Meta is raising $29B to build AI data centers across America. They’re not using public debt or their $70B cash pile... Meta is engineering one of the largest private credit deals in history. $26B in debt and $3B in equity from firms like Apollo, Brookfield, KKR, Carlyle AlpInvest, and PIMCO. Instead of funding directly, they’re using a leaseback model: Investors build and own the data centers, and Meta leases them back. Why? Because #AI infrastructure is too big and too strategic for the old playbook. Meta wants speed. Flexibility. And no balance sheet drag. So they're treating AI infrastructure like a utility and financing it like one. What it funds: • 2+ GW of new data centers (larger than many cities' power needs) • Over 1.3M GPUs by end of 2025 • A re-architected physical backbone for the Llama AI ecosystem Why it matters: - This isn’t about chips anymore. It’s about who owns the land, power, and fiber underneath AI. - Private credit is now central to digital infrastructure. - And Meta is building a compute empire that rivals Microsoft, Amazon, and Google. The new arms race won’t be won with software. It’ll be won with steel, silicon, and sovereign-scale capital. The infrastructure wars have started. And Meta just fired the next shot. #datacenters
Incredible move in treasury strategy. Meta isn't just building data centers; it's redefining capital allocation for the AI era. By choosing private credit and leaseback instead of tapping into cash or traditional debt, they’ve: 1.Preserved balance sheet strength 2.Accelerated infrastructure build-out 3.Shifted asset risk to infrastructure investors
It’s hard not to be impressed by the scale of ambition here. But 2+ GW of new data centres is a lot of power. That’s roughly the same as the entire electricity demand of countries like Namibia, Cyprus or Senegal. And we’re adding that level of consumption again and again as AI infrastructure scales. The AI race is already reshaping capital markets. It will also reshape how we think about sustainability — not just carbon, but water, materials and long-term resilience. If AI becomes foundational to everything, so will the environmental cost of running it. Are we ready for that?
That bit about steel, silicon, and sovereign-scale capital beyond software nails it. The infrastructure that powers the ecosystem is the heart of the race.
Classic old school OPM...
Check this out, Mikaela Shannon !! 💡
Hedging his bets obviously. Too much competition in this AI hyper scaling race
This is fascinating because Meta is bypassing traditional funding to orchestrate a massive private credit deal—treating AI data centers like critical utilities, not tech capex—giving them speed and scale without burdening their balance sheet. It signals a new era where control over power, land, and physical infrastructure—not just chips or code—defines the next leaders in AI. Thank you for sharing! Theo
Software Engineer
2moHad always wondered when they were going to pull off infrastructure. In my opinion, they had every resource to be a strong contender in cloud but apparently never had the vision. Happy to see them taking this on