Google caps Meta access to Gemini AI models amid compute capacity crunch delaying Meta projects
Meta $META leaned on Google's $GOOGL Gemini to help automate content moderation and scam removal. Google has now capped that access after Meta sought more capacity than it could supply, delaying some Meta AI projects. Meta is shifting work to its in-house Muse Spark model, per th
A compute shortage is now reaching into how Meta $META polices its own platforms. According to the Financial Times, Google $GOOGL has limited Meta's use of its Gemini AI models after Meta sought more capacity than Google could provide, telling the company around March that it could not meet the full demand. The shortfall disrupted and delayed some of Meta's internal AI projects, including work that used Gemini to help automate safety processes such as content moderation and scam removal. In response, Meta has directed employees to optimize how they spend AI tokens and has shifted workloads to its in-house Muse Spark model. Several other Google clients have been affected as well, though to a lesser extent. The episode shows that even a company spending heavily on its own AI infrastructure still leaned on a direct rival for parts of its safety stack, and that access to compute, not just the ability to pay for it, is shaping what the largest platforms can build right now.