Were you to glance at the trajectory of top AI stocks — OpenAI, Microsoft, Nvidia, and the like — you’d be convinced the industry is making money hand over fist.
Look a bit deeper, however, and cracks start to show in that facade, betraying one massively inconvenient truth: that the AI industry has not yet figured out how to be profitable, and possibly never will.
In interviews with the New York Times, even the most enthusiastic AI proponents were unable to spin the technology (and the industry surrounding it) into something that resembles profitability.
“The raw technological horsepower [of AI] is terrific,” bemoaned MIT research scientist and AI consultancy founder Andrew McAfee, “but it’s not going to determine how quickly AI transforms the economy.”
There is, of course, one massive reason AI is not the ROI machine investors hoped it would be: that it costs a staggering amount of money to run, and is expected to get more expensive as operations advance and scale up.
As McKinsey revealed in a report earlier this year, its research suggests that by the year 2030, AI data centers will need to spend a whopping $6.7 trillion on computing to keep up with demand. With estimates from software company Hartinger pegging the AI industry’s market size at an approximate $305.9 billion by the end of this year, it’s hard to imagine trillions of dollars even coming into the industry over the next five years, much less any funds beyond that.
Though it’s an important indicator of demand, market size is not, of course, the same thing as profitability. As tech journalist Ed Zitron noted earlier this year, OpenAI spent the entirety of its $4 billion revenue on running and training models.
While the hype cycles surrounding AI suggest that each new model will be groundbreaking and bring the world closer to artificial general intelligence (AGI) or human-level intelligence, AI companies are falling far short of those goals so far. A prime example of these diminishing returns is OpenAI’s GPT-5, whose launch turned out to be a dud.
With each new dud model, it appears more and more like AI progress has reached a plateau, as critics have argued. It’s no surprise, then, that executives who invested in AI are starting to get cold feet.
In a report released in May, work management software company Asana found, upon surveying nearly 4,000 people who work in IT, that 29 percent — or one in three — companies that bought into AI in 2024 now regret the decision. As that report put so succinctly, “the 2024 rush to deploy AI has given rise to a sobering reality.”
Those regrets have seemingly led to some course-correcting actions. In a similar survey put out in March that surveyed 1,000 companies that had invested in AI, S&P Global Market Intelligence found that 42 percent had already abandoned those endeavors — a steep increase from the 17 percent who did away with their AI projects in 2024.
JPMorgan chief information officer Lori Beer told the NYT that in the wake of the bank’s decision to restrict staff from using ChatGPT, she has kiboshed hundreds of additional AI projects.
“We’re absolutely shutting things down,” Beer told the newspaper. “We’re not afraid to shut things down. We don’t think it’s a bad thing. I think it’s a smart thing.”
While AI boosters may try to paint these investment failures positively — McAfee told the NYT that “innovation is a process of failing fairly regularly” — it’s hard to see AI as anything other than a massively-inflated bubble preparing to burst.
More on AI failures: Billion-Dollar AI Company Gives Up on AGI While Desperately Fighting to Stop Bleeding Money