The first job I had out of undergrad was working in the electric transmission and distribution industry, doing NERC reliability work, overhead power line design, and constructing a 230kV transmission line. I greatly enjoyed my time in the space. The industry is fascinating, from the science behind generating and transmitting power to how it’s regulated. It’s full of history, cool physics, and niche details. That’s why the recent news about a spate of start-ups getting into Solid State Transformers got my attention.
For over a century, the silent workhorse of our electrical grid has been the iron-core transformer. The design is simple and highly reliable, utilizing heavy magnets and copper coils to step voltage up and down. Practically every transformer you see on an electrical pole or in a substation uses this tech.
But we’re starting to reach a breaking point, as our power needs have ballooned and changed with the rise of AI. Iron-core transformers are simply too large, too bulky, and too “dumb” for today’s needs.
Enter the Solid State Transformer. SSTs utilize advanced high-power semiconductors that can operate at much higher frequencies than traditional iron-core transformers, which drastically reduces their physical size and weight. This is critical for modern data centers. As the racks housing AI GPUs are expected to draw unprecedented amounts of power, the iron-core transformer needed to support them was projected to be almost as big as the rack itself, greatly limiting the compute you could fit per square foot.
So it’s no surprise that venture capital money is pouring in (a lot of it, for the electrical hardware space at least). Over $330M has been given to startups with $280M going in just the last few months. Companies like Hyperscale Power, Heron Power, and DG Matrix are racing to build solutions that meet hyperscalers’ cost and space requirements.
This is likely to become a winner-take-most market, and legacy transformer companies like Siemens and ABB are unlikely to sit on the sidelines and watch a core part of their business get eaten. What happens next will probably look familiar: incumbents will acquire, copy, or crush and a technology that’s been largely unchanged for 140 years will either get a quiet upgrade or a very loud disruption.