Blackstone, peers chase $900B datacentres
- Blackstone, Blue Owl, Ares, and Apollo are scaling data-center bets as AI demand turns server halls into one of private capital’s biggest buildouts. - The telling figure is $900 billion — the near-term opening investors see in financing campuses, leases, power links, and even GPU fleets. - Money is plentiful, but power, land, and grid hookups are not — so capital alone will not solve the AI bottleneck.
Data centers are turning into the new private-equity obsession. Not because the buildings are glamorous — they are giant concrete boxes full of servers — but because AI has made compute look like a scarce industrial resource. That is the shift behind the latest rush by Blackstone, Blue Owl, Ares, Apollo, and other big alternative-asset firms to pour money into the sector. The basic bet is simple: companies want AI capacity now, and somebody has to finance the land, substations, cooling, fiber, and hardware fast enough to meet that demand. ### Why are financiers suddenly so interested? Because AI changed the economics. A normal cloud or colocation site was already capital-intensive, but AI clusters need denser power, more cooling, and much bigger upfront spending. McKinsey’s estimate is the clearest scale marker — data centers worldwide may need $6.7 trillion of investment by 2030, with $5.2 trillion tied to AI-heavy capacity. Once a market gets that large, private capital does not want to just lend around the edges. (businessinsider.com) It wants to own the pipes. ### What is Blackstone actually doing? Blackstone already has one of the biggest operating footprints through QTS and AirTrunk, and it keeps adding financing on top. QTS moved in April to raise $4.6 billion for AI-related build-out, starting with a 10-year green bond tied to Georgia campuses. Blackstone also said in July 2025 that it would invest more than $25 billion in Pennsylvania digital and energy infrastructure and help catalyze another $60 billion there. (mckinsey.com) That is not a small side bet — it is industrial policy with private money. ### Why are Blue Owl, Ares, and Apollo in it too? Because each firm can attack a different layer of the stack. Blue Owl has been building digital-infrastructure funds and closed a $7 billion vehicle in 2025 aimed at hyperscaler demand. It also backed Meta’s Louisiana AI campus through a joint venture valued at about $27 billion, with Blue Owl funds owning 80%. Ares has been expanding Ada Infrastructure and bought a 314-acre Virginia development site plus two hyperscale facilities in Leesburg. (datacenterdynamics.com) Apollo has gone one step further into the guts of AI spending — it led a $3.5 billion capital solution supporting a $5.4 billion GPU lease deal for xAI. ### So are they buying buildings or selling financing? Both — and that is the important part. The old model was easier to describe: build or buy a data center, lease it, collect rent. The new model is messier and bigger. Investors can own the campus, fund the construction, provide private credit, structure sale-leasebacks, back powered land, and even finance the compute gear itself. Think of it less like office real estate and more like a toll road system where every ramp, lane, and power feed can be monetized. (blueowl.com) ### Does all this money fix the shortage? Not fully. The catch is that cash is not the scarcest input anymore. Power is. Utilities, transmission lines, gas generation, permits, transformers, and grid interconnection queues now decide how fast AI capacity can come online. That is why Blackstone’s Pennsylvania push bundled energy and digital infrastructure together, and why so many deals now revolve around “powered land” rather than just finished buildings. (businessinsider.com) Capital can accelerate a project, but it cannot conjure 500 megawatts overnight. ### What does this mean for customers? For big AI buyers, it means more ways to rent capacity instead of building everything themselves. That should expand premium colocation and hyperscale supply over time. But in the short run, it can also make the market feel tighter, not looser, because the best sites get pre-leased early and giant customers lock up power years ahead. More money is entering the system — but so is more demand. (blackstone.com) ### What is the bottom line? Private equity is treating data centers like the next core utility. That sounds dramatic, but turns out it fits: AI needs electricity, land, cooling, and financing at absurd scale. The firms chasing this market are not just making a real-estate bet. They are trying to own the financing layer of the AI economy itself. (businessinsider.com) (mckinsey.com)
