Data Center Go-to-Market Podcast

The episode runs at a genuinely high clip of non-obvious claims: DC-only superconducting microgrids with zero transformers, the reframing of the buyer as real estate rather than technical, a specific technology reinsurance product, and levelized solar cost math. Some repetition ("cutting through the noise" recurs) and a lengthy mid-episode ad break drag the average down.

“we can build the power plant and the whole dedicated power facility, including the data center and all the redundancies, including the backup thermal generation for emergency power, at about a half to a third of the cost of a traditional grid connected data center”

“the only thing that matters is speed to power. How fast can you get it done? It doesn't matter if you're more efficient or if the, if you got a PUE that's 10% better than your competitors or that you're even, that you're cheaper, that doesn't matter as much”

Several genuinely contrarian frames land cleanly: speed-to-power as the single buying criterion (explicitly over PUE, cost, and sustainability), the real estate team as true buyer rather than technical gatekeepers, and a novel technology reinsurance stack as a GTM differentiator. The DC superconducting architecture is a non-obvious first-principles design choice. Tempered by standard AI-power-demand narrative scaffolding.

“our buyer is actually at the real estate level. The person who's managing the real estate for the data center company... And what they want to do is they want to know that when they sign on site they're going to hire somebody to build that plant”

“we've created this new stack of extended technology risk insurance that simply is not available publicly on the market”

Powell is a genuine founder-operator who lived the problem (ran an AI company in 2018 that couldn't find adequate rack density), built the solution over six years, assembled a Fortune 500 IP-sharing consortium, and actually bid on Stargate. He speaks from real skin-in-the-game experience rather than thought-leadership positioning, though Open Origin remains a pre-scale company and some claims are unverifiable.

“I ran an AI company back in 2018 and, and we ran into the problem and I couldn't find a host... So I got out of the AI business and I got into the AI energy business”

“we went from, you know, deciding to bid to finished production in 30 business days, which was phenomenal in and of itself for a $17 billion project with sign offs at the CEO level by all of our vendors”

Unusually dense with concrete numbers: solar costs by geography (8-9 cents/watt FOB China, 26-28 cents US, 15-18 cents Mexico), a $9/MWh LCOE calculation, sodium battery cost target of $20,000/MWh versus current LFP, 1 acre/megawatt land density versus 3-4 traditional, 10 MW robotic deployment in 24 hours, $145/kW/month pricing, and the Virginia near-miss grid incident. Forward projections are specific even when unverifiable.

“we should be able to do with three robots and four guys, 10 megawatts of deployment in a 24 hour period”

“sodium batteries for example, that the full expectation is by the time a sodium battery hits scale for consumption in the United States in China it's going to be 20, $20,000 a megawatt hour, which is about a third of the cheapest LFP battery is today”

The host opens space for Powell to develop ideas and lands one good follow-up (how a small startup secured Fortune 500-level insurance). But questions are largely open invitations rather than challenges—no pushback on the forward-looking LNG price prediction, the Virginia near-miss claim, or the unverified Stargate bid. The two self-promotional DCSMI ad reads further dilute the conversational quality.

“When you talk with your existing clients, your existing partners, you feel that people are ready for all of this, that they're really emotionally prepared for what a massive AI infrastructure build out looks like.”

“What's fascinating about that is how you did that without the resources of a Fortune 500 company.”