Google, owned by Alphabet, restricted Meta's access to its Gemini artificial intelligence models after Meta requested more compute capacity than Google could provide, the Financial Times reported.
According to the report, Google notified Meta in March that it was unable to fulfill the full compute capacity needed to run Gemini. The limitation delayed and disrupted several internal AI projects at Meta because the company asked employees to conserve so-called tokens, the basic units of AI usage.
The FT said other Google customers also felt the impact of tight capacity, though less severely; Meta was particularly affected because of its exceptionally high demand for Google’s models. The story highlights that, despite continued large investments in chips and data centers, technology companies still struggle to meet rapidly growing demand for AI compute.
Financial figures and consequences
Google Cloud reported $20 billion in revenue for the quarter ending in March. Sundar Pichai, CEO of Alphabet, said constrained compute capacity limited faster expansion. The capacity shortfall also contributed to Google Cloud’s order backlog nearly doubling on a quarterly basis.
Why this matters
The episode underscores how AI-driven demand is straining data center and chip resources, affecting competitors’ ability to develop new AI products and forcing changes in how companies plan capacity and manage token consumption.
(Source: Reuters; cover image is illustrative.)
