Google Reportedly Capped Meta's Gemini AI Access After Compute Crunch Hit Capacity

A new report by The Financial Times claims that Google has been forced to limit Meta’s access to its Gemini AI models after the Facebook parent requested more capacity than Google could deliver.

Per the report, Google informed Meta around March that it couldn’t provide the full Gemini inference capacity the company wanted to purchase.

The report, citing people familiar with the matter, claims the shortage ended up delaying several internal AI initiatives at Meta while forcing the company to rethink how it consumes AI resources.

The reported restrictions apparently extend beyond Meta, although the social media giant is said to have been affected the most because of its unusually large demand for Gemini models.

The report further adds that Meta has since encouraged employees to use AI tokens more efficiently as part of a broader effort to reduce compute consumption while the limitations remain in place.

Despite spending tens of billions of dollars on new AI chips and hyperscale datacenters, major cloud providers are still struggling to satisfy enterprise demand.

Speaking of AI infrastructure, this is yet another reminder that the current bottleneck isn’t necessarily about building better models anymore. It’s increasingly about securing enough GPUs, memory, power, and datacenter capacity to actually run them.