Meta teases a data center business as CEO looks for ways to manage AI infrastructure costs

Meta is indicating that it may rethink how it builds and pays for the computing power behind its artificial intelligence push. In a report published Tuesday, Yahoo Finance said CEO Mark Zuckerberg is teasing the launch of a data center business that could help address Meta’s AI spending needs.

The article frames the idea as a response to the rapid buildout of infrastructure required for modern AI models, which tend to require large pools of chips and power-hungry data center capacity. The report does not provide specific financial figures or a timetable for when any new data center offering could begin operating at scale.

Zuckerberg’s remarks, as described by the outlet, suggest Meta is considering a structure where data centers would serve not only internal operations but also a broader external business. That matters because data centers represent one of the largest categories of capital spending for AI-focused technology companies, and long-lead construction cycles can make costs hard to control.

Meta’s approach, if it moves from teaser to implementation, would place the company closer to the posture of an infrastructure supplier. Cloud operators, colocation providers, and enterprise infrastructure vendors have all built business models around selling compute capacity, often under longer-term contracts and with clear capacity commitments. Meta would face the challenge of setting pricing and service guarantees in a market that has been shifting quickly with changes in AI model requirements.

Meta has previously emphasized AI as a core driver of product improvement across its platforms, including recommendations, ads ranking, and other systems that rely on machine learning. However, Tuesday’s report centers less on product outcomes and more on the cost and supply constraints of compute capacity, which are increasingly central to AI strategy.

For now, the most concrete information in the public reporting is the direction of travel rather than the blueprint. The report does not disclose what portion of Meta’s planned compute would be made available externally, whether third-party customers would be included immediately, or whether Meta would build new facilities, upgrade existing sites, or use a hybrid approach.

It is also unclear what legal and operational structure would underpin any data center business, including where it would sit within Meta’s organization and how it would manage staffing, power procurement, and ongoing maintenance. Data center operations can be capital-intensive, and the lack of disclosed economics makes it difficult to assess how the plan could affect Meta’s future spending profile.

Investors and customers will likely watch for follow-up indicates, such as explicit references to external customers, contract terms, and the scale of any new capacity. Meta’s next earnings materials and infrastructure update statements, alongside any official company communications, may also clarify whether the initiative is intended to operate like a standalone unit or simply to support internal AI efficiency.