Meta reportedly in discussions with CleanSpark for data-center capacity in Georgia

Meta Platforms is in discussions with data center developer CleanSpark about securing space inside CleanSpark’s facilities, according to market reporting citing infrastructure research. The talks center on a potential lease arrangement that would give Meta additional data center capacity, a move that fits the broader pattern of major AI and cloud customers locking in power-hungry compute well before it is needed.

TheFly, as relayed by multiple market aggregators, said Meta is discussing a data center facility lease with CleanSpark. Another report attributed to Green Street News Infrastructure adds more specificity, saying Meta is seeking to lease space at CleanSpark’s Sandersville, Georgia facility, which has an estimated capacity of roughly 250 megawatts. In data-center terms, megawatts describe the scale of electrical power available for running servers, networking equipment, and cooling systems, which is often the limiting factor for new AI buildouts.

CleanSpark is the operator and developer of data center assets, though the exact contractual structure being discussed in these reports was not detailed. The reporting stops short of saying whether any agreement is already in place or whether the talks are exploratory, with pricing, timelines, and the amount of power Meta would ultimately contract for still unclear.

Meta, for its part, has not publicly confirmed that it is pursuing a lease with CleanSpark. The company also did not disclose, in the materials referenced by the reports, which specific AI systems or customer-facing services the additional capacity would support, nor did it state whether the lease would be for new build capacity at the site or for space within already energized infrastructure.

For CleanSpark, the prospect of a hyperscale or enterprise tenant leasing substantial capacity could be strategically meaningful. Large, long-duration commitments are often used by data center builders to de-risk construction and financing plans, especially when power availability and grid capacity take time to secure. CleanSpark’s ability to attract major computing customers would also announcement demand resilience in a sector that has been balancing fast growth with intense scrutiny on energy costs and utilization.

More broadly, this kind of reported lease conversation underscores how AI is pulling on the physical infrastructure beneath digital services. Training and inference workloads, and related storage and networking requirements, translate into high and sustained demand for rack space, dedicated power, and cooling. For companies like Meta, securing capacity can reduce risk that hardware availability or power constraints slow down product development, model iteration, or scaling decisions.

Still, investors and industry watchers will likely want more than headlines before drawing conclusions. The reports do not provide deal size, pricing, expected commencement date, or whether Meta would contract for specific capacity in megawatts versus a broader footprint arrangement. They also do not clarify whether the facility would be used for Meta’s internal workloads, for services offered through its advertising ecosystem, or for other compute-intensive functions.