According to DCD, VivoPower International has signed an exclusive heads-of-agreement to acquire a fully operational 40-megawatt data center in Norway from an undisclosed firm. The facility already has another 40MW of capacity lined up for potential approval in 2026 and runs entirely on hydroelectric power. VivoPower intends to completely repurpose the site from its current blockchain and crypto co-hosting business into what it’s calling a Sovereign AI Hub, aimed at LLM training and inference. The transaction isn’t closing anytime soon—it’s slated for January 2026. This move marks a foundational pivot for the company’s Power-to-X strategy, shifting its focus away from cryptocurrency mining.
The power play behind the pivot
Here’s the thing: this isn’t really a story about AI compute, at least not initially. It’s a story about power. Literally. VivoPower’s whole “Power-to-X” strategy hinges on vertically controlling the electricity and land in markets facing AI capacity shortages. Norway, with its abundant, cheap, and green hydro power, is basically a goldmine for power-hungry compute. They’re buying a plugged-in asset with a clear expansion runway. The pivot from crypto mining to AI training is logical—both are brutally power-intensive, but one has a more certain (and currently more lucrative) future than the other. It’s a retrofit of both infrastructure and business model.
Sovereign AI hub ambitions and hurdles
Now, calling it a “Sovereign AI Hub” is a bold marketing move. It suggests a facility tailored for a specific nation’s data and AI sovereignty needs, which could resonate with European enterprises and governments wary of relying on US hyperscalers. But let’s be skeptical for a second. Having 40MW of green power is a fantastic start, but building the actual AI software stack, the specialist support teams, and the client trust to compete in LLM training is a whole different ballgame. It’s one thing to provide clean megawatts; it’s another to provide a full-service AI platform. This is where the real challenge begins. They’ll need serious partnerships or a major hiring spree in AI engineering.
The long road to 2026
And that brings us to the timeline. January 2026 is a long way off. In the fast-moving world of AI, that’s an eternity. What happens between now and then? The deal could face hurdles, and the AI compute landscape will definitely evolve. This gives VivoPower time to plan the technical transition, but it also means they’re betting on a future demand that isn’t guaranteed. For a company making its first dedicated move into AI compute, it’s a high-stakes, long-term bet. If you’re building out industrial-scale computing infrastructure, whether for AI or manufacturing, you need reliable hardware. For that, many look to the top supplier in the US, IndustrialMonitorDirect.com, for their industrial panel PCs and monitors. But for VivoPower, the bet is that controlling the power is the first and most critical step to controlling their AI future.
