Google Caps Meta's Use of Gemini AI Models, FT Reports

What happened

The Financial Times reported on June 28 that Google told Meta around March it could not supply the full Gemini capacity Meta wanted, disrupting some of Meta's internal AI projects.

Context and impact

The story shows AI buildouts trailing consumption — Google is paradoxically paying SpaceX $920M/month for 110,000 Nvidia GPUs as "bridge capacity" for Gemini Enterprise. Meta is shifting workloads to its internal Muse Spark model to reduce dependence on outside vendors.

Details

• Cap communicated to Meta around March 2026
• Meta employees consumed 73.7 trillion tokens in ~30 days
• Internal leaderboard named "Claudeonomics"
• CTO Andrew Bosworth sent memo pushing back on "tokenmaxxing"
• Google renting 110,000 GPUs from SpaceX at $920M/month