San Francisco, June 30, 2026
According to information from the Financial Times, Google has limited Meta's access to its Gemini AI models because the requested computing capacity cannot be fully provided.
The restriction took effect in March 2026 and affects the largest internal AI user at Meta. As the Financial Times reported, citing three people familiar with the matter, Meta demanded an unusually large amount of computing power from Google—more than the provider could deliver. Other Google customers also received restrictions, albeit to a lesser extent than Meta.
According to the newspaper, Meta uses the Gemini models for a range of business-critical tasks: in customer communication via chatbots, in advertising consulting, in programming, and in internal process automation, including the detection of fraud attempts and the filtering of harmful content. Due to the throttled capacities, Meta asked its employees to use AI tokens more sparingly.