Particle.news

Google Limits Meta's Access to Gemini Over Cloud Compute Shortage

The company has imposed compute-based caps to ration scarce GPU capacity as it pursues leased capacity and faster data-center buildout.

Overview

  • Google told Meta in March that it could not supply the full Gemini capacity Meta sought, a shortfall that has delayed some of Meta’s internal AI projects.
  • Starting in mid-May Google introduced documented compute-based usage limits for Gemini Apps, including rolling refresh windows and weekly caps to control inference demand.
  • Meta has urged employees to curb AI token use, the internal units that measure AI consumption, as the company shifts some workloads to its own models and prioritizes key programs.
  • Other Google Cloud customers have seen reduced Gemini access but not at the scale of Meta’s disruption, underscoring that the limits are a broad demand-management response.
  • Google is seeking extra capacity through leases and partnerships, with media reports of large deals to speed supply, while the shortage is driving interest in decentralized and in‑house compute alternatives.