Google Restricts Meta's Gemini AI Usage Amid Demand Surge

Google restricts Meta’s access to Gemini models due to severe infrastructure shortages, disrupting major internal AI projects. Credit: Getty Images

Google limits Meta’s Gemini access amid critical computing constraints, halting internal projects as global infrastructure demand surges.

As the world grows increasingly dependent on AI, even the companies driving this revolution are struggling to secure enough computing power.

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

This move follows Meta’s request for more computing capacity than Google could provide in March 2026, according to three people familiar with the matter who spoke to the Financial Times.

Several other Google clients have also faced infrastructure restrictions, though to a lesser extent. Meta, however, has been hit hardest due to its exceptionally high demand for Google’s models.

Infrastructure Pressures Slow Cloud Services Growth

Demand for AI computing is climbing sharply as businesses deploy chatbots, coding assistants, and AI agents across their operations. The resulting surge in inference workloads—tasks required to run models after training—has become one of the tech sector’s biggest challenges.

Driven by this demand, especially from large corporate clients like Meta, Google is racing to secure additional capacity.

Earlier this month, Google signed a $920 million per month deal to lease computing capacity from Elon Musk’s SpaceX. Anthropic, the creator of the Claude chatbot, also reached a similar agreement with SpaceX last month.

During the first-quarter earnings call in April, Sundar Pichai, Google’s CEO, said cloud revenue exceeded $20 billion for the first time. The backlog of signed but undelivered cloud contracts nearly doubled quarter over quarter, surpassing $460 billion.

Pichai noted that computing power constraints prevented even faster growth for Google Cloud: “Obviously, we are compute-constrained in the near term. For example, our Cloud revenue would have been higher had we been able to meet demand.”

Sundar Pichai, CEO of Google. Credit: Getty Images

Shifting Strategy to Reduce Cloud Dependence

The restrictions highlight how heavily Meta has relied on rival models like Gemini as the social platform invests aggressively to become an AI leader and improve its own models.

CEO Mark Zuckerberg has recently been recruiting talent and securing infrastructure to develop what he calls personal superintelligence—an advanced AI system that surpasses human cognitive abilities across multiple domains.

Unlike Google, Meta does not operate a cloud business and is rushing to build out its own data centers for training and inference. As part of this push, Meta has committed to investing $600 billion in the US by 2028.

Internally, Meta has used Gemini to automate some safety processes, such as detecting scams and removing harmful content, alongside customer service and advertising chatbots.

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

More recently, Meta has begun prioritizing its new Muse Spark model, which is seen as more competitive with Gemini. This model reduces reliance on external infrastructure for certain applications.

Mark Zuckerberg, CEO of Meta, is investing heavily in infrastructure to develop highly advanced personal superintelligence.

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

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

Due to the restrictions and a broader push to reduce AI costs, Meta has now reversed this approach. The company has encouraged staff to use AI tokens—the units that measure AI usage—more efficiently.

Meta is not alone in adjusting its strategy to manage rising technology costs. In June, Amazon dropped its internal initiative encouraging employees to embrace AI tools after it was discovered that staff were using AI to complete what the Financial Times described as “pointless” tasks.