The world’s central banking watchdog is raising red flags about the explosive growth in artificial intelligence spending — even as one of the tech world’s most powerful investors calls such concerns an offense against the very idea of AI itself.
In its most recent annual review of global financial conditions, the Basel-based Bank for International Settlements released a report Sunday that stopped short of questioning whether AI investment will ultimately pay off — but expressed serious concern about what happens when the spending wave finally peaks.
The backdrop is striking: even as military conflict between the U.S. and Israel against Iran continued in the background, American chipmaker stocks surged a record 75% during the second quarter of 2026. That rally was fueled by yet another round of upward revisions to capital spending forecasts from the largest technology companies, all racing to build out AI infrastructure — a race that has created supply bottlenecks and chip shortages along the way.
The surge pushed U.S. earnings growth projections for 2026 to nearly 25%. The five biggest technology hyperscalers are expected to collectively spend close to $1 trillion on AI infrastructure this year alone, and Goldman Sachs estimates the cumulative total could reach $7.6 trillion by 2031.
So is this a bubble?
Absolutely not, according to SoftBank chief Masayoshi Son, one of the technology sector’s most prominent and deep-pocketed supporters. Speaking just last week, Son made clear he considers the very question an insult to the potential of artificial intelligence.
“It’s blasphemy against AI if you say it’s a bubble,” Son said. “It’s just the beginning. AI’s potential will be unlocked.”
With hundreds of billions of dollars on the line should the AI boom continue, his confidence is perhaps understandable.
Still, other investors — perhaps worn down by years of debate — are starting to come around to the idea that the spending explosion and soaring stock prices may not represent the bubble they once worried about. A quarterly client survey released Tuesday by Deutsche Bank showed bubble risk perceptions for the so-called Magnificent Seven megacap stocks — the companies at the center of the AI spending wave — have fallen to their lowest level since 2021.
For the broader U.S. technology sector, however, bubble concerns remain as high as they’ve been over the past two years. That split may help explain why, even as chipmakers enjoyed their best quarter on record, June turned out to be the worst single month ever recorded for the Magnificent Seven group since it was first defined three years ago.
For many chipmaker companies, the financial results are very real. Micron Technology, now valued at $1.25 trillion, has seen its stock more than triple since March — but revenue estimates have climbed just as steeply, leaving its forward price-to-earnings ratio essentially unchanged for the year. At just eight times forward earnings, that valuation is less than half what it was two years ago.
While the near-tripling of still-loss-making Intel’s stock stands out as something of an exception, valuations for chipmakers such as Broadcom and Qualcomm remain within historically normal ranges.
Even so, it remains the job of financial stability monitors like the BIS to spell out what could go wrong — and their latest report does exactly that.
The BIS focused its concerns primarily on the long-term sustainability of the current investment pace, particularly given that a small number of firms are competing on the assumption that only a handful of players with the best technology will ultimately control the market.
That fierce rivalry, the watchdog cautioned, could push companies to pour money into AI projects whose returns are still far from guaranteed. If the promised payoffs don’t materialize, the entire sector could be left exposed.
With competitive pressure continuously driving capital spending higher, the BIS warned that the sector’s overall return on investment — after accounting for costs — could shrink or even go negative in a bad scenario.
“Disappointment in returns could trigger a sudden pullback in financing and turn the capex boom into a protracted investment bust, with potential knock-on effects on financial conditions,” the report stated.
The BIS also flagged supply shortages in power generation, electrical grid capacity, and memory chips as additional risks. These constraints could push companies into long-term supply contracts to lock in scarce resources — potentially leaving them overexposed if demand doesn’t hold up.
The report’s most striking warning, however, concerns the possibility that AI’s very success could undermine the economy that supports it. In an extreme scenario where AI proves as capable of replacing human labor as its most enthusiastic advocates believe, a growing share of national income could flow away from workers and into further AI investment. If taken far enough, workers’ share of the economy could approach zero — leaving too few people with enough purchasing power to sustain demand for what the economy produces.
Recognizing that problem in advance, the BIS suggests, rational companies would eventually stop investing further.
“Productivity stalls not because of technological limitation, but because the demand to justify further capacity expansion is missing. The demand bottleneck becomes the binding constraint,” the report concluded.
Blasphemy or not, the BIS is reminding the world that an enormous amount of money in today’s AI arms race is still being wagered on faith.
The opinions expressed are those of Reuters columnist Mike Dolan.
