Google Rations Gemini Access to Meta as Compute Crunch Bites

Summary: Google informed Meta in March that it could not supply the full volume of Gemini computing capacity the social media giant had requested, and began enforcing quota-based limits in May — forcing Meta to slow internal AI projects and restrict how its employees consume AI tokens.

Key Points

• Google notified Meta in March 2026 it couldn't fulfill the requested Gemini compute volume
• Quota-based usage limits took effect May 17; Meta told staff to use AI tokens more efficiently
• Google Cloud posted $20B in Q1 revenue; CEO Sundar Pichai noted even stronger growth was possible with more compute
• The world's five largest hyperscalers are on track to exceed $1 trillion combined in AI infrastructure spend through 2025–2026

Why It Matters

When a company the size of Meta can't get all the AI compute it needs from Google, it illustrates how completely demand has outrun supply. Infrastructure access is now as strategically important as model quality — those who locked in data center capacity early hold a durable advantage. This also signals that cloud providers must increasingly ration capacity, which could reshape enterprise AI partnerships across the industry.

Further Reading

• Google Caps Meta's Use of Gemini AI Models — Bloomberg
• Google is rationing Gemini access to Meta — The Next Web