Amsterdam's Vertical Data Center Exposes a Policy Gap

The Netherlands' hyperscale definition fails to account for vertical builds, revealing deeper tensions between compute demand and grid capacity.

What Happened

A large data center under construction in Amsterdam, set to be fully used by Microsoft, has reignited political debate in the Netherlands. The facility consists of three towers reaching 85 meters in height, with a total grid connection of 78 megawatt. When fully operational, it would consume roughly as much electricity as a small city on an annual basis.

The project was permitted five years ago, before the Dutch national government introduced its 2022 policy restricting so called hyperscale data centers. Demissionary Minister Mona Keijzer confirmed in parliament that the existing permit cannot be reversed. Several members of parliament questioned whether current rules are fit for purpose, given that the facility technically falls below the surface area threshold that defines a hyperscale under Dutch policy.

The current definition requires both a minimum of 70 megawatt in electrical capacity and at least 100,000 square meters of land use. Because the Amsterdam facility occupies only around 23,000 square meters of ground space through its vertical design, it does not qualify as a hyperscale. Lawmakers are now asking whether the definition should be revised to use an "or" rather than "and" logic between the two criteria.

Structural Context

The Dutch hyperscale restriction was introduced in response to public criticism of a proposed Meta facility in Flevoland, which was ultimately abandoned. The policy was designed to limit very large single tenant data centers, particularly in regions already facing grid congestion and housing shortages. It reflected a broader concern across the Netherlands about how scarce grid capacity should be allocated.

What the Amsterdam case reveals is that policy tools built around physical footprint do not capture the actual constraint, which is power. A vertically constructed facility can deliver the same compute density and draw the same electrical load as a sprawling campus, while fitting neatly within the square meter limits of existing regulation. The real variable is not land use. It is energy consumption and the capacity of the local grid to serve it.

The Netherlands faces some of the most acute grid congestion in Europe. Transmission capacity is limited, new connections face years of delay, and competing demands from housing, industry, and electrification are intensifying. In this context, any large new power load in a constrained urban area amplifies existing tensions, regardless of the building's shape.

The Enki Perspective

This situation illustrates a broader pattern across Europe. Data center policy is often built around physical characteristics, such as size, location, or zoning, rather than the underlying infrastructure constraint: access to power. When policy targets the visible footprint of a facility instead of its energy profile, the result is regulatory gaps that neither slow expansion nor resolve the underlying tension.

The structural answer is not to restrict compute, but to rethink where and how it is powered. Grid constrained regions like Amsterdam face a genuine bottleneck. Adding significant new electrical load to already saturated networks creates friction for all connected users. But compute demand does not disappear when a permit is denied. It simply moves elsewhere, often to locations with less oversight and less efficient energy use.

Project Enki operates from a different premise. Rather than competing for grid connections in congested urban corridors, Enki positions AI infrastructure at the point of energy generation, converting stranded and curtailed renewable power into productive compute capacity. This model avoids the transmission bottleneck entirely. It does not require municipalities to choose between housing and data centers, because it draws on energy that would otherwise go unused.

Vertical construction, modular deployment, and innovative cooling are all part of the industry's evolution toward more efficient infrastructure. But efficiency at the facility level does not resolve the system level question of where the power comes from. That requires a shift in how infrastructure is sited: not where the grid allows, but where generation already exists.

What This Signals

The Amsterdam case is unlikely to be the last of its kind. As AI workloads grow and operators find creative ways to deliver density within existing regulatory frameworks, the gap between policy intent and practical outcome will widen. Lawmakers are already signaling interest in tighter definitions, but adjusting thresholds alone will not address the core issue.

What the market is beginning to recognize is that the future of compute infrastructure depends on solving the energy equation. Capital is available. Demand is accelerating. The binding constraint is power, and more specifically, the ability to deliver it at scale without overloading shared networks. Infrastructure models that align compute with generation, rather than routing it through congested grids, represent a structural advantage that will only become more relevant as AI capacity requirements continue to grow.

For Europe, this is also a question of digital sovereignty. Compute that depends entirely on constrained urban grids and lengthy permitting cycles cannot scale at the pace the market requires. Energy aligned infrastructure offers a path that is faster, more repeatable, and more compatible with the continent's renewable energy ambitions.

Source: 27 januari 2026 https://nos.nl/artikel/2599925-keijzer-kan-niks-doen-tegen-enorm-datacenter-voor-microsoft-in-amsterdam