In March of this year, Google significantly reduced Meta's access to its flagship AI models of the Gemini family. The reason is a shortage of computing power to fully meet Meta's requests. This decision led to the disruption and significant delays in the implementation of a number of Meta's internal projects related to artificial intelligence. As of today, the restrictions remain in place.

Reasons and consequences for Meta

Internal sources at Meta confirm that the company has faced strict limits on the use of Gemini tokens. In response, Meta's management urged employees to use resources more sparingly, which is a direct consequence of both overall cost reduction and the restrictions imposed by Google. The situation is exacerbated by the fact that Meta actively uses Gemini to train and test its own models, and disruptions in access directly affect the pace of development.

Overall market picture

Other major Google Cloud customers have also felt the imposed limits, but to a much lesser extent. Meta, as one of the largest consumers of computing resources, has been hit particularly hard. This underscores the growing dependence of even industry giants on third-party cloud providers and their infrastructure.

Analytical commentary: The situation demonstrates a fundamental problem in the modern AI sector — a shortage of computing power. Even with significant budgets, access to advanced models can be limited by the physical capabilities of data centers. For Meta, this is not just a technical inconvenience, but a strategic challenge that could push the company to develop its own hardware solutions even more actively, which in the long term will change the balance of power in the cloud computing market.