Meta seeks flexibility in data center financing

As evidenced by the hundreds of billions being spent this year on data center buildouts, hyperscalers are moving fast to bring massive amounts of computing power to bear on running AI.

Don’t just take our word for it, either.

“From our perspective, it’s strategically incredibly important that we secure access to this [compute] capacity as quickly as possible and in as large a quantity as we can get,” Michael Hart, director of corporate finance at Meta, told attendees at the 2026 Berkeley Forum on Corporate Governance on Oct. 14. “And so, that’s what is driving both a lot of the things we’re doing on the corporate balance sheet as well as some of the partnerships that we’ve struck with investment firms.”

It’s all hands on deck for finance teams at Meta, AI frontier model developers, and their partners to find the capital needed to fund these data center buildouts. The modern data centers needed to power AI technology aren’t like the data centers of old, according to Mona Dajani, another panelist who’s a partner at law firm Cooley and co-chair of its infrastructure, energy, and real estate practice. Before, data centers were “basically a real-estate transaction.” A modern data center development “integrat[es] real estate, energy…and then technology together.”

Sharing the load. Financing strategies range from corporate bond issuance to complex joint ventures, according to Hart and other experts who spoke during a panel session on digital infrastructure.

Structured as joint ventures between stakeholders, Dajani said, data center financing can include the developers, utility companies, and chip manufacturers. “It’s become much more integrated, much faster, [and] more at stake.”

One of Meta’s large projects is structured as a JV between Meta and Blue Owl Capital. They are developing a five-gigawatt data center in Richland Parish, Louisiana. Meta announced the $27 billion deal about a year ago. Blue Owl-managed funds own an 80% stake in the joint venture and Meta owns the other 20%. As part of the deal, Meta is overseeing construction and the facility’s operations when it’s done.

“Meta entered into operating lease agreements with the joint venture for use of all of the facilities of the campus once construction is complete,” according to a press release. “These lease agreements will have a four-year initial term with options to extend.”

“Meta also provided the joint venture with a residual value guarantee for the first 16 years of operations,” according to the press release—if the lease is terminated or not renewed, under certain conditions Meta would make a capped cash payment to the JV.

That type of deal “gives us a lot of long-term flexibility,” Hart said. It’s structured so that “we’re matching the timing when we incur those costs with when the data center is up and running” and generating revenue for Meta.

But such a deal doesn’t make sense for every development, he added.

“It’s a model that took many months and many dozens of people within Meta and our partner companies to bring that to life,” Hart said. “And that’s great for a $27 billion project, but it doesn’t make a ton of sense for a 50-megawatt [10 times smaller] data center.”

Meta’s other financing options include working “with cloud platforms to gain access to additional capacity in different locations or at faster speed to market.”

“Overall, that has brought together a mix of sources of capital to support the physical infrastructure that is powering these consumer experiences that we’re very proud to bring to the market,” Hart said.

The long game. Hyperscalers like Meta are taking on a load of debt to build out their digital infrastructures. That may feel like a big problem at a time of rising interest rates, with more hikes possible before year’s end.

“I feel like I probably wouldn’t be doing my job as a treasury professional if I said rates didn’t matter,” Hart said. However, current interest-rate changes don’t affect these projects much on the whole, he said. Because these are long-term investments, Meta will make “multiple trips to the market over time and take a dollar cost averaging approach” to debt structuring.

“From that perspective, at any one point in time, we want to be thoughtful about optimizing our execution, but we don’t necessarily solve for [the] short-term view on where rates have gone or where they are going,” he said.