bytevyte
bytevyte
Language
ai-beats —

BlackRock and IFM Near $25 Billion Stack Data Center Deal in Asia-Pacific

Stack data center deal

BlackRock and IFM Investors are in exclusive negotiations over a Stack data center deal that would hand the consortium Stack Infrastructure's Asia-Pacific portfolio for as much as $25 billion. The talks entered an exclusive phase this week and remain at the due-diligence stage, with no binding agreement signed and the final figure still open to repricing. The assets in play sit in Tokyo, Osaka, Sydney and Melbourne, four markets where AI and cloud demand has tightened capacity enough to command hyperscale valuations from financial buyers.

The consortium pairs the Artificial Intelligence Infrastructure Partnership (AIP), the BlackRock-sponsored vehicle, with Australia's IFM Investors. Stack Infrastructure is owned by Blue Owl Capital. A price anywhere inside the $20 billion to $25 billion band would make this a transfer between two different pools of private capital rather than a fresh commitment of new capacity.

Exclusivity matters because it removes competing bidders for a defined period while the buyer inspects the books. It does not guarantee completion, and a portfolio of this size carries enough complexity that a reprice is a live possibility rather than a remote one.

What the Stack Data Center Deal Covers

The portfolio consists of operating data centers, not land banks or shell sites, and that distinction sets the price. Completed, leased capacity in Tokyo and Sydney carries a different risk profile from a greenfield build: revenue is contracted, power is secured, and the construction risk has already been absorbed by the previous owner.

ItemDetail
TargetStack Infrastructure's Asia-Pacific data center portfolio
MarketsTokyo, Osaka, Sydney, Melbourne
BuyerAIP (BlackRock-backed) with IFM Investors
Current ownerBlue Owl Capital
Valuation bandAbout $20 billion to $25 billion
StatusExclusive talks, due diligence under way, no binding agreement

Stack Infrastructure has built its business on hyperscale tenants pre-committing to space, and that model is what makes the Asian assets saleable at this scale. A buyer is not underwriting speculative vacancy. It is underwriting the credit of tenants that have already signed.

The four-market spread does more than diversify. Tokyo and Osaka serve Japanese enterprise and cloud demand; Sydney and Melbourne anchor the Australian market, where power procurement and planning approvals have become the binding constraint on new supply. Assembling all four in one transaction is the part a rival bidder cannot easily replicate, because the alternative is four separate negotiations with four separate owners.

A Buyer Made of Its Own Suppliers

AIP's founding members are BlackRock, Global Infrastructure Partners, MGX, Microsoft and NVIDIA. Temasek and the Kuwait Investment Authority sit among its anchor investors. The vehicle launched in late 2024 with a mandate to deploy more than $30 billion into AI-related infrastructure, and a purchase at the top of the range would consume the bulk of that mandate in a single move.

That composition explains the strategy. The consortium contains a chip designer and a hyperscaler whose own procurement decisions determine what the Asian sites are worth. Nvidia's accelerator shipments and Microsoft's cloud expansion are, indirectly, the demand curve behind the rents that justify a $25 billion cheque. Owning the real estate does not give AIP control over that demand, but it does give the vehicle exposure to it without carrying the operating risk of a technology company.

IFM's presence adds a second layer. The Australian fund manager brings long-duration pension capital that matches the useful life of a data center better than a hyperscaler's own balance sheet does. Microsoft and Nvidia contribute strategic insight and, potentially, anchor tenancy. IFM, Temasek and the Kuwait Investment Authority contribute the patient equity that a ten-year hold requires.

Why Compute Is Being Repriced as an Asset Class

The transaction matters less for what changes in Tokyo than for what it says about how AI infrastructure is financed. The earlier model was vertical integration: hyperscalers built, owned and operated their own capacity and treated it as a cost centre. The emerging model splits the stack. Compute is rented, the buildings are owned by infrastructure funds, and the equity behind them comes from pensions, insurers and sovereign wealth.

That split redistributes risk. The hyperscaler keeps demand risk, because it can scale workloads up or down. The infrastructure owner takes residual-value risk, because a facility engineered for one power density and cooling design may be worth less if accelerator architectures change faster than the building's useful life.

Asia sharpens the trade. Capacity in Tokyo, Osaka, Sydney and Melbourne is scarce relative to US markets, and land, power and approvals are harder to assemble. Scarcity is what allows a financial buyer to pay a price previously reserved for strategic acquirers. If the regional supply pipeline catches up, that premium compresses.

The timing is consistent with AIP's own schedule. A vehicle launched in late 2024 with a $30 billion mandate needs to deploy, and the fastest route to scale is to buy assets that already generate revenue. That the Stack data center deal surfaced this week, and was still live a day later, suggests the diligence window is being used to test price rather than to test intent.

The Trade-Offs

Three routes were available to AIP. It could develop greenfield capacity, which takes years and carries permitting and power risk. It could lease from an existing operator, which keeps capital light but leaves no residual value. Or it could buy finished assets, which is the path the consortium has chosen. The third option is the only one that converts a deployment deadline into immediate, contracted revenue, and it is also the only one where the price is set by whatever the seller will accept rather than by construction cost.

The counter-argument is leverage. Blue Owl is a credit-oriented manager, and the assets it holds have been financed accordingly. Whether AIP assumes that debt, refinances it, or structures the purchase with fresh sovereign equity determines the actual return profile, and none of that has been disclosed. The distance between a $20 billion and a $25 billion headline is wide enough that the capital structure, not the real estate, may decide the final number.

Concentration risk deserves a mention too. AIP's founding members include companies whose leasing commitments would underpin the portfolio's cash flows. A buyer whose own shareholders are also its tenants has an informational advantage in underwriting, and a structural conflict if those tenants ever seek to renegotiate rents.

The deal is not done. Due diligence on operating assets in four jurisdictions is slow work, and the absence of a binding agreement leaves room for a rival bid, a price cut, or a decision by Blue Owl to hold. Each of those outcomes would be read as a verdict on the valuation itself.

Why this matters

The transaction converts hyperscaler lease commitments into the underwriting benchmark for a $25 billion valuation, and that benchmark will be applied to every comparable Asian portfolio that reaches the market. If it holds, compute capacity has completed its shift from an operating cost inside technology companies to a traded asset class owned by private capital, with sovereign funds setting the price. The signal to watch next is whether the Stack data center deal becomes a binding agreement at the top of the range or reprices after diligence.

✔Human Verified


Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.