Swiss startup rents GPUs with renewable energy: €16M

What happened?

AI Infrastructure Capital AG, a Swiss startup founded in June 2026 by Cédric Waldburger and Roald Parmentier, has closed a funding round of approximately €16 million led by Valyou investment foundation. The company buys enterprise servers with the latest NVIDIA chips, installs them in Iceland — where electricity is 100% renewable thanks to geothermal and hydroelectric power — and leases compute capacity long-term through multi-year contracts. According to the official statement, the company already has signed lease agreements, indicating immediate demand. The board of directors includes Yonten Wagma, managing director of Valyou, and entrepreneur Fabian Villiger, both from Rapperswil, Switzerland.

Why is this important?

Demand for AI compute is growing exponentially, but physical infrastructure — servers, energy, cooling — is the main bottleneck. As the company itself notes, “every request to ChatGPT or Claude runs on a physical server that consumes energy and needs cooling.” The shortage of NVIDIA GPUs, exacerbated by lead times of up to 12 months, has driven up prices. AI Infrastructure Capital AG addresses this by buying servers directly and placing them in Iceland, where the combination of cheap renewable energy and cold climate drastically reduces operating costs. This model not only lowers the carbon footprint but offers a structural advantage: outside cold air allows free cooling most of the year, something that in traditional data centers accounts for up to 40% of electricity consumption. If the model is replicated in other Nordic regions or areas with high renewable penetration (like Quebec or Norway), it could transform the geography of cloud computing.

Market consequences

• More GPU supply for startups: Companies that cannot afford to buy their own servers (unit cost up to €300,000 per NVIDIA H100 GPU) will access compute capacity through leasing, lowering barriers to entry in AI.
• Downward price pressure: If the model scales, it could compete with hyperscalers like AWS, Azure, or Google Cloud, which currently charge between €1 and €3 per GPU hour. A long-term lease in Iceland could be 30-50% cheaper, according to industry estimates.
• Incentive for green data centers: The initiative will pressure other providers to locate in areas with clean energy, accelerating the energy transition of the IT sector. Additionally, governments in countries with renewable surpluses could offer tax incentives to attract such infrastructure.
• Impact on NVIDIA: By increasing direct demand for complete servers, NVIDIA could see a new sales channel, but also risks its GPUs becoming a commodity if independent leasing models proliferate.

What you need to know

The company already has signed lease contracts, though clients have not been disclosed. Iceland offers a structural advantage: cold air for cooling and cheap geothermal energy (€0.04/kWh vs. €0.12/kWh average in Europe). Founder Cédric Waldburger is a partner at Tomahawk.VC, a fund investing in early-stage startups, and co-founder Roald Parmentier has experience in data center operations and hardware procurement. The €16 million round is modest compared to the billions invested by hyperscalers, but shows that the niche model can attract capital. However, the main risk is scalability: Iceland's capacity to host data centers is limited by network infrastructure and industrial land availability. Additionally, reliance on NVIDIA for chip supply introduces geopolitical vulnerability, especially if export restrictions to certain regions tighten.

“Location and cooling determine whether the operation makes economic sense. In Iceland, renewable energy is combined with outside cold air. Over the machines' lifetime, that's a structural advantage,” said Roald Parmentier.

Compared to similar initiatives, such as Meta's data center in Luleå (Sweden) or Google's plans in Finland, AI Infrastructure Capital AG is a small but agile player. Its success could validate a new market segment: leasing AI infrastructure with green energy, something previously offered only by major cloud providers. For end users, this means more options and potentially lower prices, but also the need to evaluate the reliability of a smaller provider. In the long term, if the company expands to other locations with similar conditions (like Chile or Scotland), it could help decentralize AI infrastructure, reducing dependence on a few massive data centers.