SAN FRANCISCO (Realist English). Alphabet, Google’s parent company, has imposed restrictions on Meta’s use of its advanced Gemini artificial intelligence models.
According to the Financial Times, citing informed sources, the decision was made around March 2026 and was a direct consequence of Google’s inability to meet the growing demand for computing power (also known as “cortex”) that Meta had placed on it.
According to the publication, Meta was seeking significantly greater access to Gemini than Google could provide. The shortage of computing resources has already led to disruptions and delays in several internal Meta projects using AI.
Causes: Explosive Growth in Demand and Infrastructure Shortage
The restrictions come against the backdrop of colossal pressure on global AI infrastructure. According to sources, between March and August 2025, the number of queries to the Gemini API doubled. This forced Google to reconsider its strategy for allocating the most sought-after resource — computing power.
Since May 17, 2026, the company has introduced a system of strict limits for all Gemini users, which some compare to a mobile data plan. Instead of unlimited queries, there are now weekly quotas and “renewal windows.”
Even for a corporation like Google, the cost of expanding infrastructure to meet unlimited demand has proved “frightening.” The current situation clearly demonstrates that despite multi‑billion‑dollar investments in data centres and chips, the industry cannot keep up with the pace of generative AI development.
Impact on Meta: Project Delays and Search for Alternatives
Meta has suffered more than other Google clients due to its exceptionally high demand for Gemini models. The company uses AI for a wide range of tasks — from content moderation and fraud detection to internal workflows and code development.
In response to the restrictions, Meta has taken a number of steps:
- Internal economisation. Employees have been urged to use “tokens” — the unit of measurement for AI computing volume — more efficiently.
- Accelerating development of its own models. The company has stepped up its transition to its new Muse Spark model, which developers believe can already compete with Gemini in performance. This reduces dependence on an external supplier.
Google’s Response: Searching for External Resources
Experiencing enormous strain, Google is forced to look for unconventional solutions. In June 2026, the company signed a contract with Elon Musk’s aerospace corporation SpaceX to lease computing power. The deal is worth $9.2 billion per month. AI lab Anthropic had previously signed a similar agreement with SpaceX.
Google’s financial reports confirm the severity of the problem. In the first quarter of 2026, Google Cloud revenue reached $20 billion, but according to CEO Sundar Pichai, capacity shortages prevented even higher growth. Moreover, the backlog of cloud service demand almost doubled compared to the previous quarter.
Google’s decision to restrict Meta’s access to its AI models has become a clear indicator of a systemic crisis in the industry. Demand for AI computing resources is rapidly outstripping supply, forcing even market leaders to take unprecedented measures — from strict rationing of resources for their largest clients to leasing capacity from aerospace companies.