MUMBAI: In the AI race, it’s no longer just about smarter models, it’s about who has the biggest engine under the bonnet. Meta and Anthropic are reportedly exploring a partnership that could reshape the economics of artificial intelligence, with the companies holding early-stage discussions over leasing AI computing capacity.
According to a person familiar with the matter who spoke to CNBC, Anthropic is in preliminary talks with Meta to access the social media giant’s AI infrastructure. The discussions gained further attention after The New York Times reported that the proposed arrangement could be worth around $10 billion, prompting Meta’s shares to recover from earlier session lows on Friday.
Meta declined to comment on the reported negotiations.
The discussions come only weeks after Anthropic struck a separate agreement with Elon Musk’s SpaceX to tap computing resources at the Colossus 1 data centre, a move aimed at expanding capacity for users of its paid AI services.
For Anthropic, the latest talks underscore a broader challenge confronting AI developers: securing enough high-performance computing power to train and operate increasingly sophisticated models. Access to Nvidia’s AI chips remains constrained across the industry, forcing companies to seek partnerships rather than rely solely on their own infrastructure.
The shortage has already affected Anthropic’s services. The company has introduced usage limits on some of its advanced AI models, including Fable, as demand continues to outpace available computing resources.
For Meta, the potential partnership reflects an evolving AI strategy that extends beyond developing models and consumer products. The company is increasingly looking at ways to monetise the vast AI infrastructure it has built.
In May, Meta CEO Mark Zuckerberg said the company was evaluating opportunities in the cloud computing market as it seeks new revenue streams from its growing investments in artificial intelligence.
Meta is expected to spend as much as $145 billion on capital expenditure in 2026, with a significant portion earmarked for AI infrastructure, data centres and computing capacity.
The company is also strengthening its leadership bench. Separately, CNBC confirmed that Dave Brown, a longtime senior executive at Amazon Web Services (AWS), is set to join Meta, adding further cloud computing expertise as the company expands its AI ambitions.
Zuckerberg had hinted at growing demand for Meta’s infrastructure as early as October last year, revealing that companies had approached Meta seeking access to its computing capacity, with some even willing to pay a premium to secure scarce AI resources.
The reported negotiations highlight a broader shift in the AI industry, where access to computing power is emerging as one of the sector’s most valuable assets. As demand for advanced AI continues to surge, the companies with the largest data centres and deepest reserves of AI chips are increasingly positioned not just as technology developers, but as infrastructure providers powering the next generation of artificial intelligence.