Google has placed limits on Meta’s use of its Gemini AI models after the social-media company requested more computing capacity than Google could provide. Meta is directing staff to use AI tokens more efficiently and is accelerating development of its own Muse Spark model.
Google announced that it would restrict Meta’s access to its Gemini generative-AI models following a request from the social-media firm for additional compute resources that exceeded Google’s available capacity [1]. The limitation was communicated to Meta in March 2026, and the news was reported on June 28, 2026 [1].
Meta Platforms Inc. and Google LLC, a subsidiary of Alphabet Inc., are the primary parties involved in the restriction [1]. Google indicated that it could not meet the full compute capacity Meta sought to purchase for Gemini, prompting the company to advise Meta to conserve AI token usage and to consider alternative internal models [1]. The decision also affects other Google cloud customers who rely on Gemini for high-performance AI workloads [2].
Timeline and Communication
Google’s internal capacity assessment identified a shortfall in the high-performance computing (HPC) resources required to support Meta’s projected Gemini usage [3]. In March 2026, Google notified Meta that the requested compute allocation could not be fulfilled and that the company would need to operate within reduced limits [1]. The restriction was publicly disclosed by the Financial Times on June 28, 2026, and subsequently reported by multiple technology news outlets [1][2].
The timeline reflects a broader industry trend where demand for specialized AI hardware has outpaced supply, leading cloud providers to prioritize existing commitments and manage resource allocation more tightly [3].
The decision also affects other Google cloud customers who rely on Gemini for high-performance AI workloads [2].
Scope of the Gemini Access Limits
Google Limits Meta's Access to Gemini AI Models Over Compute Capacity
Google’s limitation applies to Meta’s consumption of the Gemini 1.5 and Gemini Pro models, which are among Google’s most advanced generative-AI offerings [1]. The cap reduces the number of AI tokens Meta can process per month, effectively throttling the volume of inference requests the company can submit [2]. Google did not disclose the exact token quota but indicated that the restriction would affect ongoing development projects that rely on Gemini for content generation, recommendation algorithms, and internal tooling [1].
Uttar Pradesh cabinet approved a new startup policy and a ₹1,000‑crore fund on July 6, 2026, aiming to expand entrepreneurship support across the state.
Meta’s internal memo, as reported, instructed engineering teams to prioritize token-efficient prompts and to explore alternative model architectures, including the company’s own Muse Spark model, which is under accelerated development [2]. The memo also noted that the limitation could delay certain product rollouts that were dependent on Gemini’s capabilities [1].
Immediate Impact on Meta and Other Clients
The reduction in Gemini access is expected to delay Meta’s AI-driven initiatives that were scheduled for deployment in the second half of 2026 [1]. Teams working on automated content moderation, ad-targeting enhancements, and virtual-assistant features have been directed to adjust timelines and to reallocate workloads to internal models [2]. Meta’s shift toward Muse Spark reflects a strategic move to reduce reliance on external AI providers amid industry-wide compute shortages [4].
Other Google Cloud customers that have integrated Gemini into their products may also encounter similar capacity constraints, prompting them to evaluate token-usage policies and to consider hybrid AI strategies that combine Google’s models with in-house solutions [3].
Key Facts
What: Google limited Meta’s access to Gemini AI models due to insufficient computing capacity.
Teams working on automated content moderation, ad-targeting enhancements, and virtual-assistant features have been directed to adjust timelines and to reallocate workloads to internal models [2].
When: Notification in March 2026; public report on June 28, 2026.
Impact: Meta must use AI tokens more efficiently, shift to its Muse Spark model, and may experience project delays; other Google clients could face similar restrictions.
Google limits Meta’s use of its Gemini AI models, FT reports – CNBC – https://www.cnbc.com/2026/06/28/google-limits-metas-use-of-its-gemini-ai-models-ft-reports.html
Google limits Meta’s use of Gemini AI models, delaying AI projects – CyberNews – https://cybernews.com/ai-news/google-meta-gemini-constraints/
Google Limits Meta’s Access to Gemini AI Models Amid Computing Capacity Crunch – Creati.ai – https://creati.ai/ai-news/2026-06-28/google-limits-meta-access-gemini-ai-models-computing-capacity-crunch/
Google is rationing Gemini access to Meta because it cannot meet demand – The Next Web – https://thenextweb.com/news/google-caps-meta-gemini-compute-shortage
NewsNews
