SWI Group is accelerating its transformation into a digital infrastructure and AI compute company, combining more than 4 GW of planned power capacity across Europe and the U.S. with NVIDIA-accelerated computing and plans for an enterprise AI cloud platform.
The race to build AI infrastructure is increasingly becoming a race for something more fundamental: access to power. SWI Capital Holding, or SWI Group, is positioning itself around that constraint, moving from digital infrastructure ownership toward a vertically integrated platform spanning power, data centers, GPUs and AI cloud services.
In its interim financial report for the six months ended June 30, 2026, the Amsterdam-listed group reported €4.4 billion in total assets and €2.3 billion in Adjusted NAV, up 53% from the end of 2025. The company says its digital infrastructure portfolio now represents approximately 4 GW of power capacity across Europe and the United States.
The strategy goes beyond building or leasing data centers. SWI is assembling infrastructure that can support AI workloads directly, with plans to operate an in-house AI cloud platform for enterprises, research organizations and AI developers.
That strategy is closely tied to the growing scarcity of power for AI data centers. Gartner estimates global data-center electricity consumption will reach 565 TWh in 2026, a 26% increase from 2025. The research firm also forecasts worldwide data-center power demand will rise to 290 GW by 2030, with AI-optimized servers accounting for an increasingly large share of consumption.
SWI’s portfolio is being assembled through two primary platforms. AiOnX, its European digital infrastructure operation, has approximately 2.3 GW of planned capacity across several markets, including a campus leased to a hyperscale customer. In the U.S., SWI Digital, formerly Genesis Digital Assets, provides approximately 1.2 GW of secured grid connections.
The latter acquisition also illustrates the changing economics of existing computing infrastructure. SWI says it plans to optimize existing SWI Digital facilities and convert suitable bitcoin-mining sites into large-scale AI and high-performance computing infrastructure.
That conversion reflects a broader trend in AI infrastructure: developers are looking for locations where power and grid access already exist rather than waiting years for entirely new capacity. In an increasingly power-constrained market, the ability to control energized sites can become a competitive advantage.
SWI is now adding the compute layer. In August, the company announced that it had become a Preferred Partner in NVIDIA’s Cloud Partner program, giving it access to NVIDIA reference architectures and validated configurations for accelerated infrastructure. The partnership supports workloads ranging from model training and fine-tuning to production inference and agentic AI.
The move puts SWI closer to the infrastructure model pursued by specialized AI cloud providers and “neoclouds,” which combine data-center capacity with GPU infrastructure and cloud services. Unlike a conventional real-estate or data-center operator, SWI wants to capture value across multiple layers of the AI compute stack.
Its planned technology organization is intended to support that transition. SWI says it is recruiting a team with experience at NVIDIA, Amazon, Intel and major hyperscale operators to develop and operate its AI cloud platform.
The business model will depend heavily on converting infrastructure pipelines into contracted demand. SWI says it is in advanced discussions with hyperscalers and AI developers about long-term offtake agreements whose potential aggregate contractual value is in the tens of billions of dollars.
Those figures remain prospective rather than contracted revenue, making execution critical. The company must secure customers, financing, grid capacity, GPUs and the operational capabilities required to run AI infrastructure at scale. Data-center operators are also facing greater investor scrutiny as the capital requirements of AI infrastructure increase. Reuters recently reported that data-center companies pursuing public-market transactions are being evaluated more closely on contracted demand and financing requirements.
SWI says it intends for digital infrastructure to represent more than 90% of total assets by 2027, with non-core assets being prepared for sale.
If successful, the transformation would give SWI a position spanning AI infrastructure, accelerated computing and AI cloud platforms, rather than simply providing the physical facilities in which AI workloads run. The larger question is whether control of power and sites can translate into a durable advantage once GPU supply, financing, customers and cloud operations become equally important.
For the AI infrastructure market, however, SWI’s strategy highlights a central shift: as AI workloads expand, infrastructure companies are increasingly moving upstream from land and power toward compute capacity and the software services that monetize it.
Market Landscape
AI infrastructure is becoming increasingly constrained by electricity availability. Gartner forecasts global data-center electricity consumption of 565 TWh in 2026, up 26% year over year, while worldwide data-center power demand is expected to reach 290 GW by 2030.
That environment is encouraging data-center developers, hyperscalers and AI cloud providers to secure power, grid connections and suitable sites years ahead of demand. At the same time, infrastructure providers are moving toward vertically integrated models combining facilities, GPUs, networking and cloud services.
SWI’s strategy places it within this emerging AI infrastructure segment alongside hyperscalers, GPU cloud providers and specialized data-center operators. Its differentiation will depend on converting its power pipeline into operational AI capacity and securing long-term customers.
Top Insights
- SWI is moving beyond data-center ownership toward an integrated AI infrastructure model spanning power, facilities, GPUs and cloud services.
- Its 4 GW infrastructure pipeline gives the company exposure to the increasingly valuable intersection of power availability and AI compute demand.
- The NVIDIA Cloud Partner relationship supports deployment of accelerated infrastructure for training, inference, fine-tuning and agentic AI workloads.
- SWI Digital could repurpose suitable bitcoin-mining infrastructure for AI and high-performance computing, potentially shortening deployment timelines.
- The strategy remains execution-heavy, requiring SWI to convert prospective customer discussions into contracts, financing and operational AI capacity.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
