Amazon and Google are winning the AI capex race — but what’s the prize?

Generally, it might seem to be the AI business is racing to see who can spend probably the most cash on knowledge facilities. Whoever builds probably the most knowledge facilities could have probably the most compute, the considering goes, and thus be capable of construct the very best AI merchandise, which can assure victory within the years to return. There are limits to this mind-set — historically, companies ultimately succeed by making extra cash and spending much less — however it’s confirmed remarkably persuasive for big tech corporations.

If that’s the sport, Amazon does appear to be successful.

The corporate introduced in its earnings on Thursday that it initiatives $200 billion in capital expenditures all through 2026, throughout “AI, chips, robotics, and low earth orbit satellites.” That’s up from the $131.8 billion in capex in 2025. It’s tempting to attribute the entire capex finances to AI. However in contrast to most of its rivals, Amazon has a big bodily plant, a few of which is being transformed to be used by costly robots, so the non-AI bills aren’t really easy to wave away.

Google is shut behind. In its earnings on Wednesday, the corporate projected between $175 billion and $185 billion in capital expenditures for 2026, up from $91.4 billion the earlier 12 months. It’s considerably greater than the corporate spent on fastened belongings final 12 months, and considerably greater than most of its rivals are spending.

Meta, which reported final week, projected $115 billion to $135 billion in capex spending for 2026, whereas Oracle (as soon as the poster baby for AI infrastructure) initiatives a measly $50 billion. Microsoft doesn’t have an official projection for 2026 but, however the latest quarterly determine was $37.5 billion, which pencils out to roughly $150 billion, assuming it retains up. It’s a notable enhance, and one which has led to investor strain on CEO Satya Nadella — however it nonetheless places the corporate in third place.

From throughout the tech world, the logic right here is easy. The revolutionary potential of AI goes to show high-end compute into the scarce useful resource of the longer term, and solely corporations that management their very own provide will survive. However whereas Google, Amazon, Microsoft, Meta, Oracle, and others are frantically prepping for the compute desert of the longer term, their traders aren’t satisfied. Every firm noticed its inventory value plummet as traders balked on the a whole bunch of billions of {dollars} being dedicated, and firms with increased spends tended to drop extra.

Crucially, this isn’t only a drawback for corporations like Meta that haven’t discovered their AI product technique but. It’s everybody — even corporations like Microsoft and Amazon with a sturdy cloud enterprise and an easy tackle tips on how to generate income within the AI period. The numbers are just too excessive for investor consolation.

Investor sentiment isn’t every thing — and on this case, it could not do a lot to vary the business’s thoughts. When you imagine AI is about to vary every thing (and the argument is fairly compelling at this level), you’d be a idiot to vary course simply because Wall Road acquired jumpy. However going ahead, Huge Tech corporations might be underneath numerous strain to downplay how costly their AI ambitions actually are.