MIT spinoff Apollo Atomics raises $31M to build factory-made nuclear reactors for AI data centers

Apollo Atomics has raised $31 million in seed funding to turn decades-old nuclear reactor technology into something the industry has struggled to deliver: a reactor that can be built in a factory, shipped by truck, and deployed in less than two years.

The Cambridge, Massachusetts-based MIT spinoff is developing compact pressurized water reactors aimed at data centers, industrial facilities, utilities, and other large electricity users. Apollo says it already has more than 20 gigawatts of signed letters of intent in its commercial pipeline, a striking figure for a company still working through testing and regulatory approvals.

The oversubscribed round was led by FCVC, with participation from Y Combinator, Telesoft Partners, Alumni Ventures, Robinhood Ventures, Nucleation Capital, Pelion VC, and Duke Capital Partners. Individual investors included Y Combinator co-founder Paul Graham, Evan Meagher, Philip Johnston, Matteo Franceschetti, and Ray Rothrock.

The timing is hard to ignore. AI infrastructure is forcing technology companies to confront a problem that bigger models and faster chips cannot solve on their own: electricity.

Data centers are consuming growing amounts of energy, pushing developers and hyperscalers to secure generation years before facilities come online. Nuclear has emerged as one possible answer, particularly for operators seeking around-the-clock electricity without relying on intermittent generation. Yet building conventional nuclear plants can take years and expose developers to huge construction costs and schedule risks.

Apollo thinks the answer isn’t to reinvent nuclear physics. It’s to rethink how reactors get built.

Shrinking the reactor without abandoning proven technology

Apollo’s approach starts with pressurized water reactor technology, or PWR, which has decades of commercial operating history. Rather than introduce an entirely new fuel cycle or exotic reactor architecture, the startup redesigned one of the nuclear steam system’s largest components: the steam generator.

Its proprietary compact steam system delivers roughly 10 times the energy density of conventional designs, the company says. Apollo says that lets it shrink the overall reactor footprint by about 40 times without sacrificing output.

Credit: Apollo Atomics

That smaller footprint changes the economics Apollo is betting on. Instead of treating each nuclear plant as a massive construction project assembled largely at its final location, Apollo wants to manufacture reactors using repeatable factory processes, transport them by truck, and deploy them in less than 24 months.

The company is developing three commercial systems: the 10-megawatt electric A-10, 50-megawatt A-50, and 300-megawatt A-300. That range could put Apollo in front of customers ranging from industrial operators to large AI data center campuses and utilities.

Apollo was founded by Assil Halimi, who earned a Ph.D. in nuclear engineering from MIT and has worked on reactor operations and advanced PWR designs, and Drew Walker, a hard-tech founder and former White House director. The startup has built and tested a working reactor-system demonstrator inside MIT’s Department of Nuclear Science and Engineering.

Now comes the harder part: proving that the technology can survive commercial operating conditions and pass regulatory scrutiny.

Apollo submitted a regulatory engagement plan to the U.S. Nuclear Regulatory Commission earlier this year. It is seeking NRC authorization by the end of 2026 to use its selected fuel configuration commercially. The startup says the fuel has already achieved criticality at full output.

The new funding will finance Apollo’s next demonstration facility, called the A-1, a 1-megawatt commercial demonstrator. The company plans to conduct long-duration reliability testing, bring key manufacturing processes in-house, grow its engineering and operations teams, and continue work with the NRC.

Apollo is entering an increasingly crowded race to solve one of AI’s biggest physical constraints. The company isn’t betting on a radically new form of nuclear energy. Its wager is almost the opposite: take reactor technology the industry already knows, dramatically shrink the hardware around it, and manufacture it more like a product than a megaproject.

If Apollo can turn its 20-gigawatt pipeline from letters of intent into deployed reactors, the bigger story won’t be the $31 million seed round. It will be whether nuclear plants can finally move from construction sites to assembly lines.