Google restricts Meta's Gemini AI usage, giants face computing bottleneck

Alphabet-owned Google has placed strict limits on Meta Platforms' usage of its advanced Gemini artificial intelligence models after the social media giant requested more computing capacity than Google could physically supply, the Financial Times reported on Sunday.

Google reportedly informed Meta around March that it could not fulfill the massive volume of Gemini capacity Meta intended to purchase. According to sources familiar with the matter, this severe infrastructure bottleneck has disrupted and delayed several of Meta's internal AI development projects, News.Az reports, citing Reuters.

While other enterprise clients have also experienced restricted access to Google's cloud computing resources, Meta has borne the brunt of the rationing due to its exceptionally high demand for the Gemini models.

Neither Google nor Meta immediately responded to requests for comment outside of normal business hours, and the Financial Times report has not been independently verified.

In response to the computing limitations, Meta management has reportedly instructed its staff to practice stricter efficiency regarding "tokens"—the fundamental units used to process and measure AI data usage.

The restriction underscores a broader, systemic challenge facing Silicon Valley. Despite tech conglomerates funneling hundreds of billions of dollars into high-end microchips and massive data centers, the industry is still struggling to manufacture enough raw computing power to satisfy the exponential demand for AI services.

The infrastructure strain is actively impacting balance sheets. While Google Cloud’s revenue surged to $20 billion in the first quarter of the year, Alphabet CEO Sundar Pichai noted that acute computing power constraints actively held back even higher growth, contributing to a near-doubling of the cloud division's project backlog quarter-over-quarter.