In a sign of mounting pressure on AI infrastructure, Google has reportedly restricted Meta's access to its Gemini artificial intelligence models after the Facebook parent requested more computing capacity than the company could provide. According to a Financial Times report, Google informed Meta around March that infrastructure constraints prevented it from fulfilling all of the Gemini AI capacity the company wanted to purchase.
Around March 2026, Google informed Meta it could not fully meet the company's requested compute quota for the Gemini model. These restrictions disrupted and delayed the timelines of multiple internal AI projects at Meta, which subsequently instructed employees to use AI tokens more sparingly and improve usage efficiency.
Meta had initially relied on Gemini, which proved more capable than its own open-source Llama models for certain tasks, to automate safety processes such as removing harmful content and eliminating scams. The company has since been shifting those workloads to Muse Spark, a new internal model, as it moves to reduce dependence on external AI providers.
The restrictions come as AI companies struggle to secure enough computing power despite spending billions on chips and data centres. According to the report, Meta was among Google's largest Gemini customers and was hit particularly hard because of its massive demand for AI resources.
As companies deploy more AI tools, computing costs are also rising, prompting some tech firms to limit usage. Amid this computing shortage, to ease the crunch, Google has reportedly signed a deal worth about $920 million a month to lease additional capacity from SpaceX, while Anthropic has entered into a similar arrangement.
Google's Own Constraints
Google's inability to serve Meta's demand is not a matter of pricing or commercial disagreement, it is a direct consequence of infrastructure running at its limits. Google CEO Sundar Pichai acknowledged the crunch in the company's Q1 results: "Obviously, we are compute-constrained in the near term. Our Cloud revenue would have been higher if we were able to meet the demand."
Google Cloud generated more than $20 billion in quarterly revenue, up 63% year-on-year, but still faces a nearly $460 billion backlog of unmet demand. The company plans to spend $180–190 billion on infrastructure in 2026 and is leasing additional capacity from SpaceX and xAI. It has also introduced compute-based usage limits for Gemini applications, replacing unlimited access with weekly quotas.
Meta has cut 8,000 jobs and is investing up to $135 billion in AI infrastructure, with the Gemini restrictions accelerating its shift toward in-house models. The episode also highlights a broader industry challenge: demand for AI computing power is outpacing available infrastructure, forcing companies to scramble for additional capacity.