Google imposes compute caps on Gemini API amid surging demand
In brief
- Google imposed compute-based usage limits on Gemini API May 17, 2026, citing surging demand.
- API requests more than doubled between March and August 2025, straining capacity.
- Meta was exploring Gemini with Google Cloud for ad targeting before caps took effect.
- Usage limits apply to all customers via rolling 5-hour windows and weekly caps.
- No evidence shows Google designed restrictions specifically targeting Meta.
The demand surge
Gemini API requests more than doubled between March and August 2025, creating real scarcity in Google's compute pipeline. Google reported continued sales growth for its Gemini AI products into 2026, yet capacity constraints forced the company's hand. The new limits apply broadly to all customers, not a targeted move against any single firm.
Google has documented rate limits and spending tiers designed to ensure fair API usage across all customers. These guardrails are now the mechanism through which Google rations access during peak demand. It's a pragmatic solution, though one that complicates the calculus for companies betting on Gemini as a core component of their infrastructure.
The Meta angle
The timing reveals something stranger. Meta was in discussions with Google Cloud about leveraging Gemini models for its advertising business, and in September 2025, the company was actively discussing how to integrate Google's Gemini AI to improve ad targeting. Yet there's no public evidence that Google designed bespoke restrictions targeting Meta specifically.
This illustrates a peculiar dynamic in Silicon Valley. Bitter rivals are quietly becoming each other's customers in the AI arms race. Meta has its own open-source Llama models, yet still explored Gemini for ad optimization. Google caps access broadly, catching Meta in the same net as everyone else.
The situation highlights how deeply interdependent the tech industry has become. Companies that compete fiercely in consumer products now rely on each other's AI infrastructure, creating a tension between commercial rivalry and mutual dependence. Capacity constraints force that tension into the open.
