Google began limiting Meta’s usage of its Gemini AI models around March 2026 because Meta's demand for computing power surpassed what Google could provide [1, 2, 3]. The reduction disrupted and delayed some of Meta’s internal AI projects [1, 2, 3]. Other Google Cloud clients also faced similar capacity constraints, but the impact was less severe than for Meta [1, 2, 3].
CEO Sundar Pichai said Google Cloud’s Q1 2026 revenue reached about $20 billion, but computing power limitations capped growth. "Computing power constraints prevented even higher growth and contributed to the cloud unit's backlog nearly doubling quarter on quarter," he stated [1]. Despite billions invested by tech companies in chips and data centers, demand for AI computing continues to outstrip supply [1, 2, 3].
Meta does not operate its own cloud infrastructure and relies on providers like Google while planning a $600 billion investment to expand its cloud computing capacity over two years [4]. Internally, Meta has urged employees to use AI tokens more efficiently to manage limited AI usage stemming from these restrictions [1, 3]. Token prices for AI services have surged amid constrained supply and rising demand, pushing companies including Meta to optimize usage [5, 4].
Meta uses Gemini AI alongside other models such as Anthropic’s Claude for safety automation, customer service chatbots, ad support, and coding tasks [6, 7, 8, 4]. Google first notified Meta in March 2026 that it could not meet the full capacity Meta sought for Gemini AI and began setting usage limits then [1, 2, 3].
The Financial Times published a report on June 28, 2026, detailing Google’s restrictions on Meta’s Gemini AI usage [1, 2, 3]. With the cloud backlog doubling and capacity tight, competition for AI resources remains fierce.
Meta’s planned large-scale cloud computing investment over the next two years signals an effort to relieve these constraints and build more control over its AI infrastructure [4].
