Meta is quietly exploring a deal to lease some of its massive data centers to Anthropic, the AI startup behind the Claude chatbot, according to a report from The New York Times. The talks are still early, but a potential agreement could run as high as $10 billion over two years. The move would mark a major departure for the Facebook and Instagram parent, which has long relied on digital advertising for nearly all its revenue.

The discussions follow earlier indications from Meta that it was considering selling access to its computing infrastructure. During an earnings call last year, CEO Mark Zuckerberg said the company fields requests for data center space almost every week and has called it an option worth pursuing. Meta has been pouring tens of billions into AI data centers to support its own model-building efforts—it expects to spend between $125 billion and $145 billion in 2026 alone. Leasing that capacity to a compute-hungry partner like Anthropic could turn a massive cost center into a new revenue stream.

While Meta and Anthropic compete in the AI race, the arrangement would benefit both sides. Anthropic has an insatiable need for computing power, and it has already struck a similar deal with SpaceXAI worth $45 billion over three years, announced ahead of that company's IPO. After that deal closed, Anthropic immediately raised rate limits for subscribers of its Claude Code product. For Meta, the partnership would be a first step into cloud services, a business that could help justify its enormous infrastructure spending.

Source: Engadget