Norway’s data-centre bottleneck is becoming an allocation regime

Norway’s data-centre bottleneck is becoming an allocation regime

Norway’s data-centre market has reached a point where access to power is no longer only a connection problem. It is becoming an allocation decision.

On 10 November 2026, the Norwegian government said it intends to introduce a licensing regime covering both existing and future data centres. One day earlier, transmission-system operator Statnett temporarily stopped making new grid-capacity reservations for data centres and paused parts of the technical approval process. The suspension is expected to end no later than 1 August 2027.

The timing matters. This is not a distant policy consultation arriving after the market has already adjusted. It is an immediate constraint followed by a prospective selection framework. Developers, operators, investors and large users now have to evaluate Norwegian projects against two gates: whether the grid can accommodate them and whether the proposed use is likely to satisfy a public-interest test.

That changes the competitive logic of the market.

The headline numbers describe different stages, not one pipeline

Statnett reports four separate categories for data-centre demand:

• approximately 800 MW already connected;
• 4,200 MW with reserved grid capacity;
• 9,100 MW in the connection queue; and
• a further 18,000 MW of connection requests.

Norway’s peak electricity demand is just over 25,000 MW.

Those figures should not be added and presented as a single executable data-centre pipeline. They represent different levels of maturity in the grid process. They are electrical connection figures, not IT load. A request is not a reservation, a reservation is not an energised facility, and a connected electrical load is not the same measure as sellable IT capacity.

The distinction is commercially decisive. A project with a large request can still lack a firm delivery date, a technical approval, completed planning, financing or a customer commitment. Conversely, an operating campus may have valuable contracted expansion capacity even when its public IT figure is far smaller than a national queue headline.

PowerlandMap’s current Norwegian records illustrate that spread. The Aker Nscale project in Skien carries a 96 MW grid-capacity application, explicitly classified as requested rather than secured, connected or IT capacity. ASP’s K11 expansion in Stavanger records 6 MW of contracted IT capacity, with future delivery separated from the already operating parent site. Nscale’s Glomfjord campus reports 30 MW of operational power capacity, expandable to 60 MW, but that number is not relabelled as IT capacity.

These are three different assets and three different measurements. Treating them as comparable “MW” would erase the very risk that the new Norwegian policy is forcing investors to price.

The government is moving from connection order to project selection

The government’s proposed licensing regime is intended to apply to existing as well as future data centres, with transitional arrangements still to be designed. The stated criteria will consider digital security, preparedness and the availability of AI computing capacity in Norway, the Nordic region and Europe. Cryptocurrency-related activity will not be treated as beneficial to society.

That signals a move beyond the traditional question of whether a connection application meets technical and planning requirements. Norway is preparing to ask what the electricity is for.

The government also plans changes that would allow Statnett to impose conditions for responsible connection, review which categories of consumption deserve capacity reserved as ordinary consumption, modernise the grid companies’ connection obligation and require major new consumers to bear a greater share of network and system costs.

For developers, the permit narrative therefore has to become an operating proposition. A site will need more than low-carbon power, cold climate and land. It may need to demonstrate the strategic value of its workloads, security arrangements, resilience contribution, grid behaviour and cost responsibility.

For customers, this could make the identity of the workload more important to site selection. Sovereign cloud, public-sector services, healthcare, emergency systems, defence-related capacity and credible European AI infrastructure may receive a more favourable policy reading than an undifferentiated high-density compute proposal. The exact hierarchy is not yet law, but the direction is clear enough to affect current negotiations.

The temporary pause changes option value now

Statnett’s temporary suspension covers new capacity reservations for data centres and new applications for approval of facility functionality under the system-responsibility rules. It also pauses treatment of received applications for facilities seeking direct connection to the 420 kV grid.

That does not mean every Norwegian project has stopped. Existing connections remain connected. Projects at other stages may sit outside parts of the measure, and site-specific outcomes will depend on the grid level, prior approvals and the maturity of each case.

But the pause changes the value of development options. Land control without an advanced power position is less valuable than it appeared a week ago. A queue position without clarity on technical conditions is less bankable. An announced capacity envelope should carry a larger discount until its connection status, delivery path and regulatory treatment are evidenced.

The practical diligence order should now be:

1. identify the exact connection case and grid level; 2. distinguish requested, queued, reserved and connected power; 3. verify whether facility-function approval is complete or paused; 4. separate total electrical capacity from IT capacity; 5. test whether the planned workload supports the emerging public-interest criteria; and 6. model the developer’s contribution to grid and system costs.

This is not administrative housekeeping. It changes capital allocation, transaction pricing and customer delivery risk.

Norway still wants data centres—but not all demand equally

The policy is not a rejection of the sector. The government describes data centres as critical infrastructure and as necessary for secure digital services, digital independence and artificial intelligence. Its objective is controlled growth rather than zero growth.

That distinction is supported by the economic evidence published by the government in late October. A commissioned Menon Economics study counted 87 active data centres in mid-2026, including 49 commercial facilities with an aggregate reported capacity of 761 MW. It estimated that the sector used 3.2 TWh of electricity in 2025, about 2.3% of national consumption, and generated approximately NOK 3.2 billion in direct value added.

The same study also underlines the policy tension. Its scenarios reach 17–37 TWh of data-centre electricity use by 2035 and NOK 150–350 billion of investment, while identifying grid capacity as the near- to medium-term constraint. A market can be economically attractive and still exceed the infrastructure available to serve every applicant.

The likely result is not simply fewer projects. It is a wider gap between projects that can prove strategic relevance and deliverability, and projects that remain speculative capacity claims.

A serious counter-argument: licensing can add another bottleneck

There is a legitimate risk that the licensing regime becomes an additional layer of uncertainty. Criteria such as “beneficial to society” can be difficult to apply consistently. Existing operators will need predictable transitional rules. Investors will need to know whether licensing decisions can be appealed, transferred with an asset or conditioned on changing workload mixes. Customers will want confidence that a future change in tenant or compute type does not reopen the whole approval.

If the framework is vague, Norway could discourage precisely the long-duration, security-sensitive infrastructure it says it wants. Capital may move toward markets where power is more expensive but the approval path is more legible.

The government can reduce that risk by publishing measurable criteria, clear timelines and a transparent relationship between the licence, grid reservation and technical approval. It should also distinguish between a campus’s maximum development envelope and the capacity actually required in each phase. Phased connections with enforceable milestones may serve the system better than all-or-nothing awards.

What the market should watch next

The main principles of the licensing regime are expected in autumn 2027, but several earlier signals will matter.

First, Statnett must define the technical conditions that allow the reservation pause to be lifted, with the end date currently set no later than 1 August 2027. Second, regulatory amendments will show how responsible connection conditions and cost contributions are applied. Third, the treatment of advanced projects will reveal whether queue maturity still creates meaningful priority. Fourth, the first licence criteria will show how Norway weighs AI capacity, security, preparedness, local value creation and grid flexibility.

Until then, the most useful market question is not “How many gigawatts are planned in Norway?” It is: which projects have evidence at each step from land and planning to power request, reservation, technical approval, connection, contract and live IT delivery?

Norway’s new approach makes that evidence hierarchy unavoidable. The winners will not necessarily be the projects with the largest announced number. They will be the ones that can prove both system compatibility and public value.

---

Sources

• Norwegian Government, “Government seeks controlled growth of data centres in Norway,” 10 November 2026
• Statnett, temporary measures for continued responsible system operation, 9 November 2026
• Norwegian Government, Menon Economics study summary, 29 October 2026

*Analysis by Matthieu Gallego, PowerlandMap. Power figures retain their original scope and maturity; no grid-request figure is treated as IT capacity or confirmed supply.*