The purpose of the LOI is to assess the feasibility of developing a GPU-as-a-Service platform, where Fossefall will provide infrastructure and platform capabilities for training and running AI models, and Done.ai will test the environment and provide structured input based on its application needs and to secure compute capacity.

Under the LOI, the parties will jointly evaluate capacity planning for the coming years and provide testing of inference capabilities, enabling Done.ai to access and test GPU infrastructure and provide structured feedback. The goal is to establish a basis for a broader commercial relationship.

Done.ai has an option with Fossefall to secure adequate compute in line with its business objectives.

Strategic Partnership

As demand for AI compute capacity grows and access tightens across Europe, securing reliable infrastructure is becoming a structural competitive advantage. By partnering early with Nordic infrastructure builders, Done.ai aims to de-risk long-term capacity planning and improve unit economics as it scales its AI workloads.

Fossefall’s infrastructure is designed around sovereign compute principles, with all data processed and stored within Nordic jurisdiction in compliance with applicable Norwegian and Swedish regulatory standards. For Done.ai, this means customers receive demonstrable data residency guarantees, a critical requirement for adoption in regulated segments.

Compute capacity

Access to renewable-powered Nordic compute at scale enables Done.ai to deliver AI-powered capabilities to its SME customers at a cost that would not be achievable through traditional hyperscaler routes. By combining clean Nordic energy with sovereign data infrastructure, the model supports both lower-carbon AI workloads and strong data governance standards. Making AI both affordable and responsibly powered becomes a genuine differentiator in the SME segment, enabling smaller businesses to access advanced technology while supporting a more sustainable digital economy.

In the emerging AI economy, access to reliable compute capacity is becoming a structural competitive advantage. Just as cloud infrastructure defined the previous generation of software platforms, sovereign AI infrastructure will underpin the next. By partnering early with Nordic infrastructure builders such as Fossefall, Done.ai is positioning itself alongside the compute layer that will power the next generation of AI-driven enterprise software.

Strategic ownership investment

Done.ai has a minority investment in Fossefall representing ownership of ca. 2.4 percent of the shares in the company. The shareholding reflects Done.ai’s long-term strategic interest in contributing to and benefiting from the development of Nordic AI infrastructure, while maintaining an arm’s-length, commercially structured relationship between the parties.

Management comments

“Compute capacity is constrained, and those who partner early with the right infrastructure builders will have an advantage that is difficult to replicate later. For Done.ai, this collaboration enables improved unit economics, keeps our customers’ data in the Nordics under full regulatory alignment, and leverages renewable-powered infrastructure to support more sustainable AI workloads. This positions us to offer AI capabilities at attractive price points as we scale” – Staffan Herbst, CEO of Done.ai

“We are seeing increasing interest from companies that want reliable, high-performance compute capacity combined with strong data governance and sustainable power. A collaboration with Done.ai marks an important step as we realise our strategy to meeting the accelerating demand for sovereign AI infrastructure in the Nordic region,” says Øyvind L. Vesterdal, CEO of Fossefall.

About Fossefall
Fossefall AS is a Nordic AI and digital infrastructure platform focused on building sovereign, high-performance “AI factories” powered by clean Nordic energy. The company develops, owns and operates secure, scalable infrastructure designed for advanced AI workloads, serving enterprises, governments and other mission-critical users with strict requirements for performance, security and data sovereignty.

As part of its long-term strategy, Fossefall is pursuing an ambition to establish up to 500 MW of clean AI infrastructure capacity across Norway and Sweden by 2030, supported by surplus renewable energy, including hydroelectric power, and a pipeline of strategically selected sites. The company currently has at least 11 AI factory sites under development, with its first projects expected to be operational by the end of 2026, reflecting its goal to rapidly scale energy-efficient and locally anchored AI compute capacity in the Nordics.