Google Rates Gemini AI a Meta, Backlog en la nube casi se duplica a $ 20B

Google told Meta in March that the company couldn't have all the Gemini compute capacity it wanted—and that shortfall directly delayed Meta's internal AI projects. The Financial Times broke the story, and the numbers explain why: Google Cloud’s backlog nearly doubled quarter over quarter while revenue hit $20 billion in Q1. CEO Sundar Pichai said capacity constraints prevented even higher growth.

Capacity Crunch Hits the Biggest Customer

Meta wasn't just any Gemini buyer—the social media giant sought exceptionally high capacity, according to FT’s sources. Google said no around March, and the disruption wasn't minor: several internal AI projects at Meta were delayed. Other Google cloud clients also felt the squeeze, but none as hard as Meta.

Token Rationing Becomes Meta's New Reality

With the spigot turned down, Meta instructed its teams to use AI tokens more efficiently. That’s a polite way of saying each project now fights for a smaller slice of a limited pie. Tokens are the unit of measurement for Gemini usage; when you can't buy more, you optimize or stall.

Google's Cloud Growth Hits a Ceiling

Google Cloud’s $20 billion quarter sounds like a victory lap, but the near-doubling of its backlog tells a different story: demand is outstripping supply across the board. Google is spending billions on chips and data centers, but so is everyone else. The capacity wall Meta hit is a preview of what other enterprises will face as AI compute demand keeps climbing.

Google’s choice to ration a rival rather than expand faster might make short-term business sense, but it puts a hard ceiling on how much AI work the industry can actually ship. The question is whether Google can scale capacity faster than demand from both rivals and customers—or whether more companies will hit the same wall Meta just ran into.

Source: Google limits Meta's use of its Gemini AI models
Domain: cnbc.com