If you are searching for AI data center acquisition news today, the most important thing to understand is that the market is moving on two tracks at once. First, there are actual acquisitions, like Ecolab’s $4.75 billion move for liquid-cooling company CoolIT and Equinix plus CPP Investments buying Nordic operator atNorth for $4 billion. Second, there are capacity takeovers and financing deals that are not technically acquisitions, but still reshape who controls AI infrastructure, such as Microsoft stepping into a 700-megawatt Texas project and Meta locking in billions of dollars of AI capacity from Nebius. As of March 25, 2026, both tracks matter because they show where real value is concentrating: not just in chips, but in cooling, power, land, and data center platforms that are already “ready.” (Reuters)

A practical way to think about this is simple: in the earlier AI boom, people obsessed over models and chatbots. In this phase, buyers are increasingly paying for the less glamorous assets that make those models possible. That is why this story is starting to feel less like a software trend and more like an industrial land rush.

What counts as AI data center acquisition news today?

The freshest confirmed acquisition-specific headline is Ecolab’s agreement to buy CoolIT Systems from KKR for $4.75 billion in cash. CoolIT makes liquid-cooling systems used by major chip ecosystems including Nvidia and AMD, and Ecolab said the deal is about expanding into the cooling layer that next-generation AI data centers increasingly need. Reuters reported CoolIT is projected to generate about $550 million in revenue over the next year, with the acquisition expected to close in Q3 2026 and become accretive to Ecolab’s adjusted earnings by 2028. (Reuters)

The other major recent acquisition is Equinix and CPP Investments buying atNorth for $4 billion. That deal matters because it is not just a financial trade. It gives the buyers exposure to a Nordic platform with eight operational data centers, several development sites, and about 1 gigawatt of secured power planned for expansion. Partners Group said contracted EBITDA at atNorth increased 14-fold since 2022, which tells you why mature, power-secured platforms are commanding premium valuations. (DataCenterKnowledge)

At the same time, the biggest today-style headline in the broader AI data center news cycle is not a formal acquisition. Reuters reported that Microsoft agreed to lease a 700-megawatt data center project in Abilene, Texas, after Oracle and OpenAI stepped away from negotiations with developer Crusoe. That is a reminder that “control” over AI infrastructure can change hands through leasing, capacity agreements, joint ventures, and financing structures, not only through full-company acquisitions. (Reuters)

For readers tracking the wider infrastructure backdrop, AI Tribune’s earlier piece on AI data center news today fits neatly alongside this acquisition-focused angle.

The biggest deals are all buying bottlenecks, not just buildings

The smartest way to read the current market is that buyers are not simply purchasing square footage. They are targeting the real bottlenecks.

In Ecolab’s case, the bottleneck is cooling. AI racks are denser, hotter, and harder to run efficiently with legacy air-cooling alone, which is why liquid cooling has become one of the hottest acquisition targets in the sector. Reuters also reported before the deal closed that KKR’s sale process for CoolIT was expected to value the company at more than $3 billion, underscoring just how strategic cooling has become in the AI stack. (Reuters)

In the atNorth transaction, the bottleneck is power-ready regional capacity. The Nordics are attractive because of climate, energy mix, and the ability to support high-density compute more efficiently. Data Center Knowledge reported the transaction also came with a $4.2 billion financing package, which shows how capital-intensive even “existing platform” deals have become. (DataCenterKnowledge)

And in the case of Microsoft’s move in Texas, the bottleneck is plainly available megawatts. A 700MW project next to a major AI campus is not ordinary real estate. It is strategic capacity. That is why even a lease reassignment becomes major news. (Reuters)

This is also why older AI Tribune coverage like Why OpenAI Is Burning Cash While Google and Anthropic Aren’t as Much remains relevant here: the economics of AI increasingly run through infrastructure ownership, power access, and capital structure, not just model quality.

Why valuations keep climbing

The underlying math is brutal, and it helps explain why acquisition prices look so aggressive.

Data Center Knowledge, citing S&P data, reported that 113 data center transactions were completed globally in 2025, representing more than $69 billion in total deal value, about $8 billion above the previous annual record. The same report said Synergy tracked 575 data center-oriented M&A deals since the beginning of 2024 with an aggregate value of $151 billion, and 84% of the deal value came from private equity funding. That is not a niche market anymore. It is a global capital stampede. (DataCenterKnowledge)

The broader demand case is even bigger. McKinsey says hyperscalers and colocation providers have announced plans for more than 2,600 new data centers, with roughly one-quarter planned for cities that do not currently have a data center footprint. McKinsey also notes that power constraints and permitting delays are increasing the value of existing sites with assured electricity access. (McKinsey & Company)

Morgan Stanley goes further, estimating about $2.9 trillion in global data center construction cost through 2028, with more than 80% of that spending still ahead. Goldman Sachs Research has projected data center power demand could rise about 175% by 2030 versus 2023 levels. When you put those forecasts next to confirmed M&A activity, the premium buyers are paying starts to make more sense. They are effectively paying to skip years of waiting. (Morgan Stanley)

A relatable example here is real estate. In a cold housing market, buyers negotiate hard. In a market with almost no inventory, they pay up for homes that are already livable. AI data center acquisitions look a lot like the second scenario: ready power, ready cooling, and ready permitting are the equivalent of “move-in ready.”

What online reaction is saying right now

There are not many meaningful consumer-style reviews for deals like these, because this is enterprise infrastructure, not a laptop launch. What you do see instead is a mix of analyst reaction, investor commentary, and trade-press interpretation.

For Ecolab, Barron’s framed the CoolIT deal as a clear push into AI infrastructure but also highlighted that the purchase price works out to about 29 times next year’s EBITDA, while Ecolab shares slipped about 1.5% after the announcement. In other words, the market appears to like the strategic direction but is still debating whether the price was rich. (Barron’s)

For Nebius, which is not an acquisition story but is highly relevant to infrastructure allocation, Reuters reported Meta signed agreements worth up to $27 billion over five years, including $12 billion of AI capacity by 2027 and a conditional $15 billion add-on. Meanwhile, Bank of America initiated coverage with a Buy rating, arguing Nebius is positioned as a “share-taker” in AI compute. That mix of customer demand plus bullish analyst coverage shows how investors are rewarding platforms that can secure and monetize AI infrastructure quickly. (Reuters)

The most objective takeaway is this: online reaction is not screaming that every deal is cheap. But it is consistently signaling that scarce infrastructure assets are being treated as strategic, not optional. (Barron’s)

What to watch next in AI data center acquisition news

The next wave of deals will likely center on four areas: cooling, power-secured campuses, regional operators, and financing vehicles.

Cooling is obvious after CoolIT. Regional operators are in play after atNorth. Financing structures are becoming just as important as outright purchases, with Morgan Stanley noting that AI infrastructure is increasingly pulling in private credit, securitized funding, and joint-venture capital. And the trade press has already reported that Blackstone is preparing a publicly traded AI data center acquisition vehicle, which, if it materializes, would further institutionalize the race for these assets. (Morgan Stanley)

Geopolitics also matters more than many readers realize. If you want the international angle, AI Tribune’s Can China Win the AI Race? is a good companion read because the AI infrastructure contest is increasingly about domestic compute, sovereign capacity, and control over supply chains as much as it is about models. Morgan Stanley explicitly frames AI as a geopolitical competition across chips, compute, energy, and data. (Morgan Stanley)

FAQ

What is the biggest AI data center acquisition right now?
The most prominent recent confirmed acquisition tied directly to AI data center infrastructure is Ecolab’s $4.75 billion purchase of CoolIT Systems, announced on March 20, 2026. (Reuters)

Is Microsoft’s Texas move an acquisition?
No. Reuters reported it as a leasing arrangement for a 700MW Texas project that Oracle and OpenAI had previously been discussing with Crusoe. It is major infrastructure news, but not a company acquisition. (Reuters)

Why are AI data center acquisitions getting so expensive?
Because buyers are paying for scarce assets: existing capacity, secured power, cooling technology, and faster time-to-market. Recent market data and research all point to power constraints and huge capital requirements as key drivers of valuation. (DataCenterKnowledge)

Are investors still interested in data center M&A in 2026?
Yes. Data Center Knowledge reported that 2025 set a new record with 113 transactions and more than $69 billion in value, and analysts expect the market to remain robust in 2026 despite power and AI-cycle risks. (DataCenterKnowledge)

Final take

The best way to read AI data center acquisition news today is not as a random pile of deals. It is a map of what the industry now values most. Right now, buyers are paying up for what is hardest to build from scratch: cooling expertise, power access, operational campuses, and strategic regional platforms. That is why the headline-grabbing AI story is no longer just who has the smartest model. It is who can secure the physical backbone behind it. (Reuters)

Do you think the next big AI winners will be model companies, cloud providers, or the quieter infrastructure owners buying up the pipes, power, and cooling behind the scenes? Drop your take in the comments, because this is one of those moments where the boring part of AI might quietly become the most valuable part.