The great AI spending spree has entered its reckoning phase. Amazon's latest earnings call, in which the company disclosed yet another quarter of ballooning capital expenditures on artificial intelligence infrastructure, sent its shares tumbling and crystallized a fear that has been building across markets for weeks: Big Tech may be building cathedrals for a congregation that hasn't materialized.
The Nasdaq 100 has shed roughly ten percent since early October, erasing hundreds of billions in market capitalization from the very companies that were supposed to be the unassailable winners of the AI revolution. The proximate cause is straightforward — quarterly reports revealing that Amazon, Microsoft, Alphabet, and Meta are spending at rates that would have seemed hallucinatory even eighteen months ago, while revenue from AI products remains stubbornly incremental.
The arithmetic of faith
The numbers tell a story of industrial-scale conviction. Combined AI-related capital expenditure among the four largest cloud and consumer AI players is on pace to exceed $200 billion for the full year. Data centers are sprouting across the American interior at a pace not seen since the fiber-optic mania of the late 1990s. Electricity demand from these facilities has become a genuine constraint, with utilities scrambling to provision power for facilities that consume as much energy as small cities.
Yet the revenue side remains conspicuously modest. Microsoft's AI-augmented Office suite adds a few dollars per seat per month. Google's AI search features have yet to demonstrate meaningful monetization. Amazon Web Services is growing, but not at rates that justify the infrastructure buildout. The gap between investment and return has become impossible to ignore.
A different kind of bubble
This is not, strictly speaking, a repeat of the dot-com collapse. The companies spending most aggressively are genuinely profitable, with core businesses that throw off enormous cash. They can sustain losses on AI ventures for years without existential risk. The question is whether shareholders will tolerate that patience.
The answer, increasingly, appears to be no. Fund managers who spent 2023 and 2024 celebrating the "Magnificent Seven" are now rotating toward value stocks, energy, and anything that offers near-term earnings visibility. The AI trade has become crowded, consensus, and suddenly uncomfortable.
Our take
The market is not wrong to demand accountability, but it may be premature in its timing. Transformative technologies rarely monetize on investor-friendly schedules. The internet itself took a decade to produce sustainable business models after the infrastructure was laid. What's different now is the scale of the bet — and the concentration of it among a handful of companies whose share prices had already priced in perfection. The correction is healthy. The question is whether it becomes contagious.
