Local impact

Reading order

Read local impact as six files, not as one warmth. Land: what the detaljplan allows and what else could have used the plot. Power: what the network owner has written, not what a community brochure hopes. Water and noise: what investigations exist. Jobs: construction versus permanent, company estimate versus observed payroll. Heat: designed versus delivered. Write scenario or observed on every line before you let the sentences sit in the same paragraph.

Keep the welcome letter and the decision paper apart. A municipality can want the tax base and still owe neighbours a noise and traffic file. A company can fund a local project and still not have published megawatts. Walk municipal planning and environmental pages after you have named the effect, not before. If the effect is spare grid capacity, stop. That cell belongs to the network owner and to Svenska kraftnät’s formal process, not to this page.

• Construction jobs versus permanent roles.
• Scenario versus observed.
• Company community spend versus municipal budget impact.
• Heat designed versus heat delivered.

How local effects enter Swedish decisions

Land and neighbours sit first with the municipality. Boverket’s Planning and Building Act guidance explains detaljplan, building permit, technical consultation and start notice as different papers. A comprehensive plan is not a building right. Outdoor coolers, substations and construction traffic are part of what neighbours will see. Inspektionen för strategiska produkter may screen certain investments; that screening is not a local-impact assessment and not a substitute for PBL.

Power impact is a network-owner file. A hall can raise local load, trigger reinforcement, or sit waiting. Svenska kraftnät’s indicative transmission map does not show underlying-network capacity and does not replace the formal connection process. Bidding areas SE1 Luleå, SE2 Sundsvall, SE3 Stockholm and SE4 Malmö explain wholesale-price differences. They do not tell Gävle, Falun, Falkenberg or Luleå that spare megawatts exist on a named street. A national energy mix is not a local spare-MW gift and not a jobs argument.

Water, noise and heat sit with environmental and utility counterparties. Naturvårdsverket’s Environmental Code guidance is the starting point when an investigation is triggered. Heat that is only designed does not warm apartments. Heat that GleSYS said had started to flow to Falkenberg Energi on 1 October 2020 is a dated local effect with unpublished later volumes. atNorth’s and Conapto’s Exergi wording is a Stockholm-area mention, not a measured household benefit. EcoDataCenter’s Falun recovery wording with Falu Energi & Vatten is likewise a named-counterparty sentence, not a town heat total.

Jobs and community spend are usually operator-published. Construction employment is time-limited. Permanent operating crews are smaller and are rarely a municipal labour-market programme by themselves. Company-funded local projects can be real and still not appear in the municipal budget as net new revenue. Microsoft’s community materials are the dated example in this set: local projects around named municipalities, employment figures as company estimates. They are not a Statistics Sweden census and not a disclosure of facility megawatts.

What named primary sources show

Microsoft said the Sweden region opened on 16 November 2021 and named Gävle, Sandviken and Staffanstorp. Facility megawatts are not in the cited launch materials. The later community PDF discusses local projects and employment figures as company estimates around those municipalities. Read it as dated company community material. It does not fill the MW gap and it does not convert construction narrative into a permanent payroll census.

GleSYS documented a local heat effect: delivery to Falkenberg Energi starting 1 October 2020, with 3 MW utility power in that release. The later operator page states 5 MW and continues to describe district-heat recovery. Neighbours in Falkenberg therefore have a dated heat-start claim and an unresolved public explanation of the megawatt difference. Annual delivered heat after 2020 is not on those pages. That is a local-impact gap, not a licence to invent a town total.

Meta’s Luleå campus is the long-running northern example: operator-described as operational from 2013, outdoor-air cooling, local hydropower, operator PUE 1.09, current MW not on the cited page. The existence of the campus is documented. Grid-agreement documents and a current operating-crew figure are not in the public sources used here. atNorth SWE01 in Kista states 12 MW site power and heat recovery with Stockholm Exergi without occupied IT load or delivered-heat volume. Conapto’s 5 September 2024 opening of Stockholm 4 South states a 24 MW campus with Stockholm 2 South and mentions Exergi in related wording.

EcoDataCenter’s Falun campus is a local expansion story with several electricity figures that must not be added: 41 MW after C and D, 65 MW once E and F are operational, 80 MW available power on the marketing page. Buildings E and F were under construction in the 2025 annual report; commissioning dates there are targets. Recovery with Falu Energi & Vatten is described; delivered heat is not quantified. evroc’s Arlandastad notice is a land and building-rights purchase with a GPU design envelope, not a documented construction peak or operating crew.

Common misreads

The first misread is to use local-impact prose as a grid-capacity proof. A community PDF, a hydropower sentence, or a municipal welcome does not reserve megawatts at a cadastral unit. The second is to treat construction-job estimates as permanent local employment. The third is to treat company community spend as a municipal budget line. Microsoft’s figures stay company estimates, dated, not a census.

The fourth misread is to treat designed heat as a warmer town. Exergi and Falu wording without a delivery date or a volume is not an observed household effect. The fifth is to add Falun’s megawatt sentences or GleSYS’s 3 MW and 5 MW into an impact score. The sixth is to read SE1 as a local blessing and SE4 as a local burden. Bidding areas price transmission constraints. They do not grade municipalities.

• Inventing spare grid capacity from a community paragraph.
• Equating construction jobs with permanent roles.
• Treating company estimates as a census.
• Counting designed heat as an observed local benefit.
• Using bidding areas as a municipal score.

What to ask next

Ask the municipality which detaljplan applies, what use it allows, and which noise, traffic and stormwater investigations exist. Ask the network owner for the latest written position and forbid yourself from answering that question with a sustainability brochure. Ask the heat company whether anything is contracted or delivered. Ask the operator to split construction and permanent roles, and to date each figure.

Then open municipal-engagement and noise pages for process, the heat-reuse page for commercial heat states, and the power pages for connection evidence. If Microsoft-style community numbers appear, label them company estimates. If GleSYS-style heat starts appear, ask for the next measured year. If EcoDataCenter-style campus totals appear, keep the rows un-added. Empty cells stay empty, and a missing operating-crew figure stays missing.

• Which PBL paper applies, and what use does it allow?
• What has the network owner written — not hoped?
• Construction versus permanent roles, each dated.
• Heat state: prepared, contracted, or delivered?
• Which effect is still a scenario, not an observation?

What this does and does not prove

This page proves that local-impact claims in the reviewed set must be split by land, power, water, noise, jobs and heat, and by scenario versus observed. It proves that Microsoft published dated community estimates without facility megawatts, that GleSYS published a 2020 heat-delivery start, that Meta’s Luleå campus is documented without a current MW on the cited page, and that Falun’s expansion figures are separate electricity rows.

It does not prove a net local-welfare ranking of Swedish municipalities. It does not publish a national jobs total or a national heat-market total. It does not treat operator PUE figures as local-benefit scores. Those PUE figures remain operator numbers from different years and methods. It does not invent spare grid capacity, unpublished operating crews, or live traffic counts. It does not turn evroc’s GPU design into a construction-employment forecast.

Where the source is silent, the impact is unknown. Point to the municipal paper, the network-owner letter, or the named operator gap. A later primary source may add an observed effect. Until it does, the welcome sentence and the measured effect remain different documents, and a community estimate stays an estimate rather than a census of local jobs or heat.