Google Limits AI Token Usage for Meta Amid Computing Power Constraints

Google has restricted the usage of its Gemini AI model for its competitor, Meta.

The move comes as Google intervened after Meta requested computing capacity that exceeded what Google could provide.

The Financial Times reported that in March, Google notified Meta that it could not fulfill the entire volume of Gemini capacity that Meta sought to purchase.

The Financial Times noted that while other corporate clients of Google were also affected, Meta was hit particularly hard due to its exceptionally high demand for Google's models.

Meta is known to heavily utilize Gemini for key internal tasks, such as automating security processes and blocking harmful content, as well as for internal development work. It is reported that some of Meta's internal AI projects have faced setbacks due to Google's recent measure.

The Financial Times further added that in response to these constraints and to control AI costs, Meta is encouraging its employees to use AI tokens—the unit of AI usage—more efficiently.

This is being cited as an example of how companies, despite pouring money into AI chips and data centers, are still struggling to secure the computing power necessary to handle the skyrocketing demand for AI services.

For instance, while Google Cloud's revenue reached 20 billion dollars in the first quarter, CEO Sundar Pichai stated, "Constraints in computing power prevented higher growth, and as a result, the cloud division's backlog nearly doubled compared to the previous quarter."

Reported by Kim Taewon | Video by Lee Eui-seon | Graphics by Lee Jung | Produced by SBS Digital News
※ Please note: This article was translated by AI and may contain errors.