Meta Weighs Leasing Excess AI Data Center Capacity as Costs Surge and AI Ambitions Grow

Meta CEO Mark Zuckerberg is grappling with a strategic dilemma: whether to monetize the company’s growing stockpile of AI data center capacity by leasing it to external firms, or to reserve the compute power for Meta’s own advancing AI models. The decision comes at a time when Meta’s capital expenditures have soared to rival those of its largest peers, yet the company still lacks a mature cloud services business to offload unused infrastructure.

During internal discussions, Zuckerberg acknowledged the pressure to generate new revenue streams beyond advertising, especially after Meta posted a weaker-than-expected revenue forecast and a 90% drop in free cash flow, driven by heavy spending on AI infrastructure. The company is now exploring the option of leasing computing power to select partners, including AI startup Anthropic, as a way to recoup some of its massive investment. “We need to build new enterprise sales capabilities,” Zuckerberg said, “while ensuring we retain ample compute for our own AI models, which are advancing rapidly under AI chief Alexandr Wang.”

The move highlights a broader tension facing large tech companies with sprawling data center footprints: how to balance the immediate financial return from selling excess capacity against the long-term need to fuel internal AI development. Meta’s AI research, led by Wang, has accelerated in recent months, requiring substantial and predictable access to GPU clusters and networking infrastructure. Leasing capacity to outside firms could provide short-term cash relief but might also limit Meta’s ability to scale its own AI projects quickly.

Industry observers note that Meta’s lack of a public cloud platform—unlike rivals Amazon, Microsoft, and Google—puts it at a disadvantage when trying to monetize spare capacity. However, the growing demand for AI compute from startups and enterprises could create a lucrative niche. If Meta successfully builds a leasing business, it could help offset the billions spent on data center construction and energy contracts, while also positioning the company as a key player in the AI infrastructure ecosystem.