Google Caps Meta's Access to Gemini AI Over Capacity Limits

Google has imposed restrictions on how much of its Gemini AI technology Meta can access, marking a notable friction point between two tech giants both racing to scale artificial intelligence capabilities.

According to a Financial Times report, Google informed Meta around March that it could not provide the full Gemini model capacity the social media company had requested to purchase. The computing power shortfall has disrupted and delayed some of Meta's internal AI projects, though the specific initiatives affected were not disclosed.

While several other Google clients have experienced similar capacity constraints, Meta has been particularly impacted due to what the Financial Times characterized as "exceptionally high demand" for Google's models. The restrictions have prompted Meta to encourage employees to use AI tokens—the units measuring AI usage—more efficiently.

Why it matters

This episode illustrates a fundamental bottleneck in the AI economy: even as companies pour billions into chips and data centers, computing infrastructure cannot keep pace with surging demand for AI services. The constraint affects not just startups but relationships between the industry's largest players, potentially reshaping competitive dynamics and partnership strategies across the sector.

Capacity crunch hits cloud revenue

The capacity limitations extend beyond Meta. Google Cloud generated $20 billion in revenue during the first quarter ending in March, but CEO Sundar Pichai acknowledged that computing power constraints prevented even stronger growth. The cloud unit's backlog nearly doubled quarter-over-quarter, suggesting robust demand that Google's infrastructure couldn't immediately fulfill.

The situation highlights an industry-wide challenge. Despite massive capital investments in AI infrastructure, tech companies continue struggling to secure sufficient computing resources to meet customer needs. This scarcity creates strategic vulnerabilities for companies like Meta that depend on external AI model providers while building their own capabilities.

Strategic implications

The capacity restrictions may accelerate Meta's efforts to develop proprietary AI infrastructure and reduce dependence on external providers. The company has already invested heavily in its own AI research and model development, including the open-source Llama family of large language models.

For Google, the capacity constraints present both a challenge and an opportunity. While the company cannot currently meet all customer demand—a revenue limitation—the situation also demonstrates strong market appetite for its Gemini models and positions Google as a critical infrastructure provider in the AI ecosystem.

Neither Google nor Meta responded to requests for comment on the Financial Times report, which cited people familiar with the matter. Reuters could not independently verify the details.

The capacity crunch underscores how AI infrastructure has become a strategic chokepoint, with implications for competition, partnerships, and the pace of AI adoption across the technology industry.

This story was first reported by the Financial Times.