A hype makes a bubble: AI has been a magnet for heavy investments, which are now being feared as a function of unrealistic promises, creating the so-called ‘AI bubble.’ If AI fails to deliver, that bubble will be considered as having popped. Whether it really is a bubble is largely disputed among tech executives and investors, making the topic a high-stakes bone of contention. If the bubble pops, it will also hurt people outside the tech business and stock markets, as warned earlier by OpenAI CEO Sam Altman. He had said that if the turbulence in AI investments worsens, it could bring down the entire economy.
At the Web Summit in Lisbon (Portugal), which started on November 11, Microsoft president Brad Smith was asked if AI had become a bubble. “From a long-term perspective, I think the answer is no … I think we’ve got years, if not decades, ahead of us to grow,” he responded.
But according to J.P. Morgan’s AI CapEx (Capital Expenditure) Report, the AI industry will have to make $630 billion in annual revenue to deliver the expected 10% return on investment through 2030, which the report called “an astonishingly large number.” The report was shared by Technology Analyst Max Weinbach on X.
In another major announcement amid the AI bubble scare, SoftBank, a Japanese investment holding company, disclosed that it had recently sold its entire stake in chip giant Nvidia for $5.8 billion. But Yoshimitsu Goto, SoftBank’s chief financial officer, publicly said, “I can’t say if we’re in an AI bubble or not,” adding that the sale had “nothing to do with Nvidia itself.” After the sale, SoftBank’s profits spiked.
How did it all start?
The ‘Magnificent 7’ (or ‘Mag 7’) is the nickname given to the group of the largest tech companies in the US stock market, comprising Apple, Microsoft, Alphabet (Google’s parent company), Amazon, Nvidia, Meta, and Tesla. This group has been leading the US markets for at least the last decade.
“The reason why the Mag 7 did well is because of one simple reason: they were running a very low-risk business model. Most of them were the only game in town, and all these companies became monopolies because of their tech edge, not their balance sheet edge; that is a critical thing to understand,” Shankar Sharma, seasoned financial analyst, told HT. Sharma is often referred to as “The Alchemist of Dalal Street” by Forbes magazine and is currently based in Dubai.
“[Mag 7] had such a great edge because of research and development, and innovation that they were hard to compete with. Now, they are changing their business models and becoming asset-heavy businesses because of this huge CapEx that they are going to be doing on AI,” he said. “Anything that has CapEx becomes effectively a manufacturing business. CapEx is all about big assets on the ground, which reduces your business flexibility, and markets don’t like that.”
Sharma explained that the Mag 7 used to be service businesses that focused on OpEx (Operating Expenses), but that’s now changing. “Suddenly, from being asset-light, tech-driven companies, [Mag 7] are becoming more like traditional companies with huge balance sheets and CapEx. That is where the core of this [AI bubble] problem lies. And it’s still not certain to anybody whether that CapEx is going to generate any worthwhile return on investment,” he said.
Would you invest your money in a Mag 7 for the next five years?
Overinvestment is also a problem, according to Sharma. “We had overinvestment in financials in the 2007-08 crisis, and we had overinvestments in the dot-com boom, undersea cables, broadband, etc., and all those went to virtually zero. The markets fear that AI may be another giant bubble of overinvestment in infrastructure and CapEx, which may not end well. Any CapEx-heavy boom ultimately ends badly.”
Sharma says that he won’t be betting on any tech company long-term. “I would not be betting anything close to five years in any tech name right now. They are changing their business models dramatically, and I believe that anything that requires so much capital expenditure with such poor payoff, horizons, and spectrum is not worth taking a five-year view on.”