Google has limited Meta Platforms’ access to its Gemini artificial‑intelligence models after Meta’s request for additional compute capacity exceeded what Google could supply. The restriction was communicated to Meta in March and remains in effect, causing delays to several of Meta’s internal AI initiatives. Meta has responded by urging employees to use AI resources more efficiently and by moving some workloads to its proprietary Muse Spark model, thereby reducing reliance on third‑party models.
The capacity shortfall reflects broader infrastructure constraints in the AI industry, where demand for compute power outpaces the supply of chips, data‑center space and electricity despite large capital investments. To address its own shortfall, Google announced in early June that it will lease additional compute capacity from SpaceX in a deal valued at roughly $920 million per month.
During Google’s first‑quarter earnings call in April, Chief Executive Sundar Pichai said Google Cloud revenue surpassed $20 billion for the first time and that the backlog of signed but undelivered cloud contracts had nearly doubled from the prior quarter to exceed $460 billion. Pichai added that near‑term computing capacity remains constrained and that cloud revenue would have been higher had the company been able to meet all customer demand.
Meta, under CEO Mark Zuckerberg, has committed up to $600 billion in U.S. AI infrastructure investment through 2028 to expand its data‑center capacity. Internally, Meta has employed Google’s Gemini models for coding, customer‑service automation, advertising tools and content‑moderation, but is now shifting certain functions to its own Muse Spark model.
Other Google customers have also experienced similar capacity limits, though the impact has been most pronounced for Meta because of its unusually high demand.