While most eyes are on stock prices, Sebastian Siemiatkowski, the chief executive of Klarna, is looking at the physical infrastructure of the AI boom—and he doesn’t like what he sees. In a candid warning, he expressed deep nervousness regarding the billions of dollars being poured into data centers and computing chips. He fears that the industry is building out capacity “without some more thoughtful thinking,” potentially creating a massive glut of useless hardware.
This “infrastructure bubble” theory is gaining traction. Siemiatkowski argues that as AI models become more efficient, they will require less computing power, rendering the massive server farms currently under construction obsolete. If this happens, the trillions of dollars in capital expenditure by big tech companies will turn into massive losses, dragging down the entire stock market.
The market seems to be waking up to this reality. Nvidia, the primary beneficiary of this spending, has seen its valuation questioned even as it hits $4 trillion. The ripple effects are already hitting the crypto market, which utilizes similar hardware; the sector has shed $1 trillion as investors re-evaluate the “compute at any cost” narrative.
Broad market indices are reacting with caution. The FTSE 100 and Stoxx 600 have posted significant losses, reflecting a move away from capital-intensive growth stocks. When a fintech leader questions the valuation of the very technology supposed to revolutionize finance, it signals a crack in the consensus.
If Siemiatkowski is right, the current correction—predicted by JP Morgan’s Daniel Pinto—is just the beginning. The unwinding of a hardware bubble takes years, leaving a trail of bankruptcies and empty data centers in its wake.
