Sesterce Targets Finland With $11 Billion AI Buildout

French AI infrastructure company Sesterce plans to invest more than €10 billion, or about $11.2 billion, in Finland. The company will develop a large AI campus at the former Kaipola paper mill site in Jämsä. The project will start with 200 MW of capacity and later expand to 600 MW. Sesterce ultimately wants to build more than 1 GW of AI capacity across Finland. The first phase is scheduled to begin in 2026, with site clearance and interior work already underway. Sesterce says the project has an anchor customer for its initial development. The company has not publicly named that customer. CEO Youssef El Manssouri told Finnish broadcaster Yle that it is one of two leading AI laboratories.

The choice of Jämsä reflects a broader shift in AI infrastructure development. Sesterce selected an industrial site with existing power, water, buildings and logistics infrastructure. That approach can reduce the need for entirely new infrastructure. It can also shorten the path from site selection to usable compute capacity. “For more than six decades, the Kaipola mill produced paper for the world. Now, a new generation of industry is being planned for the site. Our goal is to preserve the area’s industrial identity while making use of existing infrastructure, including the power grid, water systems, buildings and logistics connections.”

A 200 MW Start With a Path Toward 600 MW

The initial 200 MW phase will establish the campus’s first major computing block. Sesterce plans to expand that capacity to 600 MW during the second phase. Its longer-term Finnish ambition goes beyond 1 GW of AI capacity. That scale puts electricity at the center of the project’s development strategy. The Kaipola site connects to Elenia’s high-voltage electricity network. That connection provides the foundation for the project’s first phase. Jorma Myllymäki, CEO of Elenia, said: “A functioning, reliable and resilient electricity network is a key enabler of vitality, economic growth, and large-scale industrial investments. The Kaipola industrial site is connected to Elenia’s high-voltage electricity network, which currently provides a solid foundation for implementing the first phase of the project.”

A grid connection does not solve every power challenge for a large AI campus. Sesterce will still need enough generation, transmission capacity and operational flexibility as the site grows. The jump from 200 MW to 600 MW will create additional requirements across the power system. Each expansion stage will need matching infrastructure beyond the data halls. Sesterce also plans a major renewable energy commitment around the campus. The company says it will enable 1 MW of newly built renewable generation for every 1 MW consumed by the campus. The commitment will apply within 10 years of the campus becoming operational. Sesterce says independent third parties will audit and publicly report the results each year.

Liquid Cooling Moves Into the Core Design

The campus will use a closed-loop liquid cooling system for its AI infrastructure. Sesterce also plans to use harvested rainwater and recycled water. The company says it will not use drinking water for cooling. That approach addresses a growing infrastructure issue as AI racks reach much higher thermal densities.

“Our planning is based on energy-efficient and flexible operations. The Sesterce Campus projects will use liquid cooling, deliver high energy efficiency, and feature controllable and flexible power consumption. Opportunities for utilizing waste heat are also being considered as part of the design.”

The focus on flexible power consumption carries strategic importance. Large AI campuses can create major and continuous electricity demand. Flexible workloads can give operators more options during periods of grid stress or changing power conditions. That capability could become an important part of future AI capacity contracts. Cooling also changes how customers should evaluate AI capacity. A customer may contract for a large amount of compute without understanding the thermal limits behind it. Liquid cooling can support high-density deployments, but the cooling system still needs enough flow, heat rejection and operational headroom. At large AI campuses, cooling capacity can become as important as accelerator capacity.

Sesterce is also considering ways to reuse the heat generated by the campus. The company has not disclosed a specific heat reuse project for Jämsä. Finland already has infrastructure and district heating systems that can support heat recovery projects. A large AI campus could therefore become part of a wider local energy system.

Former Industrial Land Becomes AI Infrastructure

The redevelopment of Kaipola reflects another major trend in European AI infrastructure. Sesterce is not starting with an undeveloped site. It is converting a former industrial property into a new computing hub. The paper mill closed in 2021, leaving behind infrastructure that can support another energy-intensive industry.

“Many supercomputer projects destroy a lot of nature. Finland has excellent industrial sites that we can convert into supercomputer centres. That guided our choice,” El Manssouri said.

The industrial history of Kaipola gives the project a different development profile. Existing roads, buildings and utility infrastructure can reduce some construction requirements. The site also gives Sesterce access to an established industrial location rather than a greenfield development. That distinction matters as European communities face growing pressure to manage land use around AI infrastructure.

The economic impact could extend beyond construction. Sesterce expects the project to create about 2,000 construction jobs. The company also expects around 300 permanent positions once the campus begins operations. Sesterce says it plans to work mainly with Finnish contractors and subcontractors during construction.

Jori Reijula, mayor of Jämsä, has welcomed the potential economic impact while stressing the need for the project to meet local requirements. “Jämsä needs new investments, business activity and jobs. The Sesterce project is an interesting opportunity to strengthen the vitality of the region, provided it progresses from plans to implementation and meets the applicable requirements. The redevelopment of the Kaipola site would be a welcome opportunity to give the area a new lease on life and support job creation in Jämsä. The City wants to advance investment projects in cooperation with businesses and will assess each project carefully as it moves forward.”

Sesterce Links AI Capacity to Local Commitments

Sesterce plans to create a €10 million community fund when the Jämsä campus begins operations. The City of Jämsä and local residents will help select projects for the fund. The company also plans public facilities that could include restaurants and education spaces. Those facilities will remain separate from the secured AI infrastructure. The commitments give the project a broader local dimension. Sesterce is positioning the campus as an AI facility and an industrial redevelopment project. It also connects the campus to local employment and infrastructure development. That model could influence how other European communities evaluate large AI projects.

For AI customers, the project highlights a larger capacity challenge. Compute capacity depends on much more than the number of GPUs installed at a site. Power availability, cooling performance, network connectivity and operational flexibility all influence usable capacity. A 200 MW campus therefore represents a complex infrastructure system rather than a simple measure of compute supply.

The planned expansion makes that issue more important. Moving from 200 MW to 600 MW requires more power and cooling infrastructure. It also requires additional transmission capacity and operational coordination. Customers will need confidence that each layer can scale alongside the computing environment.

Finland’s AI Infrastructure Race Accelerates

Sesterce’s announcement comes as Finland attracts major AI infrastructure investments. Google has also announced plans for significant investment in Finnish AI infrastructure. The country offers a combination of industrial sites, renewable electricity and a cool climate. Those factors have helped make Finland increasingly attractive for energy-intensive computing. The growing pipeline creates a new challenge for the country. Finland must match data center expansion with generation, transmission and grid capacity. Large AI campuses can increase electricity demand quickly. Their development can therefore affect both infrastructure planning and the economics of power supply.

Sesterce’s renewable commitment gives its project another strategic dimension. The company wants new renewable generation to grow alongside its consumption. That approach links AI expansion with additional electricity production rather than treating renewable supply as a separate issue. It also gives customers a clearer framework for evaluating the energy strategy behind future compute capacity.

The Jämsä project also demonstrates why existing industrial sites are becoming strategically valuable. AI infrastructure developers need locations that can support power, cooling, connectivity and construction at the same time. Former industrial facilities can offer some of those assets before development begins. That advantage could become more important as suitable greenfield sites become harder to secure.

The Bigger AI Infrastructure Question

Sesterce’s Finnish strategy goes beyond a $11 billion investment announcement. The company is combining industrial redevelopment with high-density AI computing and large-scale power infrastructure. It is also building liquid cooling and renewable generation into the broader project strategy. That combination reflects how AI infrastructure is becoming increasingly dependent on systems outside the traditional data hall. For end users, the important question is not simply how many megawatts Sesterce can build. Customers need to know how much compute those megawatts can support consistently. They also need visibility into cooling headroom, power flexibility, network capacity and expansion plans. Those factors will determine whether announced capacity becomes dependable AI capacity.

The Jämsä project could become an important test for Europe’s next phase of AI infrastructure growth. Its success will depend on more than construction progress. Sesterce must coordinate power, cooling, renewable generation, industrial redevelopment and customer demand at very large scale. If those pieces move together, Finland could strengthen its position as one of Europe’s most important locations for AI computing.

The project also points to a wider shift in infrastructure strategy. AI developers can no longer treat power, cooling, land and compute as separate planning exercises. Every additional megawatt creates requirements across several connected systems. For AI buyers, that means the real value of future capacity will depend on how well those systems work together, not simply on the headline size of the campus.