Two US power conversion plants, both promising 40 GW a year
Serguey Asael Shinder2 min readAndy Colthorpe reports two US power conversion stories on the same day, and they carry the same number.
Spain-headquartered Power Electronics held a groundbreaking in Houston, Texas on 24 September for a 53-hectare campus combining logistics, production, electronics manufacturing, R&D and training, expected to create more than 400 jobs. The company says the site will reach 40 GW of annual production capacity when fully ramped, will be its most automated inverter plant, and will bring its worldwide total to 105 GW. The report is explicit about what was not said: no initial capacity, no product split between solar inverters and battery PCS, and no date by which ramp-up is expected.
Heron Power - founded by Drew Baglino, formerly Tesla's senior VP of powertrain and energy engineering - has landed a pilot with RWE for its Heron Link, a 5 MW medium-voltage PCS that integrates bidirectional inverters with solid-state transformers for large-scale energy and data centre projects. Baglino names limited access to grid-expanding technology as the bottleneck on meeting demand growth. The target is up to 40 GW a year - some 10,000 units - from Morgan Hill, California, by late 2027, backed by a stated intention in August to invest more than $100 million in the facility. The company has raised $200 million across two rounds.
What it means
Forty gigawatts appearing twice in one day, from a Spanish incumbent and a two-round startup, is the detail to sit with. It is not evidence that either is wrong; it is evidence that headline nameplate capacity has stopped discriminating between very different propositions. One of these numbers describes a built campus with an unstated ramp; the other describes a 2027 intention funded by $200 million of venture capital and $100 million of announced capex. Both are "40 GW". Anyone using announced PCS capacity to reason about supply in 2028 is adding figures that mean different things.
The omissions in the Power Electronics announcement are the more useful part of the report, and worth crediting the reporter for surfacing rather than smoothing over. Without an initial capacity, a product split or a ramp date, "40 GW when fully ramped" is a ceiling with no schedule attached - and the product split matters especially, because inverter lines serving solar and battery PCS lines are not freely interchangeable in practice. A buyer's real question is how many megawatts of battery PCS leave Houston in a given quarter, and that number has not been published.
Heron's technical bet is the separate thing to track. Integrating the transformer into the conversion unit as an SST removes a separately-procured, long-lead component from the project critical path - which, if it works at 5 MW in RWE's pilot, addresses exactly the bottleneck Baglino describes. That is a claim a pilot can actually settle, unlike a capacity figure, so the RWE result is the more informative of today's two announcements.