Google is tightening access to its Gemini AI platform. Demand from developers and tech giants has surged, with API requests more than doubling between March and August 2025.
Starting May 17, 2026, Google is imposing compute-based usage limits on Gemini Apps. Users now operate under rolling 5-hour refresh windows and weekly caps, similar to a cellular data plan. These documented rate limits and spending tiers apply broadly to all customers.
Among those affected is Meta. The advertising giant has been in discussions with Google Cloud about using Gemini models to improve ad targeting. This highlights a new dynamic in Silicon Valley, where fierce competitors are becoming customers in the AI race.
The demand surge and subsequent capping underscore the steep adoption curves for frontier AI models. Even for Google, the capital required to meet unconstrained AI demand is significant enough to warrant management strategies.
Meta's exploration of Gemini, despite having its own open-source Llama models, suggests Google's offerings provide advantages for specific use cases.