Nordic datacenter tracker
The Nordic countries are currently attracting substantial interest from datacenter developers due to cheap renewable electricity, climate conditions and land availability. This page tracks publicly reported plans for new datacenter projects in Sweden, Norway and Finland. The tracker compiles publicly reported project information from press releases, media reports, permitting documents and company material.
The methodology is still under development. Corrections, missing projects, better source material and methodological comments are very welcome, please send them to fornborg@kth.se.
Scenario map
Totals cover projects with reported capacity. Coverage varies by country.
Scenario explainer What the four scenarios include and how to read them
| Scenario | What it includes |
|---|---|
| Low 2030 | Plans with a confirmed investment decision and an announced timetable. |
| Stated 2030 / 2035 | Plans with an announced timetable. Larger sites may also be included when a phase has a timetable and is under construction or has a confirmed investment decision. |
| High 2035 | All announced capacity, including grid requests and plans without permits, assumed fully built by 2035. |
Reading the figures. Ranges reflect uncertainty in capacity and timing. An asterisk indicates missing MW data. The method below explains the calculations.
Using the map. Blue shading shows the upper estimate. Points open project details. The project filters also apply to the map and chart.
HQ. Groups projects by developers’ global headquarters: Domestic is the project’s country, Other Europe is the rest of Europe, and Outside Europe is elsewhere. Partnerships can match several groups, with each project counted once per selection. A dash marks countries without HQ reviews. Developer and ownership details are in the project table.
New capacity, year by year
Data & assumptions
* Projects without assigned MW are not included in the quantified totals.
Figures show capacity added from 2026 onwards, including cooling and other site electricity use. Cumulative totals include additions already completed during this period. Added/year shows the increase in each year.
Lines show the upper estimates and shaded bands show the range. The scale stays fixed when you change the scenario or filters. Annual ranges should not be added together. Move over the chart or tap a year to see its values.
The scenario definitions and project filters are shared with the map. You can select a different scenario and HQ group in each view.
Loading annual capacity…
Projects can be expanded with the + sign to show multiple project phases, capacity interpretations, key assumptions, and scenario inclusion.
Country
Bidding zone
Status
Type
Scenario (summary & map)
| Project | Developer | Country | Zone | Type | Project status | Investment decision | Full build-out (MW) | Full build-out (TWh/yr) | Announced dates | Last checked |
|---|
Capacity and electricity use
Each step opens a short explanation.
Parameters and assumptions
MW denotes installed site capacity, including cooling and auxiliary loads. Full build-out covers the announced total after deductions for existing capacity and overlaps. Scenario values cover the capacity installed by the selected year.
Capacity conversion
PUE and peak-to-capacity ratios draw on EPRI (2026). The ranges are sensitivity cases. Applying the inverse peak ratio to a project's reported peak is the tracker's capacity estimate.
Annual electricity use
Annual TWh = site MW × annual load factor × 8,760 / 1,000,000
The load factor expresses average electricity use relative to installed capacity or peak power. For an explicit peak figure, site MW in the equation is the original peak, converted with PUE where it refers to IT equipment. The corresponding column below supplies the factor.
Hyperscale, colocation and peak-based factors draw on EPRI's facility sample. AI/HPC adapts the training-server operating-time assumption in Shehabi et al. (2024) to a site-level factor. Colocation with AI uses an assumed load factor. Research/SMR-linked uses a placeholder aligned with AI/HPC. The factors still need calibration against Nordic operating data.
Opening dates and expansion
A construction start implies first operation after years when no opening date is available. If neither date is usable, the calculation uses the development stage of the building or phase:
- Expansion schedule
- Capacity is added in equal amounts over calendar years, including the first operating year. A reported phase schedule or completion target takes precedence and can shorten this period. A single-phase delivery enters in its opening year.
- Campus timing derived from a phase
- A campus included in Stated through a qualifying phase expands in equal annual amounts from its earlier or later estimated opening through . Known phases retain their schedules and set a minimum. Completion at the horizon is a scenario assumption.
- Grid connection dates
- In Low and Stated, a reported grid connection date sets the earliest possible opening for the planned capacity. If the connection is still requested or its status is unclear, the lower estimate assumes that this capacity is not operating by the scenario year.
The stage intervals and expansion period are model assumptions. IEA (2025) supports a two-to-three-year development period. National Grid / Regen (2025), pp. 26–30, describes phased delivery and grid constraints.
Data coverage and country comparisons
Totals cover identified projects with a reported capacity figure. Several large projects have no public MW figure and are omitted from the quantified totals. Country totals may therefore partly reflect differences in public MW coverage.
In Norway, Statnett publishes project names and MW for grid reservations and its capacity queue. Sweden's Svenska kraftnät publishes aggregate queue figures, while Finland's Fingrid treats connection applications as confidential. Planning and permit documents provide additional project-level information, though electrical capacity is not always disclosed.
How the tracker is updated
Country updates follow a common search and review process. The record keeps the sources and unresolved questions behind each project.
01Search for new capacity
A rules-based search routine works through each country using a documented codebook. It combines operator and customer names with place names, capacity terms, construction stages and grid terminology. The search covers company announcements, planning and permit records, grid information, local reporting and other public sources.
A lead moves forward when the sources connect a named datacenter operator or end customer to a specific location and a new capacity addition. The source trail and unresolved questions are retained for the next step.
02Check project identity, phases and dates
The lead is compared with existing projects, candidate records and historic names. Addresses, municipalities, land parcels, company relationships and earlier announcements help determine whether it is a new project, a new phase or another name for an existing campus.
Campus owners, operators, tenants and individual buildings are linked without merging their evidence. Each phase keeps the MW figure, date and investment decision stated for that phase. This prevents double counting and keeps a building’s dates separate from those of the wider campus.
Historic names, superseded targets and earlier capacity claims are retained with their sources.
03Record what each source actually measures
Every capacity claim is recorded with its unit, scope and qualifier. IT equipment load, total site electricity use and grid connection capacity remain distinct. The record also keeps the source's range, project status, timing and investment decision, together with the part of the campus or phase to which they apply.
Source ranges and qualifications are retained. Conflicting figures and unresolved relationships are flagged for follow-up, with the questions linked to the relevant projects and sources. The calculations use the rules described above.
04Calculate and review the scenarios
The scenario definitions apply to the map, chart and table. Results are checked for overlapping capacity, missing information and conflicts between sources before the site is updated.
New records enter as candidates. Their identity, evidence and scenario effect are reviewed before inclusion in the totals. One approved calculation feeds the data export, annual chart, summary, table and map.
05Identify actors and global headquarters
Project announcements and site-specific sources identify the developer, operator, named customer and, where relevant, property owner or project partner. Each actor is linked to the whole project, including its phases. Partnerships keep separate actors and roles.
Global headquarters are checked against company reports, official contact pages and company announcements, with other reliable sources used when needed. A local subsidiary's registered address or a regional office does not establish the group's global HQ. Registration country, headquarters, fund manager and financial owners are recorded separately. The business type describes what the actor does.
Ownership notes retain the source and date. They describe the documented ownership without assigning an owner nationality or tracing every investor. Unknown identities and HQ countries remain unknown and do not match any HQ group. A project can still match a known partner's HQ while another actor is unresolved. Expand the project for actors, roles, HQ, sources and the review date.
The reviews cover 31 Swedish, 46 Finnish and 32 Norwegian projects as of 17 October 2026. Every relevant actor has a known HQ country for 30 Swedish, 42 Finnish and 30 Norwegian projects. The remaining cases include an anonymous operator, a company with two headquarters and actors that do not publish one global HQ. HQ selection filters the map and its totals without allocating capacity between actors.