Google Caps Meta’s Gemini AI Use as Compute Strain Hits Several Clients
Updated
Updated · Bloomberg · Jun 28
Google Caps Meta’s Gemini AI Use as Compute Strain Hits Several Clients
2 articles · Updated · Bloomberg · Jun 28
Summary
Meta’s use of Google’s Gemini models has been limited because Google could not supply the computing capacity the company wanted, the Financial Times reported.
Several Google clients have faced similar restrictions, with Meta hit hardest as AI infrastructure demand outstrips available compute.
The cap has disrupted Meta’s internal projects and prompted the company to tell staff to use AI tokens more efficiently, according to the report.
The limits date to around March and underscore a broader bottleneck in AI infrastructure even for large customers racing to expand model use.
Was Google's AI cap on Meta a resource crisis, or a strategic move to hobble a key competitor?
Is the AI boom hitting a physical wall, limited not by code but by power grids and electrical parts?
$320 Billion AI Compute Crunch: How Google, Meta, and SpaceX Are Racing to Overcome the 2026 Infrastructure Bottleneck
Overview
In March 2026, Google informed Meta that it could not meet Meta’s massive demand for Gemini AI computing capacity, leading to a usage cap. This was because Meta’s requirements were much heavier than other Google Cloud customers, making Meta the most affected by these constraints. As a result, Meta’s internal AI development slowed down, with several projects delayed. To cope, Meta teams had to use tokens more carefully to manage limited resources. This situation highlights how growing demand for AI compute is straining even the largest tech companies, forcing operational changes and revealing the challenges of scaling AI infrastructure.