Summary
- Pure DC has launched phase one of SJK01, a 110MW AI data centre campus in Seinäjoki with the first phase fully leased.
- The full Finnish campus is planned to scale beyond 550MW of IT capacity, supported by more than 700MVA of renewable power.
- The design uses repeatable 40MW AI-ready modules, direct liquid cooling, closed-loop cooling systems, and waste heat recovery for district heating.
Pure DC has launched the first 110MW phase of a planned AI data centre campus in Seinäjoki, Finland, setting out a development that could eventually scale beyond 550MW of IT capacity.
The first phase, known as SJK01, is already fully leased and has secured planning permissions and power requirements, according to the company. Pure DC says the substation for the first data hall has been constructed and is live, while the initial phase is expected to involve more than €1.5bn of investment.
Across its full planned scale, the campus would represent more than €7.5bn of investment and sit across 370 acres in western Finland. The site is being built for mass-scale AI and machine learning workloads, with a design based around repeatable 40MW AI-ready modules.
Power and cooling set the campus shape
Pure DC says the development has access to more than 700MVA of renewable power, a claim that gives the project a different starting point from AI campus plans still waiting on grid certainty. Large AI facilities are increasingly being judged by the firmness and timing of their electricity supply rather than by headline capacity alone, particularly as developers move from conventional data hall densities towards GPU-heavy infrastructure.
The cooling design is equally central. The campus is being developed around modules that can support direct liquid cooling, with data halls able to use either liquid-cooled or air-cooled approaches depending on customer requirements. Pure DC says both options are based on closed-loop systems that use no water during operation.
That combination of direct liquid cooling and closed-loop operation reflects a hardening design problem across AI infrastructure. Higher-density compute changes the balance of mechanical and electrical systems, raises heat rejection demands, and makes future hardware evolution harder to accommodate. By building around repeatable modules, Pure DC is trying to create a platform that can scale without redesigning each phase from scratch.
The company also plans to use waste heat recovery for district heating. In the Nordics, that approach has stronger practical foundations than in many other European markets because municipal heat networks provide a clearer route for exporting low-grade heat. Even so, the commercial test remains whether the heat can be captured, upgraded, contracted, and delivered at a useful temperature and volume outside the site boundary.
The Nordic capacity route
Seinäjoki extends a familiar Nordic proposition for digital infrastructure: large plots, cooler ambient conditions, lower-carbon electricity, and political appetite for inward investment. AI campuses increase the demands on that model. They require larger blocks of firm power, heavier substation coordination, stronger fibre planning, and a deeper construction and operations supply chain than many earlier cloud or colocation deployments.
Pure DC says the programme will create more than 3,000 jobs over a decade-long construction period, with local companies involved where possible. The company is also working with Seinäjoki University of Applied Sciences and vocational institutions on education and reskilling. Those commitments will sit alongside the practical delivery pressures of a campus that needs transformers, switchgear, generators, cooling plant, specialist labour, and long-lead electrical equipment across multiple phases.
The project also gives a UK-headquartered data centre developer a major AI platform in Finland. That is part of a wider European pattern in which capital and development expertise from outside the Nordics is being deployed into markets where power and cooling assumptions are more favourable than in Frankfurt, Dublin, Amsterdam, or London.
Later phases will depend on whether secured power, customer demand, permitting, and construction capacity remain aligned. SJK01 gives Pure DC an unusually large first step: a fully leased, planned, and powered phase in a market where AI infrastructure announcements often run ahead of grid reality.
