Hive is expanding its European AI Cloud operations by converting an existing Tier-1 data center in Boden, Sweden, into a Tier-3 liquid-cooled HPC facility, a strategy that, acoording to the company, provides a faster path to cash flow, typically 9 months versus 3 years for a greenfield build.
Hive says engineering and design are complete, with construction commencing this quarter. The upgraded facility will support 2,000 NVIDIA GPUs for enterprise-grade AI and GPU cloud workloads across the EU. This expansion builds on HIVE’s AI operations in downtown Stockholm, active for the past two years, which have served early enterprise clients and validated demand for green, low-latency compute in Northern Europe.
Bitcoin miners as pioneers
Frank Holmes, Hive's Co-Founder and Executive Chairman, said: “It is now widely recognized that Bitcoin miners were the original builders of Tier-1 digital-infrastructure networks – developing substations, fiber connectivity, and energy-optimization systems that laid the foundation for today’s hyperscale AI data centers. This model is now being replicated globally, including in West Texas, where projects such as Stargate, a US $500 billion HPC campus, mirror the pioneering work of early Bitcoin-mining companies that sourced stranded or surplus renewable energy.”
Aydin Kilic, President & CEO, added: “Repurposing existing infrastructure allows HIVE to reach cash-flow generation much faster than greenfield projects. Our operational strength enables us to grow both Bitcoin mining and AI Cloud services in parallel, all powered by renewable energy.”
Johanna Thornblad, Sweden Country President, commented: “Building on our two years of AI operations in Stockholm, this conversion project will deliver enterprise-grade AI capacity to the EU market faster and more efficiently than traditional data-center builds.”
Together with the BUZZ data center in Toronto – which will host an additional 2,000 GPUs in 2026 – and planned growth in New Brunswick, HIVE has secured power and land at three strategic locations for next-generation, renewable-powered HPC operations. Through its colocation partnership with Bell Canada, HIVE’s BUZZ division can rapidly deploy AI Cloud GPU infrastructure, enabling flexible scaling to meet enterprise demand. Across its global pipeline, HIVE expects to operate approximately 6,000 GPUs by 2026, serving both AI training and inference workloads.
Pictured: HIVE's 32 MW facility in Boden, Sweden.