AI spending is eating the economy's lunch

The AI money hose is still wide open

Morgan Stanley turned the dial up again on its AI infrastructure forecast, saying Amazon, Alphabet, Meta, Microsoft, and Oracle could spend about $805 billion on capex in 2026, with 2027 now penciled in at roughly $1.1 trillion. That’s a bigger number than most people use for “serious” and “absurd” combined.

David Sacks: this is not a side quest

Former White House AI and crypto czar David Sacks jumped on X and argued AI-related spending could add about 2.5% to U.S. GDP growth this year and more than 3% next year. His bigger point: the market keeps treating AI like a product cycle, but the capital spending alone is starting to look like an economic stimulus package with server racks.

Why investors should actually care

This matters for a few reasons:

• Big Tech’s AI arms race is now a real line item, not a slide-deck fantasy.
• More capex usually means more demand for chips, networking gear, data center buildouts, and power infrastructure.
• If AI spending keeps showing up in GDP and earnings, the “is this bubble-ish?” debate gets a lot more complicated.

The catch: this may be conservative

Sacks argued Morgan Stanley’s estimate could still undercount the full AI wave because it only covers five hyperscalers and leaves out startups, enterprise spending, and whatever else companies are quietly throwing at the AI bonfire. In other words, the tab may be even bigger than the headline suggests.

Big picture: when one industry starts looking like a macroeconomic growth engine, you’re no longer just trading a trend. You’re trading a new piece of the economy.