There is a familiar way to pitch a nuclear startup in 2026: promise a fuel nobody has commercialized, a coolant that isn't water, and a reactor that looks nothing like the ones running today. Apollo Atomics went the other direction. It kept the reactor that already generates roughly 80 percent of the world's nuclear electricity - the pressurized water reactor - and asked a smaller, stranger question. Which single part makes this thing so big, and what happens if we shrink only that?
The answer, in Apollo's telling, is the steam generator. In a conventional plant it is the largest component in the building, a multi-story heat exchanger that moves energy from the reactor's hot water loop into the steam that spins a turbine. Apollo redesigned it into what it calls a Compact Steam Generator, roughly 20 times smaller than the standard unit. Because that one piece dictates so much of the plant's footprint, shrinking it shrinks everything around it. The company says the whole reactor comes out about an order of magnitude smaller - small enough to build in a factory and move on a truck.
01 / The IdeaA boring technology, worn as armor
The bet underneath Apollo Atomics is that in nuclear power, the physics has not been the bottleneck for a long time. The world has run pressurized water reactors for decades, across thousands of reactor-years of commercial operation. What has been slow and expensive is everything else: the concrete, the site work, the years of construction, the licensing of designs regulators have never seen before. By sticking to light water, low-enriched uranium, and standard fuel, Apollo keeps itself on a regulatory path the U.S. Nuclear Regulatory Commission already understands.
That is the contrarian move. Where much of advanced nuclear treats novelty as the selling point, Apollo treats familiarity as the moat. The innovation is deliberately narrow - one component, redesigned - so the rest of the machine can ride on proven parts and existing supply chains. Manufacturability, not exotic science, is the hard problem it chose to own.
It is a useful frame for anyone building in a mature industry. The temptation is always to start from a blank page, because a blank page is where the impressive-sounding ideas live. But a blank page in nuclear means years of first-of-a-kind reviews, unproven materials, and a supply chain that has to be built before the first unit ships. Apollo's read is that the market does not reward the most novel reactor. It rewards the one that can be delivered, financed, and licensed on a schedule a customer can plan around. Keeping 95 percent of the machine boring is what buys the freedom to be interesting about the other 5 percent.
02 / How It WorksOne change, four consequences
- Redesign the steam generator into a compact unit roughly 20x smaller, without giving up thermal output.
- The smaller heat exchanger collapses the size of the surrounding plant by about an order of magnitude.
- A smaller plant can be built in a factory and shipped whole, instead of poured in place over a decade.
- Using standard fuel and licensed physics keeps the design on a regulatory path regulators already know.
Energy infrastructure demands reliability, rapid deployment, and cost discipline. With MIT, we're running full two-loop tests on actual hardware to research long-term performance and reduce uncertainty. Assil Halimi, Co-Founder & CEO
03 / The PeopleAn engineer and an operator
Apollo Atomics has two people on the cap table and two very different resumes. Assil Halimi, the CEO, holds a PhD in nuclear engineering from MIT and has spent about a decade in reactor design and operations. He is the physics. Drew Walker, the COO, is the logistics - he founded an electric boat company, scaled manufacturing at an electric truck maker, and handled operations at the White House. In a company whose entire thesis is "make nuclear manufacturable," pairing a reactor designer with someone who has actually run factories is less a coincidence than the strategy.
Being small is not incidental either. A two-person team in a capital-intensive field is a statement that the early work is design, testing, and licensing groundwork - not headcount. It also raises the obvious question every investor asks, which is whether two people can carry a nuclear program far enough to matter. Apollo's answer so far has been to name milestones publicly and hit them, including a target of reaching criticality with fuel before the Fourth of July, which it says it met ahead of schedule.
There is a quiet discipline in that. Deep-tech timelines are famously elastic, and "soon" tends to expand to fill whatever runway a company has. Setting a dated, checkable milestone - and clearing it early - is the kind of signal that is hard to fake and easy for a customer or an investor to read. It also fits the company's whole posture, which treats nuclear less as a science project and more as a program to be run: define the test, run the hardware, reduce the uncertainty, repeat.
04 / The ProductsThree sizes, one architecture
Apollo's roadmap runs from a small demonstrator to utility-scale machines, all built on the same compact architecture. A 1 MW demonstrator is planned to prove the design on real hardware, followed by commercial reactors sized for different customers.
The design leans on fully passive safety - systems that rely on physics like gravity and natural circulation rather than active intervention - and a commercial-grade fuel supply chain. The company points to a target cost of around 3 cents per kilowatt-hour, a number worth reading as an aspiration rather than a receipt.
05 / The CustomersWhoever needs firm power, fast
Apollo is aiming at a demand curve that barely existed a few years ago. Data centers and hyperscalers are buying electricity by the gigawatt and increasingly want it clean and always-on, on timelines that make a ten-year build irrelevant. Alongside them sit industrial facilities that need process heat or behind-the-meter power, and utilities evaluating small modular reactors. The company reports 20 GW of signed letters of intent - early, non-binding interest, but a signal of where the pull is coming from.
The business model follows the customers. Apollo is not selling a science demo; it is trying to sell power, or the machines that make it, under long-term arrangements. That means the company's fate is tied less to a single breakthrough and more to a chain of unglamorous questions: can it manufacture at volume, can it finance the first units, can it satisfy a regulator, can it hit the cost it advertises. The 3-cents-per-kilowatt-hour figure is the pivot point. If the compact design genuinely collapses construction time and cost, that number is plausible; if it does not, the whole pitch softens. This is the part worth watching over the next few years.
06 / The ProofTesting at MIT, talking to the NRC
In April 2026 Apollo announced a research collaboration with MIT's Department of Nuclear Science and Engineering to run full two-loop experimental tests of its compact reactor under conditions matching large commercial plants. The point is to validate the computational models on real hardware and chip away at the uncertainty that regulators and customers care about. On the regulatory side, the company says it submitted an engagement plan to the NRC in early 2026 and is targeting a construction permit application around 2028.
None of this makes Apollo a finished company. It is early, it is small, and the gap between a working demonstrator and a licensed, factory-built reactor fleet is exactly where a lot of nuclear ventures have stalled. What is notable is the shape of the bet: not a new atom, but a new supply chain wrapped around an old one.
07 / The MarketWhere it sits
Apollo Atomics lands in a crowded field of advanced and small modular reactor developers - names like Aalo Atomics, Oklo, NuScale, X-energy, Kairos Power, Radiant, and TerraPower - each with a different theory of what will make nuclear cheap and fast again. Some are betting on new fuels and coolants. Apollo's differentiation is almost the inverse: same reactor type, same fuel, same regulators, and a single mechanical redesign meant to make the whole thing small enough to mass-produce. Whether that is enough to win is unproven. As a strategy, it is at least legible.