The race for AI infrastructure has often been framed as a battle of chips, models, and larger data centers. However, the announcement of CPP Investments and Equinix agreeing to acquire atNorth — with a business valuation of $4 billion and a financing package of $4.2 billion — reveals a more pragmatic reality: AI is increasingly becoming a matter of electricity capacity, thermal design, and permits, rather than just a tech marketing narrative.
atNorth may not be a household name outside of infrastructure circles, but it operates a platform that fits perfectly into the new demand pattern: eight data centers spread across Denmark, Finland, Iceland, Norway, and Sweden, with additional sites in the pipeline – Kouvola (Finland), Ølgod (Denmark), and a campus in Stockholm – along with a crucial asset: 1 GW of secured energy capacity, plus potential future capacity.
The deal, announced from Reykjavik on February 27, 2026, leaves CPP Investments with a 60% stake and Equinix holding 40%, while atNorth will continue to operate under its brand. At first glance, it appears to be another consolidation in a growing market. In reality, it is a strategic repositioning aimed at winning a war where the scarce resource is not square meters but rather usable megawatts and the capacity to dissipate heat as densities reach industrial levels.
The Acquisition Buys More Than Just Square Meters: It Purchases Electricity, Cooling, and Time
When reviewing a business model, I separate the “shell” from the supporting structure. In data centers, the shell consists of the buildings; the supporting structure is the triangle of energy–thermal characteristics–timelines. AtNorth has built its value proposition around that triangle: relevant energy availability (1 GW), operation in countries with natural cooling advantages, and an explicit focus on high-density workloads using liquid cooling across various sites.This explains why Equinix does not need to acquire 100% for the deal to be strategic. With a 40% stake, it gains exposure to capacity and growth in the Nordic countries and access to a platform ready to absorb demand from AI, hyperscalers, and enterprises without having to shoulder the entire execution and capital burden alone. CPP Investments, for its part, positions itself as the majority owner of a real asset with demand accelerating “due to cloud and AI adoption,” according to its own executive.
The key element here is “time”: the market rewards those who can turn energy and permits into operational capacity faster than their competitors. The briefing also mentions a mega-site in Sollefteå (Sweden) as part of the pre-deal growth momentum. These announcements often serve as a thermometer for bottlenecks: if an operator already has a pipeline, capital flows in to accelerate the work, not to create a thesis from scratch.
There is also a detail often underestimated: a data center for AI is not just another building; it is a reverse power plant. Liquid cooling and high-density designs require an operational engineering approach more akin to an industrial installation than a real estate asset. Those who master this operation can provide certainty, and certainty comes with a price tag.
An Alliance with Shared Responsibilities: Patient Capital and a Global Sales Network
The 60/40 structure makes mechanical sense. CPP Investments acts as the pillar that supports the weight of capital and the patience needed for infrastructure cycles. Equinix brings in another piece: global relationships and operational strength in digital infrastructure, along with the experience needed to convert demand into repeatable contracts. There is no need to speculate on specific synergies; it suffices to look at what each side requires for the arrangement to work.AtNorth provides its local platform and sustainability narrative: renewable energy, heat reuse, modular designs, and principles of circular economy. This component is not merely a reputational adornment; in markets under regulatory pressure and energy limits, operational sustainability reduces friction: it increases the likelihood of permits, diminishes local opposition, and may facilitate integrations like district heating. Each point of friction removed is a month gained in an industry where each month translates into revenue.
The “provisionally agreed” financing package of $4.2 billion underscores another reality: growth here is driven not by marketing campaigns but by capex. In infrastructure, the opportunity cost of not building is severe when demand is concentrated among a few major buyers. Capital flows in to ensure that the project pipeline does not turn into a graveyard of renderings.
Furthermore, there is a corporate design choice that appears deliberate: atNorth “will operate independently under its brand.” In systemic terms, this preserves the local speed of execution (permits, construction, operation) while allowing the shareholders to add resources without rewriting the entire organization. It’s a way of not breaking the engine while swapping out the turbo.
The Nordic Landscape: Moderate Concentration and a Non-Reproducible Advantage
The briefing places the Nordic data center market at $1.22 billion in 2025, with projected growth of 8.39% CAGR through 2033. It also notes a level of concentration where the top five operators control around 42% of installed megawatts, including Equinix, Digital Realty, Bulk Infrastructure, atNorth, and Green Mountain. This paints a sector with enough competition to force performance, yet sufficient concentration for energy and land access to make a difference.Here, the Nordic advantage is not a single factor; it’s a challenging-to-copy combination: availability of renewable energies, a climate conducive to cooling, and a history of projects that have already normalized the presence of large facilities. Yet this advantage is not automatic. The very demand that attracts capital also attracts competition, thus shifting the bottleneck: when everyone wants to build, scarcity can arise in electrical interconnections, supply chains, cooling equipment, and regulatory timelines.
Consequently, this operation is better understood as an acquisition of “capacity options.” AtNorth already has eight operational sites and three in development. This is tangible pipeline. It doesn’t guarantee flawless execution, but it does mitigate the risk of starting from scratch in markets where the maturation time can be lengthy.
Another point I usually scrutinize is the temptation to sell to everyone. In data centers, segment dispersion often proves to be an expensive mistake: each type of customer demands different densities, redundancies, and contracts. AtNorth describes itself as specializing in “high-density colocation” and “built-to-suit,” focusing on AI/HPC workloads. This product atomization — high density and tailored solutions for specific workloads — makes capex more defensible, as infrastructure is designed for a clear consumption pattern, not for “just in case.”
The Real Bet and Its Risks: Physical Execution, Regulations, and Capacity Promises
A $4 billion agreement with $4.2 billion in financing is a bet that demand will continue to absorb new capacity at prices that justify the investment. While the briefing cites momentum from hyperscalers, enterprises, and AI adoption, it also acknowledges a key limitation: there are no recent public figures on revenues, market share, or atNorth’s debt breakdown from available sources. As a result, the analysis shifts from “multiples” to mechanics.The mechanics involve three structural risks.
First, execution risk. Building data centers across multiple countries, with high-density standards and liquid cooling, is an engineering discipline, not a traditional real estate operation. Deployment speed is gained through suppliers, modular design, and standardized operation. It is compromised by excessive customization and late changes.
Second, regulatory and permitting risk. The announcement itself notes that the closure is subject to usual conditions and regulatory approvals, without a public timeline. In critical infrastructure, the timeline is not a detail: it’s the blueprint of the return load. Months of delays can turn an expansion into a chase.
Third, risk of capacity promises versus actual capacity delivery. “1 GW secured” is a powerful signal, but the business does not collect revenue for gigawatts on paper; it collects for megawatts delivered with contractual availability. The gap between the two is where capex deviations, grid limits, and operational restrictions lurk.
That said, there is an implicit defense: Equinix is not a player needing to learn from scratch how to market digital infrastructure; CPP Investments is not an owner frightened by long cycles. This combination does not eliminate physical risks but enhances the capacity to absorb them.
The Takeaway for C-Level Executives: AI Infrastructure is Earned by Securing Scarce Inputs
This announcement is not just another M&A. It is a signal that AI infrastructure is entering a phase where the strategic asset is the right to operate: energy, thermal design, permits, and a pipeline that turns capital into capacity frictionlessly. AtNorth fits this pattern with eight operational sites, three in development, and a technical posture geared towards high density with liquid cooling.From the business model architecture, CPP Investments and Equinix are acquiring a structure that already holds weight, adding financing to elevate floors without collapsing the base. Returns will not come from “being trendy” but from executing developments, signing contracts, and operating with energy and thermal efficiency in a competitive market.
Companies do not fail due to a lack of ideas; they fail because their model components do not align to convert demand into measurable value and sustainable cash flow.
