Wall Street’s AI spending arms race just got bigger

The AI tab keeps getting larger

Wall Street’s latest AI prediction is basically: remember that spending boom you thought was wild? Yeah, double it. J.P. Morgan now expects the top four U.S. cloud providers to add more than $200 billion in extra data center capex in 2026, with another $210 billion-plus increase projected for 2027.

That’s not a typo. That’s the kind of number that makes your browser tab freeze for a second.

Why investors should care

This isn’t just about big-tech flexing for the sake of it. The bank is saying demand for AI infrastructure is still accelerating, which means the companies supplying chips, power, networking gear, cooling systems, and data center buildouts could keep riding the wave.

For the hyperscalers themselves — think Amazon, Microsoft, and Alphabet — the message is a little more complicated:

• More spending can mean more future AI capacity and more revenue down the line
• But near-term margins can get squeezed while the buildout bill lands on the table
• And if everyone keeps racing to outspend everyone else, the capital intensity of cloud could stay sky-high

The part that matters for your portfolio

J.P. Morgan also lifted its 2026 growth outlook for this group from 52% to 63%, which is Wall Street-speak for: the AI infrastructure arms race is not cooling off anytime soon.

That’s good news if you own the “pick and shovel” names tied to the boom. It’s a trickier story if you’re staring at the hyperscalers and asking, “How long can these companies keep writing giant checks before the market starts side-eyeing the ROI?”

Big picture: AI is still very much in its “build the factories first, count the profits later” phase — and right now, the factory bill is getting absurdly large.