What was reported this week: a BlackRock-backed consortium is in exclusive talks to buy Stack Infrastructure’s Asia-Pacific data centers for up to $25 billion, in what would rank among the largest digital infrastructure deals ever struck in the region.
What was reported in May: the same seller was weighing a sale of the same Asian business at a valuation of more than $30 billion. Both figures came from Bloomberg. Nobody has put them next to each other.
On May 6, bankers at Morgan Stanley and Deutsche Bank were handed a mandate that looked, at the time, like the easiest sell of the year.
Their client was Blue Owl Capital, the alternative asset manager that owns Stack Infrastructure. The asset was Stack’s Asian data center business, spread across Australia, Japan and Malaysia. The number attached to it in the reporting that day was more than $30 billion. Global investors were pouring capital into Asian data centers, the AI buildout was accelerating, and the assets sat in exactly the markets everyone wanted.
Four and a half months later, the same business is in exclusive talks at up to $25 billion.
That gap is the story, and it has gone almost entirely unremarked. Coverage this week has treated the $25 billion figure as a triumphant datapoint for the AI infrastructure trade, which is understandable, because $25 billion is an enormous number and the buyer list reads like a who’s who of global capital. The consortium is built around the Artificial Intelligence Infrastructure Partnership, the BlackRock-backed vehicle, alongside IFM Investors, the Australian pension-owned manager. Both were circling the asset as far back as June.
But an asset that was pitched at north of $30 billion in May and is being negotiated at up to $25 billion in late September has not appreciated. It has repriced, by something on the order of seventeen percent, during the single most euphoric stretch the AI trade has had all year. In the same window, Meta alone added roughly half a trillion dollars of market value.
Two Numbers, Two Different Things
Honesty about what the comparison shows matters here, because the easy version of this story is wrong.
The May figure was an aspiration. Bloomberg reported then that Stack was “considering its options,” including a partial or full sale, at a valuation of more than $30 billion. That is a seller’s number, produced before a single binding bid existed, and seller’s numbers are routinely optimistic by design. The September figure is different in kind: it is the range inside an exclusive negotiation, with due diligence being prepared. It is much closer to a clearing price.
So the correct reading is not that the assets lost seventeen percent of their value. It is that the ask met the market, and the market answered five billion dollars lower.
There is a second explanation sitting in the same reporting, and it deserves equal billing. The May description of the perimeter named Australia, Japan and Malaysia. This week’s description names Tokyo, Osaka, Sydney and Melbourne. Malaysia is absent. If the Malaysian assets have been carved out, part of the difference is simply a smaller thing being sold, not a cheaper one.
That caveat is real and we cannot resolve it from outside. Neither Blue Owl nor the consortium has confirmed the perimeter. What we can say is that both readings are unflattering to the prevailing narrative, because either premier Asian AI infrastructure is clearing below its spring ask, or the most attractive slice of it was hard enough to move that it had to be separated out.
An asset pitched above $30 billion in May and negotiated at up to $25 billion in September has not appreciated during the biggest AI rally of the year. It has found its level.
The Seller Is the Part Worth Watching
Follow the direction of travel. Blue Owl is not raising capital to build more Asian data centers. Blue Owl is selling, it hired two bulge-bracket banks to do it, and it stepped up those efforts as the process ran. A firm that believed these assets would be worth materially more in eighteen months does not run a competitive auction for them now.
Meanwhile the buyers are pension money and an infrastructure partnership. IFM is owned by Australian superannuation funds. These are long-duration, yield-seeking pools of capital, and they price assets off contracted cash flows and lease terms, not off a narrative about superintelligence. When that kind of money sets the clearing price for AI infrastructure, the number it produces is the closest thing the sector has to an honest mark.
This is the second time in under a year that BlackRock’s infrastructure machine has set the reference price for a large data center platform. The $40 billion Aligned Data Centers acquisition established one benchmark. Stack’s Asian portfolio is about to establish another, and this one is arriving lower than the seller wanted.
Here is BTN’s position. The AI infrastructure trade has spent two years being valued by the people building it and the people funding the builders, which is a closed loop. What is happening at Stack is the first thing in a while that resembles price discovery: a real seller, real banks, a real auction, and buyers whose fiduciary duty runs to retirees rather than to a growth story. The resulting number came in below the ask.
That is not a crash and nobody should read it as one. Twenty-five billion dollars is a vast sum and the deal, if it closes, is a genuine vote of confidence in Asian digital infrastructure. But it is a vote cast at a discount to the spring asking price, in the middle of the most bullish AI tape since the cycle began, and the sector’s commentary has managed to report the headline figure without once mentioning the earlier one.
The people who should be uncomfortable are the ones marking comparable portfolios to May’s number. If Stack’s Asian assets clear at $25 billion, every holder of similar infrastructure now has a fresher, lower, and considerably more real comparable to explain.
Deliberations are ongoing. Reporting is explicit that talks could be prolonged or fail to produce an agreement at all. Treat the $25 billion as a live negotiation rather than a done deal.
But watch where it lands. In a sector that has had almost no honest marks, this one is going to be quoted for a long time.