In short: Dean. The flagship bring-your-own-power holding, initiated 2025-AUG-05 and still up ~545% after halving from its May peak. Microturbines of 50 kW–3 MW at ~3× revenue against Bloom's 15–20×, with the same tailwind. The asset is time to market — 1–3 months versus 18–24 months for GE Vernova-scale turbines — and it runs at ~10% of a ~1 GW/yr three-shift capacity. "Multiple ways to win": hyperscaler data halls of 20–30 MW, thousands of 1–4 MW edge data centres still on CPUs, and manufacturers that can no longer lean on the grid. ~$300–350m fully diluted; out of prepackaged bankruptcy three years ago, first profitable year last year.
A data centre is useless without electricity, and in the US the grid can't supply it: the wires were built for demand that hasn't grown in twenty years, and the queue to plug a big new load into the grid runs 18 to 24 months. So the companies building AI capacity increasingly "bring their own power" — they park generators on site and burn natural gas from the pipelines that already criss-cross the country.
Capstone makes microturbines — jet-engine-style generators shrunk from the 200–300 megawatt machines GE Vernova sells down to units of 50 kilowatts to 3 megawatts. They're less efficient per unit than the giants, but they can be delivered in one to three months instead of two years, and when your alternative is having no power at all, speed beats efficiency. The factory is running at roughly a tenth of what it could make, so if orders come the profit leverage is enormous.
Dean's core argument is a price one. Bloom Energy sells a competing on-site power product into exactly the same boom and trades around 15–20 times its revenue; Capstone trades around 3 times. Same wave, a fraction of the ticket. The catch is that this is a ~$300m company that came out of a prepackaged bankruptcy three years ago and only turned its first annual profit last year — and the stock has already more than sextupled since they bought it, then halved again in a month, so the ride is violent in both directions.
15:33So it can produce roughly a gigawatt per year running three shifts. Right now it's 90% less than that. And what we like about capstone is there's multiple ways to win. So number one they can serve a data hall for a hyperscaler. We don't expect it to be producing or supplying 300 megawatts to a hyperscaler.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.