Google has reportedly placed limits on how much Meta can use its Gemini artificial intelligence models after the social media giant requested more computing capacity than Google could provide, according to a report published Sunday by the Financial Times. Citing three unnamed sources, the newspaper reported that Google informed Meta around March that it could not supply the full amount of Gemini computing power the company wanted to purchase.

Neither Google nor Meta confirmed the report, and both companies declined to comment. As a result, the account remains based on unnamed sources and has not been independently confirmed by either company.

One of the most notable aspects of the report is the relationship between the two companies. Meta, which owns Facebook, Instagram and WhatsApp, is one of Google’s biggest competitors in artificial intelligence and has invested heavily in developing its own family of AI models. Yet the report says Meta has also been purchasing access to Google’s Gemini models, and its demand grew so large that it was reportedly affected more than any other Google customer when computing resources became constrained.

According to the report, the shortage delayed several of Meta’s internal AI projects and prompted company management to encourage employees to use fewer AI “tokens,” the units that measure how much computing work an artificial intelligence model performs each time it processes a request.

The report highlights one of the biggest challenges facing the AI industry today: computing power.

Large language models such as Gemini require enormous numbers of advanced computer chips operating inside massive data centers. Companies like Google make these models available to outside businesses through cloud services, charging customers based on usage. Every prompt, response and calculation consumes processing capacity.

When demand exceeds available computing resources, providers must either expand infrastructure or limit customer access until additional capacity comes online.

The reported restrictions underscore that even the world’s largest technology companies continue struggling to secure enough AI infrastructure. According to the Financial Times, several additional Google customers also experienced reduced access, although none as significantly as Meta.

The situation reflects a broader industry-wide shortage. Technology companies are collectively investing tens of billions of dollars in new data centers, advanced processors and AI infrastructure, yet demand continues to outpace supply.

For Meta, the reported limits illustrate the risks of relying on a direct competitor for critical technology. Although the company continues investing aggressively in its own AI systems and expanding its own computing infrastructure, the report suggests Meta still depended heavily enough on Google’s models that any reduction in access could slow product development.

For Google, the situation presents both an opportunity and a challenge. Selling access to Gemini has become an increasingly important business for parent company Alphabet, and attracting customers as large as Meta demonstrates strong market demand for its AI models.

At the same time, limiting a major customer’s usage illustrates that Google itself remains constrained by the pace at which it can build additional data centers and deploy new computing hardware. The company may also be prioritizing scarce computing resources for its own products and services before allocating additional capacity to outside customers.

Because the report relies entirely on unnamed sources, many details—including the precise timing and scale of the restrictions—should be viewed with caution. Reuters, which also reported on the story, said it could not independently verify the Financial Times account.

Regardless of whether the specific claims are ultimately confirmed, the broader issue is widely recognized across the technology industry. Access to high-performance AI computing has become one of the most valuable and limited resources in modern technology.

For businesses building products around artificial intelligence, securing reliable computing capacity may increasingly become just as important as choosing which AI model to use.

JBizNews Desk | New York
© JBizNews.com All Rights Reserved. Reproduction or distribution without written permission is prohibited.