Castellan does not start from zero. It is built on the operating DNA of the Bitkern Group — a decade of developing and running energy-intensive compute infrastructure across Switzerland, Austria, and the United States — and it inherits the one asset that decides every AI project before the first rack arrives: power.
The Bitkern Group has spent a decade in the demanding end of the compute business — crypto-mining and high-performance data centres: securing power, building high-density facilities, and operating them commercially, around the clock, across Switzerland, Austria, and the United States. It is the same operational school that produced several of the leading AI-infrastructure operators of this cycle.
Over that decade the group has operated more than 300 MW of energy-intensive compute capacity — negotiating multi-year power supply, engineering high-density electrical and cooling systems, and running facilities where every hour of downtime carries a price. That is a different education from the real-estate playbook most data-centre entrants arrive with. It takes nothing on faith: what breaks at scale, what a grid operator will actually commit to, and what a megawatt truly costs across a full market cycle.
It also produced the position most newcomers simply cannot reach: access to more than 5 GW of power, available as demand requires, alongside established energy-supply and procurement relationships. The hardest part of an AI campus is not constructing the building — it is securing abundant, affordable, interconnected power. That is why powered land sits at the base of everything Castellan offers.
Most AI-infrastructure entrants start from zero on power. Castellan starts with the hardest asset already in hand.
These are not aspirations on a slide. They are functions the group runs today — from energy sourcing through GPU procurement at scale to high-density HPC hosting — transferred into Castellan as working capabilities, with the people and counterparties attached.
Castellan was created to carry this operating base into a larger mandate: sovereign-aligned AI infrastructure at campus scale, as set out in the Castellan vision. What the group contributes is concrete — four assets that normally take a decade to assemble, available on day one.
Two companies, deliberately separate. Bitkern continues to run its own business, unchanged. Castellan is an independent Swiss company that acquires and develops its own sites — greenfield locations selected and purpose-built for AI, to data-centre standards. No mining facility is converted into an AI campus; the two asset bases do not mix. What flows between the companies is expertise, the supplier and site network — including locations that are unsuitable for mining but ideal for AI — and, in part, the team.
The result: when Castellan signs its first anchor lease, it does so with power access in hand, a supply chain that has already delivered, and a team that has commissioned and operated this class of infrastructure before. That is the difference between a development story and an operating company.
A decade of operations, +300 MW delivered, and access to more than 5 GW of power — brought to bear on sovereign AI infrastructure. Talk to the team, or start with the flagship.