Meta is reportedly in advanced discussions with several leading private credit firms to raise up to $29 billion to finance its large-scale AI data centre buildout across the United States.
According to sources, Meta is working with Apollo Global Management, KKR, Brookfield, Carlyle, and Pimco on a potential deal that would include $3 billion in equity and up to $26 billion in debt. If completed, the transaction would mark one of the largest private credit financings on record.
Morgan Stanley advises Meta and is exploring innovative structures to make the debt component more tradable and attractive to institutional investors. The deal may feature a leaseback model — investors would fund and own the data centres, leasing them back to Meta. This structure helps reduce upfront capital expenditure and mirrors strategies used by Amazon and other hyperscalers.
The debt is expected to carry terms such as SOFR plus 375–425 basis points, with a 7–10 year tenor, and could include delayed-draw tranches for added flexibility.
Why private credit and why now?
Despite Meta’s strong balance sheet (nearly $70 billion in cash and only $28.8 billion in existing debt), the company is turning to private credit for speed, flexibility, and to keep leverage off its public bond stack.
Private credit offers bespoke covenants and less public scrutiny compared to traditional public debt markets, a growing trend as U.S. data centre financing is projected to hit $60 billion in 2025.
Funding shift to support AI data centre infrastructure race
The financing effort is part of Meta’s broader push to strengthen its position in the rapidly intensifying AI infrastructure race. In May 2025, the company raised its capital expenditure guidance for the year to between $64 billion and $72 billion, up from an earlier range of $60 billion to $65 billion. The increase reflects growing investments in AI infrastructure, including next-generation AI data centres, chips, and renewable energy.
Meta CEO Mark Zuckerberg has previously said the company could spend “hundreds of billions of dollars” on AI infrastructure in the coming years. Alongside building out physical infrastructure, Meta has made major investments in AI firms such as ScaleAI, reportedly investing $15 billion, and is actively recruiting talent from competitors like OpenAI.
From hyperscaler to AI data centre infrastructure owner
Meta currently operates 28 data centre campuses globally, with more in development. The new facilities are expected to require gigawatt-scale power, some approaching the output of a nuclear plant, and are being planned in states such as Louisiana, Texas, and Wyoming. The company aims to house more than 1.3 million AI processors by 2026.
In late 2022, the company paused more than a dozen data centre construction projects to redesign them for AI-intensive workloads. Since then, it has resumed development with a renewed focus on high-performance computing and energy efficiency.
Meta has also signed its first nuclear energy agreement to support its AI operations and continues to invest heavily in renewables and subsea fibre infrastructure.
Private credit powers AI infrastructure
While Meta has historically leased data centre space, the shift toward AI-centric compute is prompting greater ownership and control of its infrastructure. In late 2022, the company paused over a dozen data centre construction projects to redesign them for AI-intensive workloads, resuming development focusing on high-performance computing and energy efficiency.
Private credit is rapidly emerging as a key financing source for digital infrastructure. In early 2025, Blue Owl committed to a $15 billion joint venture with OpenAI to fund AI data centre sites. SoftBank and Oracle are reportedly involved in a $500 billion AI infrastructure initiative. Microsoft, meanwhile, plans $80 billion in capex for 2025.
If finalised, Meta’s $29 billion transaction could result in an estimated annual interest burden of $2.3 billion, raising questions about long-term debt sustainability and the risk of overcapacity by 2027. Analysts note that while Meta’s AI ambitions are vast, monetisation of these investments is still lagging behind the pace of capital outlays.
Talks are advanced, but no binding agreement has been reached, and key terms could still change.
