Why Google is Limiting Meta’s Gemini AI Use as Demand Soars

As the world becomes more and more reliant on AI, even the companies that power the revolution are starving for compute. 

Search engine giant Google is imposing strict limits on Meta’s use of its Gemini AI models, exposing severe infrastructure bottlenecks that are now disrupting internal projects for the latter. 

The move comes after the social media giant requested more computing capacity than the rival technology group could deliver in March 2026, according to three people familiar with the matter who talked to Financial Times.

Several other Google clients have been affected by the infrastructure restrictions too, albeit to a lesser extent. However, Meta has been particularly impacted because of its exceptionally high demand for Google’s models.

Infrastructure pressures block cloud services growth

Demand for AI computing is rising sharply as businesses deploy chatbots, coding assistants and AI agents across their operations. The resulting increase in inference workloads, which are the tasks required to run models after they have been trained, has emerged as one of the biggest challenges in the technology sector.

As a direct result of these demands, particularly from large corporate customers such as Meta, Google is racing to secure additional capacity. 

Earlier this month, Google signed a US$920m-a-month deal to lease computing capacity from Elon Musk’s SpaceX. Anthropic, the maker of the Claude chatbot, also signed a similar deal with SpaceX last month.

During the first-quarter earnings announcement in April, Sundar Pichai, CEO at Google, said that cloud revenue exceeded US$20bn for the first time. The backlog of signed, but not yet delivered, cloud contracts nearly doubled quarter on quarter to more than US$460bn.

Sundar explains that computing power constraints prevented even higher growth of Google Cloud: "Obviously, we are compute-constrained in the near term. And as an example, our Cloud revenue would have been higher if we were able to meet the demand.”

Shifting the strategy to ease cloud dependence

The restrictions expose the extent to which Meta has relied on rival models such as Gemini as the social platform spends aggressively to become a leader in AI and improve its own models. 

CEO Mark Zuckerberg, has recently been tapping into talent and securing infrastructure to develop what he dubs personal superintelligence, an advanced AI system that surpasses human cognitive capabilities across multiple domains. 

Unlike Google, Meta does not possess a cloud business and is racing to build out its fleet of data centres for its own training and inference needs. As part of this push, Meta has committed to investing US$600bn in the US by 2028.

Gemini has been used internally at Meta to automate some of its safety processes, such as rooting out scams and taking down harmful content, alongside customer services and advertising help chatbots. 

It is also used for internal workflows and coding alongside other models like Anthropic’s Claude. Meta initially chose Gemini because it performed better than its own Llama open-source models.

More recently, Meta has begun prioritising its new Muse Spark model, which is viewed as more competitive with Gemini. The model reduces the dependence on external infrastructure for some applications. 

Apple has also partnered with Google to use Gemini models to power its next-generation Siri voice assistant, using custom models that are different from the versions available to the public.

Earlier this year, several tech giants urged employees to use AI tools as extensively as possible in a trend referred to as tokenmaxxing. Meta even stated that it would evaluate employee performance based on usage of AI tools.

Owing to the restrictions and a broader push to streamline AI costs, Meta has now reversed this approach. The company has encouraged staff to be more efficient with AI tokens – the units that measure AI usage.

Meta is not alone in adjusting its strategy to manage escalating technology costs. In June, Amazon dropped its internal initiative that sought to encourage employees to embrace AI tools after it emerged staff were using AI to complete what the Financial Times described as ‘pointless’ tasks.