The Financial Times reported on June 28, 2026 that Google (Alphabet) has placed limits on Meta's use of its Gemini AI models, citing Google's inability to provide as much computing capacity as Meta wanted.
June 28, 2026 · AI Infrastructure
Google Caps Meta's Gemini Access as the Compute Crunch Bites
Unable to supply as much computing capacity as Meta wanted, Google limited its rival's use of Gemini models — disrupting some of Meta's internal AI projects and exposing how even a hyperscaler can't keep pace with surging demand.
~$20B
Google Cloud revenue, Jan–Mar 2026 quarter
~2×
Order backlog nearly doubled vs. the prior quarter
Mar 2026
When Google told Meta of the access limits
Demand outruns supply
The order backlog nearly doubled quarter-on-quarter while capacity stayed constrained — so requests had to be capped.
~2×
Latest quarter backlog
How rivals got tangled in supply & demand
Meta scales up
Taps Gemini for internal projects (e.g. ad targeting), driving heavy token use
→
Capacity runs short
Global GPU server fleets for training & inference can't meet demand
→
Google caps access
Delays & interruptions hit Meta; staff told to use tokens efficiently
A symbolic case of the compute crunch
Supply still can't meet demand despite massive AI infrastructure investment.
Multiple Google customers were affected — Meta's demand was simply the highest.
Google strengthened Gemini API cost-control features around March 2026 to tighten enterprise governance.
Specific model versions, pricing and benchmark details tied to the limits were not disclosed.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…