tools like ChatGPT over expensive or cumbersome corporate systems. Their conclusion: AI is already reshaping work, just not in the way executives envisioned.
For skeptics, the results add weight to the argument that the AI frenzy bears the hallmarks of a bubble. “Sounds about right for a bubble,” said Marko Kolanovic, former JP Morgan research chief.
Yet even as doubts surface, many in Silicon Valley and on Wall Street remain confident. AI backers argue that revolutions rarely happen overnight and that meaningful returns take years to materialize.
Sam Altman, OpenAI’s chief executive, acknowledged that investors may be “overexcited” but stopped short of predicting a collapse. Meanwhile, Wedbush Securities analyst Dan Ives insists the “tech bull cycle will be well intact for another two to three years,” brushing off the sell-off as a short-term wobble.
The stakes are enormous. Morgan Stanley expects global data center investment to top $3 trillion within three years, fueled by AI demand. It also projects AI could add $16 trillion in value to the S&P 500 through cost savings and productivity gains.
Still, if MIT is right and most enterprise AI tools fail to deliver, those lofty forecasts may need revising.
In the meantime, companies are adjusting. Meta, one of AI’s biggest spenders, is reportedly reorganizing and downsizing parts of its AI division. And OpenAI’s launch of ChatGPT-5 drew a muted response, with many calling its improvements incremental.
For now, markets are in a holding pattern. The next major test comes with Nvidia’s earnings report next week. The company has consistently smashed expectations, and its results are likely to reveal whether corporate AI spending is slowing or simply taking a breather before the next wave.