Hyperscale Power

Why this team

This is the strongest founder-market fit in my set. CEO Daniel Rothmund built a solid-state transformer for data centers during his PhD at ETH Zurich's power electronics lab, and co-authored a 2019 paper on a 99%-efficient 10 kV design for exactly this use. CTO Sami Pettersson says he spent 16 years at ABB Corporate Research.

The flip side is that both founders come from research and a large incumbent. Selling to hyperscalers and running a factory are skills I would want to see them hire for early.

• Daniel RothmundCo-founder and CEOPhD at ETH Zurich's Power Electronic Systems Laboratory, building a medium-voltage solid-state transformer for data centers; co-author of a 99%-efficient 10 kV SiC design paper. research-collection.ethz.ch
• Sami PetterssonCo-founder and CTOBy his own account, 16 years at ABB Corporate Research in Switzerland, with a PhD in power electronics. ch.linkedin.com

What exists today, and what does not yet

Traction on the public record

• Closed a €5M seed co-led by World Fund and Vsquared Ventures. Mar 2026 · worldfund.vc
• Six open roles in Zurich, including grid systems, compliance, test, procurement and a partner manager, posted August and September 2026. Sep 2026 · api.ashbyhq.com

What I cannot see from outsideCustomers, pilots, letters of intent, megawatts in discussion and the status of the prototype. None are public. The hiring mix suggests a move from design toward certification and partners.

Competition, and why they could still win

Hyperscale is late and outfunded, so it cannot win on capital. It can win on engineering: higher switching frequency means a smaller, lighter unit, and in a data center hall, floor space and delivery time are worth more than a few points of price. A European base is also a real advantage for European AI campuses, where the American players are thinner.

What I would have to believe

1. The prototype reaches high efficiency at multi-megawatt power, not just in a lab demonstrator.
2. Data center buyers will qualify a new supplier for equipment this critical within two to three years.
3. There is room for a fourth independent supplier alongside three funded rivals and the incumbents.
4. The founders' intellectual property is clean of ETH and ABB claims.

Questions for the first call

1. Where is the prototype today, and when will you have efficiency data at full power?
2. Which data center operator or equipment maker has agreed to test a unit, and on what terms?
3. What does Infineon's logo on your site mean: supplier, partner or investor?
4. Who owns the intellectual property from your ETH and ABB years, and is any of it licensed?
5. Where will you manufacture, and how will you serve US campuses at 13.8 kV from Europe?
6. What does certification cost and take for a unit like this, and who on the team has done it before?

What gets them to a Series A

• A full-power prototype with third-party efficiency and reliability data.
• At least one paid pilot with a data center operator or a major equipment maker.
• A defined certification path and a contract manufacturing partner.
• A sales lead who has sold power equipment to hyperscalers.

When I would walk away

• The prototype misses efficiency or reliability targets at power by a wide margin.
• Heron, DG Matrix or an incumbent signs the large European campuses first.
• The intellectual property turns out to be encumbered.

Risks

• Late to the market and heavily outcapitalized by three funded rivals. techcrunch.com
• Pre-product: the seed funds the first prototype, and headline efficiency is a target. worldfund.vc
• Incumbents with Nvidia relationships are building the same product. powerelectronicsnews.com