Business and financial media have been reporting on the impending danger of an AI “bubble.” A bubble in this context is when investment in an industry keeps growing and growing past all hope of financial return, whereupon at some point the bubble pops, resulting in large scale financial loss.
In his Unherd article Is the AI Bubble About to Burst?, John Rapley reports that AI is now “sucking up more than half of America’s [Venture Capital] investment” (my emphasis). Referring to AI mogul Sam Altman’s goal of raising $7 trillion, Rapley remarks, “So insatiable is this bubble that there may not even be enough cash in America to satisfy OpenAI.” (The amount of cash in circulation in the United States today is about $2.4 trillion. Seven trillion dollars would be 26% of the entire U.S. Gross Domestic Product of $27 trillion.)
Rapley’s point is that investment in AI is gobbling up money that would otherwise go into our ventures, to the detriment of the nation’s economic growth. He says that if we factor out the AI companies from the stock market, corporate profits, and other economic indicators, the U.S. economy would be flat.
So much investment could be justified if the product made enough money. But that isn’t happening with AI.
A Futurism article by Joe Wilkins is entitled AI Industry Nervous About Small Detail: They’re Not Making Any Real Money. “Put into monetary terms,” Wilkins reports, “in 2024, OpenAI was burning through $2.50 for every dollar it brought in.” While start-ups often take up to three years to show a profit, OpenAI has been around for six years. The company now projects that it will start making money in 2029, if everything goes right.
But it hasn’t been going right! Heralded improvements, such as ChatGPT 5.0, have been a fiasco, and the industry faces a host of legal challenges, from the unauthorized use of copyrighted material in AI training to lawsuits over hallucinations and people dying because of AI “relationships.”
Recently, research from MIT found that 95% of the companies that invested in AI are getting zero return. The MIT researchers also found that while employees are indeed using AI, they are not using AI-enabled software from their companies or specialized products. Rather, they are just using ChatGPT, which they can use for free.
Sam Altman himself is admitting that there is an AI bubble. “When bubbles happen, smart people get overexcited about a kernel of truth.” That does not negate the value of the product they are investing in. “Someone is going to lose a phenomenal amount of money,” he said. “We don’t know who, and a lot of people are going to make a phenomenal amount of money.”
But the problem is not just with the software. AI takes an enormous amount of computer power. So new data centers have to be built. Which, in turn, take an enormous amount of electricity.
According to the McKinsey Quarterly,
Our research shows that by 2030, data centers are projected to require $6.7 trillion worldwide to keep pace with the demand for compute power. Data centers equipped to handle AI processing loads are projected to require $5.2 trillion in capital expenditures, while those powering traditional IT applications are projected to require $1.5 trillion in capital expenditures. . . . Overall, that’s nearly $7 trillion in capital outlays needed by 2030—a staggering number by any measure.
That’s where Altman’s goal of reaching that number comes from. But it isn’t just that. You have to keep spending money on those data centers.
Joe Wilkins has written another article for Futurism entitled There’s a Stunning Financial Problem With AI Data Centers. He draws on the analysis by Harris Kupperman, who is quoted in the first paragraph of this post.
Kupperman works out rough calculations for the genuine cost of a data center, factoring in the inevitable breakdown of parts over time. Each data center, he says, is essentially made up of three components: the chips, which become obsolete in just a few years; the systems connecting the chips, which need to be replaced every decade or so; and the building itself, which should last for quite a while.
Add it all up, and time is not on the data center’s side. The finance guru estimates that the “AI datacenters to be built in 2025 will suffer $40 billion of annual depreciation, while generating somewhere between $15 and $20 billion of revenue.”
Kupperman crunches some more numbers and puts them in perspective (my bolds):
For context, Kupperman notes that Netflix brings in just $39 billion in annual revenue from its 300 million subscribers. If AI companies charged Netflix prices for their software, they’d need to field over 3.69 billion paying customers to make a standard profit on data center spending alone — almost half the people on the planet.
“Simply put, at the current trajectory, we’re going to hit a wall, and soon,” he fretted. “There just isn’t enough revenue and there never can be enough revenue. The world just doesn’t have the ability to pay for this much AI.“
Photo: POP by asmartblonde via Deviant Art, CC BY-NC-ND 3.0
Ad-Free & Subscriber Log-In