NextFin News - The 140-year-old reign of the iron-core transformer is facing its most serious existential threat as the artificial intelligence boom pushes data center power density beyond the physical limits of Victorian-era engineering. Hyperscale Power, a European startup emerging from stealth on March 10, 2026, has secured €5 million in seed funding to commercialize a solid-state transformer (SST) that promises to shrink power conversion footprints by more than half. Led by World Fund and Vsquared Ventures, the investment signals a desperate pivot by the venture capital community to solve the "power wall" currently stalling the deployment of next-generation AI clusters.
The fundamental problem is one of geometry and physics. Traditional transformers rely on heavy copper coils and massive iron cores to step down voltage, a design that has remained largely unchanged since the late 19th century. While reliable, these units are increasingly incompatible with the 100-kilowatt server racks now being shipped by Nvidia, and the looming 1-megawatt racks that will soon follow. In modern hyperscale facilities, the power equipment required to feed these racks has ballooned to twice the size of the server infrastructure itself. Hyperscale Power CEO Daniel Rothmund, who developed a 99.1% efficient SST during his doctoral work at ETH Zürich, argues that the industry has reached a breaking point where the physical size of legacy hardware is actively slowing the scaling of global compute capacity.
Hyperscale Power enters a market that has suddenly become a high-stakes battlefield. In the first quarter of 2026 alone, SST startups have raised over $280 million, according to PitchBook data. The competitive landscape is already crowded with heavyweights: Amperesand, backed by Temasek; DG Matrix, supported by ABB; and Heron Power, the Andreessen Horowitz-funded venture led by former Tesla executive Drew Baglino. What distinguishes the Hyperscale Power approach is its focus on ultra-high frequency operation. By stepping power up to the tens of kilohertz range before transformation, the company can utilize significantly smaller magnetic components than its rivals, potentially offering the highest power density in the nascent sector.
The economic stakes are immense. The global data center SST market, valued at a modest $35.5 million in 2025, is projected to explode to over $700 million by the end of the decade as hyperscalers like Microsoft and Amazon seek to reclaim floor space for revenue-generating GPUs. Beyond mere size, solid-state technology offers "smart" capabilities that iron-core units lack, such as the ability to buffer power using integrated silicon carbide semiconductors. This allows data centers to stabilize the grid and integrate renewable energy sources more effectively, a critical requirement as U.S. President Trump’s administration continues to push for massive expansions in domestic energy production to fuel the AI race.
However, the transition is not without risk. The primary reason the iron-core transformer has survived for over a century is its near-immortality; a well-maintained unit can last 50 years. Solid-state electronics, by contrast, are subject to the thermal stresses and shorter lifespans of semiconductor components. For a data center operator, the trade-off between a smaller footprint and a potentially higher failure rate is a delicate calculation. Hyperscale Power must prove that its high-frequency architecture can match the ruggedness of the "dumb" iron it intends to replace.
The shift toward solid-state power is no longer a laboratory curiosity but a logistical necessity. As AI clusters move from megawatts to gigawatts, the ability to pack more power into less space becomes the ultimate competitive advantage. While the €5 million seed round is small compared to the war chests of its American rivals, Hyperscale Power’s focus on extreme miniaturization targets the exact pain point of the world’s largest cloud providers. The era of the humming, oil-filled iron box is ending, replaced by the silent, high-frequency pulse of silicon.
Explore more exclusive insights at nextfin.ai.
