Mike Schroepfer's Gigascale Capital raises $250M to back deep tech and physical world startups

Gigascale Capital raises $250M for deep tech and physical world startups

Mike Schroepfer, former CTO of Meta, has closed $250M for Gigascale Capital, an early-stage venture firm backing deep tech companies in energy, robotics, and manufacturing. The fund was oversubscribed.

Schroepfer's core thesis is that as the marginal cost of software approaches zero, the binding constraint becomes physical output — how much can be built, and how fast. That reorientation from bits to atoms is where he sees the next generation of outsized returns.

“I thought we're moving from bits to atoms. Just like the physical world mattered again. That is the marginal cost of software went to zero. The thing that really mattered is how much stuff can we build and how are we gonna build it... $2.50. Quarter bill. Portfolio includes Panthalassa for ocean data centers, Heron Power on grid electronics, Radiant Nuclear for micro-reactors, and Form Energy for long-duration batteries.”

Portfolio

The firm's existing bets illustrate the thesis. Panthalassa is building ocean-based data centers for AI inference and recently raised a round led by Peter Thiel, with Gigascale as an early investor. Heron Power applies power electronics from electric vehicles to grid infrastructure, promising faster and cheaper power delivery to data centers, with products shipping next year. Radiant Nuclear is building a container-sized micro-reactor that delivers a megawatt of power for five years without refueling. Form Energy is developing iron-air batteries capable of storing energy for up to 100 days — designed for grids that depend on intermittent renewables, at low cost because iron is the core material.

Robotics timeline

Schroepfer is bullish on industrial robotics in the near term, skeptical of home robotics in the short term. Factories and warehouses offer controlled environments where robots can tackle tasks like unboxing — every box slightly different, bags getting stuck — that are genuinely hard to automate but don't require navigating pets, spilled wine, or unpredictable home layouts. The home comes later.

The bigger productivity unlock, he argues, isn't replacing a human with a robot in the same floorplan. It's redesigning the entire facility around robotic utilization — the way factories rebuilt around electric motors rather than just swapping them in for steam engines.

Solar

Solar panels are now so cheap — down roughly 99% in cost — that the panel is no longer the most expensive element of a solar installation. The steel framing costs more. Schroepfer thinks the next solar opportunity isn't competing with Chinese cell manufacturing but rethinking installation itself: new form factors, automation, and a deployment model that treats the cell as essentially free and attacks everything else.

Fund strategy

On capital intensity, Schroepfer's near-term approach is to use Gigascale's network to get portfolio companies funded by others rather than lead every subsequent round — Panthalassa and the Thiel round being the clearest example. He signals that SPVs and follow-on vehicles are likely as companies scale, but says the priority now is staying focused on early-stage deployment. With the fund oversubscribed, he expects multiple options for how to participate as the portfolio matures.