Industrial Grid Connection: Flexibility Quota and Auction for Inflexible Loads

The Problem

Fingrid's transmission grid in Southern Finland is approaching saturation. New connections above 10 MW have been restricted in multiple areas until early 2027. The restriction reflects both current load growth and the pipeline of committed data centre investments — 17 projects with investment decisions taken, aggregate capacity potentially exceeding 2,200 MW by 2030. The restriction is not arbitrary: it operates within the flow-based market coupling framework adopted by all Nordic TSOs on 29 October 2024 (day-ahead markets) and extended to long-term capacity on 31 October 2025. Available connection capacity per region is now calculated through this harmonised Nordic methodology, which considers the full Nordic grid's physical constraints rather than simple bilateral NTC values.

The connection queue does not distinguish between load types. A 500 MW data centre and a 50 MW adjustable electrolyser queue under the same rules. This is a governance failure: the allocation mechanism does not price flexibility, and no institution guides applicants toward the system-valuable mix.

Inflexible loads — data centres running 24/7 flat profiles — occupy transmission capacity permanently. They do not offer frequency reserves, do not shift consumption in response to price signals, and do not support the grid's balancing needs during high-stress periods. Their net fiscal contribution is, according to Verohallinto Director General Heikura (May 2026), presently negative for the largest operators.

Flexible loads — industrial processes that can shift timing, electrolysers that absorb surplus, SGFA-type nodes that simultaneously produce district heat and grid reserves, CHP replacements — deliver system value beyond their electricity consumption. They reduce the compound risk described in TN-009: the coincidence of cold, still, dry conditions with declining domestic firm capacity.

Reframing: consumer to grid participant

SM-011 is not a restriction on data centres. It is a redefinition of what a large grid connection means. The optimisation function changes:

Current: fastest applicant → capacity winner
SM-011: highest resilience contribution → capacity winner

This is not a novel institutional form. Finnish society already accepts that shared infrastructure — regulated waterways, ports, rail networks, aviation slots — cannot be allocated on pure speed-of-application. Mandatory minimum flow releases on regulated rivers are not negotiated each spring; they are encoded in water permits and enforced automatically. The Vuoksi minimum flow of 300 m³/s (May 2026) is a legal constraint that operates regardless of commercial interest.

SM-011 applies the same logic to the electricity grid. A large continuous load is not a passive consumer of shared infrastructure. It is a participant in system stability. Its connection terms should reflect that participation — not through voluntary commitments or press releases, but through a rule encoded before the situation arises, not negotiated when it has already arrived. Framed as grid modernisation, SM-011 extends to the connection allocation layer a transition already underway through frequency reserves, demand response, and priority load management.

Two Design Questions

Question 1: Defining "inflexible"

The flexible connection agreement framework (Sähkömarkkinalaki §20b–20c, Energiavirasto rules effective 1 February 2026) already defines two categories: temporarily flexible (pending grid development) and permanently flexible (with Energiavirasto permit). An inflexible load is one that declines to enter either category.

No new definition is needed. The mechanism inverts the existing rule: flexible connection agreement = Track 1. No flexible connection agreement = Track 2. This is implementable through Fingrid's existing connection terms without legislative change.

Question 2: Quota size

The quota must be calibrated to avoid two failure modes:

• Too small: Data centres relocate to Sweden or Norway, where connection is easier. Finland loses investment pipeline. The correct policy response would then be a different instrument.
• Too large: Inflexible loads continue to displace flexible ones. The Heikura problem persists: inflexible data centre load consumes connection capacity that should have been reserved for flexible industrial loads and reserve-providing nodes, increasing transmission congestion and reducing system resilience — while delivering, in the words of Verohallinto DG Heikura (May 2026), presently negative net fiscal integration.

Proposed approach: the quota is dynamic and regional, calibrated annually by Fingrid against three inputs:

1. Transmission utilisation rate (WEM §12, SE1→FI TRR). When spare transmission capacity exists, the quota can be larger. When TRR is elevated, the quota tightens.
2. Flexible load queue. The quota remains small until all queued flexible applicants have been accommodated. Flexible loads are not waiting while inflexible loads connect.
3. Committed grid expansion timeline (Fingrid 2026–2035 main grid plan). As new capacity is confirmed and dated, the quota can grow in anticipation.

Initial calibration suggestion: 10–20% of new connection capacity per region allocated to inflexible quota. Review annually. Provide data centres with 3–5 year forward visibility on quota trajectory to support their investment planning.

Regulatory Basis

SM-011 does not require new legislation. The flexible connection agreement infrastructure is in place. The remaining elements — quota, auction, and proceeds allocation — are within Fingrid's operational and governance authority. TEM can reinforce through a non-binding recommendation without initiating a legislative process.

Recommendations

1. Fingrid board decision. Fingrid adopts a connection allocation policy that: (a) confirms flexible connection agreement commitment as the criterion for Track 1 priority; (b) establishes an annual inflexible quota per region; (c) implements a sealed-bid or ascending auction within the quota; (d) directs auction proceeds to a dedicated grid investment fund.

2. TEM guidance. TEM issues a recommendation supporting Fingrid's policy, signalling that inflexibility pricing is consistent with national energy policy objectives. No legislation required.

3. Transparency. Fingrid publishes annually: the quota by region, auction clearing prices, queue lengths by load category, and proceeds directed to investment. This creates market-facing accountability and allows data centre investors to plan.

4. Pilot year. Implement in one region (Southern Finland) for 2027, then evaluate before national rollout.

Yhteydet muihin analyyseihin

Liittymäkapasiteetin kohdentamisongelma on tunnistettu laajemmin: joustamattomat kuormat syrjäyttävät systemaattisesti joustavia liittyjiä, koska nykyinen prosessi ei hinnoittele joustavuutta. Tämä analyysi ehdottaa konkreettisen institutionaalisen vastauksen olemassa olevan sääntelyn varassa.

Säädettävät ja useita palveluja tuottavat kuormat — kaukolämpöä tuottava CHP-retrofit yhdistettynä reservimarkkinaosallistumiseen tai sähköpolttoainetuotantoon — ovat juuri se kapasiteettikerros, jonka liittyminen tuottaa sähköjärjestelmälle eniten arvoa. Nämä eivät saa joutua kilpailemaan 24/7-tasaisesta datakeskuskuormasta.

Sähköjärjestelmän kriittisin haavoittuvuus on olosuhteiden yhteisvaikutus: kylmä, tyyni ja kuiva sää yhtä aikaa laskevan kotimaisen varmatuotantokapasiteetin kanssa. Joustavuuskiintiö suojaa juuri ne kuormat, jotka kykenevät reagoimaan tällaisiin tilanteisiin.

SE1→FI-siirtoyhteyden reaaliaikainen käyttöaste (TRR) on luonteva dynaamisen kiintiön kalibrointiparametri: kun siirtokapasiteetti lähestyy saturaatiota, joustamaton kiintiö tiukkenee automaattisesti.

Track 1 -kelpoisuuden arvioinnissa voidaan hyödyntää mitattavia joustavuusindikaattoreita: reservimarkkinaosallistuminen, kysyntäjousto hintasignaaleihin ja hukkalämmön hyödyntäminen ovat konkreettisia kriteerejä, jotka Fingrid voi sisällyttää liittymissopimuksen ehtoihin.