SWI Group pivots 4 GW digital empire toward AI compute platform

The Euronext-listed group is converting bitcoin mining sites into AI infrastructure, targeting 90% digital-asset exposure by 2027.

By Disrupts Editorial Team Sat, 10 October 2026 AI level 2 of 5

AI level 2 of 5: drafted by our AI editorial assistant from the company's announcement; fact-checked, edited and signed off by Mark Walker, Editorial Director. What the levels mean

A vast open-pit mine features multiple excavators and dump trucks operating on terraced roads under bright daylight.

SWI Capital Holding, listed on Euronext Amsterdam under the ticker SWICH, has reported half-year results for 2026 that lay bare one of the more ambitious infrastructure pivots in the current AI investment cycle. The Singapore- and Amsterdam-headquartered group recorded €4.4 billion in total assets and €2.3 billion in adjusted net asset value as of June 30, a 53% uplift since the end of 2025, with a period profit of €631.6 million driven largely by the recognition of value on its investment in Genesis Digital Assets, subsequently rebranded SWI Digital. The headline numbers are less interesting than the strategic architecture behind them.

SWI is assembling a vertically integrated AI compute platform spanning approximately 4 gigawatts of power capacity across Europe and the United States. That infrastructure is held through two vehicles: AiOnX, its European platform with roughly 2.3 GW of planned capacity including one campus already leased to a hyperscale tenant; and SWI Digital, which provides a further 1.2 GW of secured US grid connections, largely inherited from former bitcoin mining sites now being converted into high-performance computing and AI infrastructure.

From bitcoin mines to GPU farms

The conversion of bitcoin mining capacity into AI compute is the clearest expression of a broader capital rotation that has been building since late 2023. Bitcoin mining sites are, at their core, purpose-built power infrastructure with high-density electrical capacity and established grid connections. As AI training and inference workloads have driven insatiable demand for power-dense compute environments, that legacy mining infrastructure has become surprisingly attractive raw material.

SWI formalised its ambitions in August 2026 when it was designated a Preferred Partner for Compute, Networking and Enterprise Software in the NVIDIA Partner Network. The designation allows SWI to deploy NVIDIA-accelerated infrastructure for workloads ranging from model training to production-scale inference. The group says it is assembling a technology team drawing on alumni from NVIDIA, Amazon, and Intel, targeting an in-house AI cloud platform for enterprises, research institutions, and AI developers. By 2027, SWI intends for digital infrastructure to represent more than 90% of total assets, with selected non-core holdings, including mixed-use development land and hospitality projects, classified as held for sale.

Capital flows and the AI infrastructure race

The macro context here is significant. Sovereign and institutional capital is increasingly flowing into a small number of vertically integrated AI infrastructure operators that can credibly claim to control the full stack from land and power to accelerated compute and cloud services. SWI's model, combining European and US grid capacity with GPU deployment and a nascent cloud services layer, positions it alongside a cohort of well-capitalised challengers to the hyperscaler-owned compute model, rather than purely as a co-location landlord.

The group says it is in advanced discussions with hyperscalers and AI developers over long-term offtake agreements, describing the potential aggregate contractual value as running into the tens of billions of US dollars across their respective terms. That figure, while company-issued and unverified, underlines the scale of demand SWI believes it can capture if it converts pipeline to contracted capacity. The group is also evaluating a US equity capital markets transaction to fund the next phase of growth, subject to market conditions, which would bring a Euronext-listed vehicle into the US public markets at a moment when AI infrastructure listings are drawing significant investor attention.

For cross-sector investors, the story cuts across energy, real estate, and AI capital markets simultaneously. The 4 GW pipeline is not merely a compute story; it is a power-security story. Access to grid capacity at scale has become one of the most contested resources in the AI build-out, sitting alongside GPU allocation and hyperscaler offtake commitments as the three variables that determine whether an infrastructure platform can execute. SWI's ability to lock in long-term power agreements across two continents will ultimately be the harder-to-replicate competitive advantage, regardless of how the GPU and cloud services layers develop.