This Week in Voltage
The nuclear-AI energy story got a new wrinkle this week, and it comes with a cautionary tale baked in. Fermi Inc. — the Rick Perry-backed startup that went public at a $19 billion valuation promising 17 gigawatts of AI-ready power near Amarillo — fired its CEO after failing to sign a single client. The stock is down 84% from its peak. The Texas panhandle site remains mostly unfinished.
The lesson isn't that nuclear-for-AI is a bad idea. It's that vaporware gets punished eventually, even in a bull market. "Going from zero to a 17-gigawatt AI hyper campus in one leap before they had a single tenant or a dollar of project finance — that was a lot in hindsight," energy analyst Timm Schneider told the LA Times. Contrast that with Blue Energy and GE Vernova, which announced a disciplined gas-to-nuclear conversion strategy this week: gas turbines online by 2030, BWRX-300 SMRs converting the site to nuclear by 2032, with a final investment decision gated to 2027. That's how you build credibility in this market — milestones, not market caps.
The abundance thesis is intact. The hype-without-execution playbook is not.
Deep Charge: The Orbital Grid Is Closer Than You Think
Here's a number that reframes the space-based solar conversation: launch costs have fallen from roughly $10,000 per kilogram in 2000 to just over $1,000/kg today, according to SBSP pioneer John Mankins. Citi Research forecasts costs could reach $100/kg within two decades. That single cost curve is what separates space-based solar from a physics lecture and puts it on an infrastructure timeline.
The concept is straightforward in principle: put photovoltaic arrays in geosynchronous orbit where the sun shines 24/7, transmit the power to Earth via microwave or laser, receive it at ground stations. No weather. No night. No intermittency. The IEA projects U.S. data centers will need roughly 130% more energy by 2030 — and that demand doesn't care whether the electrons come from a reactor in Wyoming or a solar array in geostationary orbit.
What's changed is that hyperscalers are starting to act on this. Meta has signed an agreement with Overview Energy to explore powering AI data centers with space-based solar, according to PV Magazine. The strategic logic is explicit: orbital generation bypasses terrestrial grid constraints and the interconnection bottleneck that has paralyzed hundreds of gigawatts of conventional projects. The domestic interconnection queue hit 241 GW in 2025 — that's not a queue, that's a wall. If you can beam power directly to a data center's receiving station, you've routed around the entire problem.
I'd argue this is the real civilizational unlock hiding inside the space-based solar story. It's not just about clean energy or baseload reliability — it's about decoupling energy infrastructure from geography entirely. A terrestrial grid requires right-of-way, permits, transmission corridors, and decades of regulatory combat. An orbital power system, once the launch economics work, can serve load anywhere on Earth with a receiver antenna. Mankins' SPS-ALPHA concept envisions arrays capable of dispatching tens of megawatts to several gigawatts from a single geostationary position — potentially serving markets from South Africa to Northern Europe from one installation.
The remaining bottlenecks are real: project financing at orbital scale, wireless power transmission efficiency at commercial distances, and spectrum allocation through the International Telecommunications Union. None of these are physics problems. They're coordination problems — which means they're solvable on a faster timeline than building a new transmission corridor through three states.
By the Numbers
- ~$100 million in private capital invested in space-based solar in roughly a six-month window as of mid-2025, per SpaceNews
- $1,000/kg approximate current launch cost, down from ~$10,000/kg in 2000 — a 10x reduction that changes the financing math
- 130% projected increase in U.S. data center energy demand by 2030, per the IEA
- 241 GW domestic interconnection queue as of 2025 — the grid bottleneck that makes orbital bypass strategies attractive
- 84% decline in Fermi Inc.'s stock from peak — the market's verdict on ambition without execution
What We're Fighting For
The Fermi collapse and the Meta-Overview deal are two sides of the same coin. Hype without milestones gets destroyed. But genuine infrastructure breakthroughs — the ones that route around the constraints everyone else is fighting over — get funded.
Space-based solar is not ready to power civilization today. But the trajectory is unmistakable: launch costs collapsing, hyperscalers signing early agreements, and a grid interconnection crisis creating demand for exactly the kind of geography-independent power delivery that orbital systems promise. Watch for Overview Energy's first demonstration deployment timeline and whether the ITU spectrum allocation process accelerates under pressure from commercial players. Those are the two gates between sci-fi and infrastructure.
The future belongs to builders who think in orbital mechanics, not just transmission corridors.