Finland's draft grid rules would rank small connections ahead of large data centers

Finland's government has drafted an amendment to the Electricity Market Act that would retire the country's first-come, first-served grid connection system and replace it with four priority groups, reordering who stands where in the queue and moving large data centers behind small connections, as Data Center Dynamics first reported. The stated aim is to motivate high electricity consumers to take measures that lower electricity prices and reduce the risk of shortages, and to encourage large consumers to shift into less congested parts of the network.

The draft assigns priority by group rather than by date of application, with the first group reserved for small-scale connectors — sites with a power intake capacity of no more than 3MW — and for critical infrastructure, electricity storage facilities with capacities of 100kVA or less, and storage connected to power plants not charged from the network. Data centers are excluded from that segment unless rated at 1MW or below, and for these projects the old rule survives, with connections issued on a first-come, first-served basis all year round.

The composition of that top group hints at the draft's values: priority goes to small-scale connectors, critical infrastructure, and storage, including storage tied to power plants that draw nothing from the grid, while data centers are the one class named and then fenced off, admitted only at 1MW or below. The ladder is built around what a project contributes to the network's balance.

The second group is where the draft does its most consequential work, holding data centers of more than 1MW and no more than 10MW and setting out a production commitment whose terms are local generation and flexibility obligations: 50% of the site's estimated annual electricity consumption covered by local production, a local power plant commissioned no more than 36 months before the data center and sited in the same area, and flexible production capacity offered on an organized marketplace, as electricity production or demand-side flexibility, continuously and without interruption, at an output of at least 10% of the site's peak. The same passage excludes from the second group any facility holding electricity production capacity under a flexibility agreement of at least 80% of peak output, though which group such a site would occupy is not stated.

Above 10MW the terms soften in an unexpected direction: data centers larger than that enter the third priority group without the production commitment, while a fourth group absorbs everything else, and the second, third, and fourth groups run through a six-month approval process with first allocation scheduled for December 2027. The amendment is set to come into force on January 1, 2027, and would apply to connection requests submitted before its enactment where no agreements have been reached.

Read together, the four groups do something more specific than the framing of data centers being dropped down the queue conveys: a data center above 10MW is not placed last but takes the third of four rungs, above the catch-all that collects all other projects. The change is to the reference point, because first-come, first-served rewarded the developer who arrived earliest, while the grouping rewards the one that is small — a data center of 1MW or less keeps the old first-come treatment inside the top group — or that accepts generation and flexibility obligations, so scale, once neutral, now works against a project. The fourth group collects all other projects, meaning a data center above 10MW still ranks ahead of every applicant that fits no named class, and what the draft does to large sites is demote them relative to small connections and small data centers, not send them to the back of the line.

The design also places the heaviest obligations where the load is smallest: the production commitment appears in the second group, among sites of 1MW to 10MW, while the largest consumers above 10MW are moved down a rung but are not asked to build generation, so if the intent is to reduce strain on the network, the draft asks the middle band to do the work and the biggest sites mainly to wait — and, on the government's account, to consider less congested locations.

Interconnection policy, seen this way, becomes load-class policy: the draft works by classification, sorting applicants into classes and attaching different obligations to each, which makes the binding constraint on a project its class, and a rule book that ranks by class is also one that can be re-ranked when the classes change — which is why the timing provisions carry more weight than the tiers themselves.

That middle band is where the draft's terms are least resolved: a site of 1MW to 10MW sits in the second group, and the carve-out for facilities holding a flexibility agreement points to a class of project that pairs its load with local production, though the destination of that class is not stated. Set the ambiguity aside and the direction is clear enough — the draft is trying to make local generation and demand flexibility the price of a better position, but whether a shell at the low end of the band can be financed on terms that include a co-located plant, a 36-month commissioning window, and a continuous offer of flexible output is a question that will be settled project by project.

The six-month windows add a second kind of delay on top of the ranking itself: projects in the second, third, and fourth groups do not receive connections by arriving, but only when an allocation period opens, and the first such period is set for December 2027, so the practical timetable for a large data center in Finland runs not from the date it files but from the date the state next decides who is served.

The provision with the most immediate commercial weight concerns timing: because the amendment would apply to connection requests submitted before enactment if no agreements have been reached, it puts the value of an unpapered queue position in question, and this publication has argued that grid access and queue position are becoming a form of deal currency, optioned ahead of offtake because the queue is the binding constraint. Finland's draft tests who actually holds that currency, since a place in line is an asset only if the body that issues it cannot reshuffle it, and the agreement condition suggests the positions most exposed to reallocation are the ones not yet converted into contracts — developers who treated their arrival date as the entitlement have a reason to sign before January 2027.

For anyone underwriting Finnish data center development, the practical effect is a change in what a queue position proves — before the amendment, a filed application at an early date was evidence of priority, while afterward that evidence rests on an agreement, and the report does not state what form of agreement would insulate a request from the new ranking. Until that is settled, a project's standing in the Finnish queue is likely to be read on the strength of its paperwork and not only the date it was filed.

Geography is the other lever, because the government's aim includes steering large consumers into less congested areas, turning queue priority into a siting instrument that sits alongside the planning process, and consent, not capital, has long been seen as the gate for data center development. Finland's draft broadens what consent covers to include the network operator's willingness to allocate capacity on the applicant's terms, not only the local planning decision, so where the two gates align a site advances, and where they do not a developer can hold a signed lease and still have no path to power.

There is also a hole in the calendar: the amendment would bite on January 1, 2027, while the first allocation under the new groups is not scheduled until December 2027, leaving nearly a year in which the ranking binds but the process that ranks has not yet run, and the coverage does not say how requests arriving in that window would be handled — for a developer deciding whether to file early or wait for clarity, the draft offers no guidance either way.

Set against the wider buildout, the proposal is a concrete instance of an argument the market is already having: data center capital has been pricing the anchor tenant and the megawatt, and Finland's draft inserts a variable between a site and its power in the form of defined obligations that determine where the site stands in line. Projects that can carry those obligations start to look less like pure load and more like co-developed infrastructure, with a generation asset beside them and a flexibility line attached, while projects that cannot may wait, relocate, or — for the smaller sites — find an advantage in staying under the 1MW threshold that still buys first-come treatment.

Nothing about the proposal is settled — it is an amendment in draft, and the coverage gives no reading on how much of it will become law — but it establishes where the pressure will fall: on the middle of the size range, where the generation and flexibility obligations attach, rather than at the top, where the draft's instruction is to wait or to move. The first allocation under the new groups is scheduled for December 2027, and by then the applications themselves will show which behavior the draft produces: local generation paired with data centers, relocation toward less congested areas, or projects clustering below the 1MW line to keep the queue they know.