Startup Spotlight: Apollo Atomics Is the Nuclear Startup to Watch in the AI Power Panic

Why it matters

Apollo Atomics is targeting the same bottleneck now shaping AI infrastructure and heavy industry: power timelines. Its bet is that using proven pressurized water reactor technology, while shrinking one key component, can make nuclear deployable fast enough for customers who cannot wait for new grid capacity.

The YC-backed Cambridge startup says its Compact Steam Generator can shrink proven pressurized water reactor architecture enough for factory production, truck transport and customers who need firm power before the grid catches up.

Apollo Atomics has also started putting more of its timeline in public. In a recent LinkedIn post, the company shared a slide that appears to lay out Commercial Pathways for Apollo Reactors: a Compact Steam Generator design and reactor announcement, an MIT Compact Reactor Demonstrator A-0 under construction as a first full-scale CSG demonstration in 2026, a megawatt A-1 demo for multicoolant testing in 2027, an A-10 reactor for data centers in 2028, an A-50 reactor for small grids in 2029, an A-300 reactor for large grids in 2031, and A-100C reactor systems for large multi-unit campuses. The same slide appears to point to additional applications in fusion steam systems and maritime propulsion.

Source: LinkedIn

zoomed in

That public roadmap matters because Apollo Atomics is not just pitching a better component. It is pitching a deployment sequence.

The company has a public video overview, but the more important story is the commercial architecture behind it. Apollo Atomics is one of the hard-tech companies worth putting on the watch list now, before the nuclear conversation gets louder.

The Cambridge, Massachusetts startup is not leading with a new fuel cycle, a novel coolant or a far-off render of an atomic future. It is taking the most deployed reactor architecture on earth, the pressurized water reactor, and attacking one very specific piece of the plant: the steam generator.

That sounds boring until you understand the claim.

Apollo Atomics says its Compact Steam Generator is 20 times smaller than a conventional steam generator while delivering the same thermal power. That single component, the company argues, lets it make a pressurized water reactor roughly 40 times more compact overall. The result is supposed to be a reactor that moves from the nightmare category of bespoke megaproject into the much more interesting category of factory-built machine.

That is the whole bet.

Not nuclear, but with vibes.

Not energy solved by a beautiful deck.

The startup's wager is that if you can shrink the right component inside the most boring and bankable nuclear architecture, you can change the deployment model without asking regulators, fuel suppliers, utilities and customers to believe in an entirely new physics stack.

For a world running out of patience with the grid, that is a very big claim.

The company

Apollo Atomics was founded by CEO Assil Halimi and co-founder and COO Drew Walker. The company is part of Y Combinator's Spring 2026 batch and publicly describes itself as building the most compact nuclear reactors.

The pitch is aimed directly at the new power panic.

Data centers need 24/7 firm power. Industrial operators need electricity and process heat. Utilities are staring at load-growth forecasts that suddenly look less like normal planning exercises and more like a slow-motion emergency. The AI boom did not create every grid problem, but it made the timing brutal. Customers do not need a white paper about power in 2040. They need credible megawatts on a timeline that does not require waiting for an entire transmission buildout to show up like a miracle.

Apollo Atomics' proposed reactor ladder includes:

• A-10: 10 MWe, with the public roadmap pointing to data centers in 2028
• A-50: 50 MWe, with the roadmap pointing to small grids in 2029
• A-300: 300 MWe, with the roadmap pointing to large grids in 2031
• A-100C reactor systems: large multi-unit campuses, according to the public slide

The sizing tells you who the company is really chasing. A 10 MWe unit can be relevant for smaller behind-the-meter or industrial use cases. A 50 MWe plant starts to look like real data-center power. A 300 MWe plant moves closer to utility-scale generation while still sitting far below the gigawatt-class nuclear builds that have defined the industry's cost and schedule nightmares. The A-100C framing pushes the same logic toward campuses that need multiple units rather than a single reactor.

The company says each reactor is designed to fit on a truck.

That is the phrase investors will repeat. The more important phrase is factory-built.

Because the dirtiest secret in nuclear is not that the technology does not work. It works. The dirty secret is that projects keep turning into custom civil-construction epics with terrifying financing risk, long schedules, regulatory uncertainty and a workforce problem that gets worse every time the industry goes quiet for a generation.

Apollo Atomics is trying to make nuclear look less like building an airport and more like buying critical infrastructure from a qualified factory line.

That is why the startup matters.

The trick: make PWRs small without making them weird

The company's core claim starts with a practical observation: pressurized water reactors already have the deepest commercial history in nuclear power.

In a conventional PWR, the reactor heats water in a high-pressure primary loop. That heat transfers through a steam generator into a secondary loop, producing steam to drive a turbine. This is not science fiction. It is the mainstream.

Apollo Atomics' pitch is that most advanced reactor startups introduce too many new variables at once. Exotic fuels. Unproven coolants. New supply chains. New regulatory questions. New manufacturing processes. New operating regimes. Sometimes the whole thing looks less like a company and more like a stack of heroic assumptions wearing a logo.

The startup is trying to narrow the miracle.

The company says it keeps the familiar ingredients: light water, low-enriched uranium, existing qualified components and the PWR architecture regulators have seen many times before. The innovation is concentrated in the steam generator.

Its Compact Steam Generator replaces the large hand-built steam generator in a conventional PWR with a much smaller component that Apollo Atomics says can deliver the same thermal power. In the company's telling, this is not a side optimization. It is the wedge that changes the plant.

A smaller steam generator means a smaller reactor system. A smaller system means transportability. Transportability means factory production. Factory production means fixed-price contracting becomes at least imaginable. Fixed-price contracting means customers can maybe buy nuclear without signing up for a decade of just-trust-us procurement hell.

That is the dream.

And yes, nuclear dreams have a body count.

Which is why the company's narrowness is interesting. Apollo Atomics is not trying to reinvent nuclear from the neutron up. It is trying to make one hard thing dramatically better and ride the existing credibility of the rest of the system.

That is a real startup strategy.

The team

The most important thing about Apollo Atomics is not the video. It is not the truck line. It is the founder-market fit.

Assil Halimi is the kind of founder nuclear investors want to see before they let themselves get excited. Apollo Atomics' YC profile says Halimi has a PhD in Nuclear Engineering from MIT, 15 peer-reviewed publications and two patents on reactor design and fuel optimization. The same profile says he previously worked as a reactor operator at Engie, as an advanced reactor designer at EDF, and worked with Westinghouse, NuScale and Rolls-Royce on PWR development.

That background matters because nuclear is not a learn-it-on-YouTube-and-ship-an-MVP category. It is a field where the distance between sounds plausible and can be licensed, manufactured, financed, insured and operated is enormous.

MIT's own profile of Halimi, from before Apollo Atomics' launch, is almost comically on-theme. His research focused on high burnup fuel assessment and small modular reactor design. He was already thinking about the economic trap of SMRs: smaller reactors can be faster and more repeatable, but if you simply shrink a reactor without improving the design, you may lose economies of scale and make the economics worse.

That is exactly the trap the company is trying to avoid.

Halimi's line of attack is high power density. More energy from less plant. Less site work. Less complexity. Less schedule risk. Nuclear people can argue about the details, and they will, because that is their love language. But at least the founder's research history points directly at the company's thesis.

Drew Walker brings the other half of the company-building puzzle. Apollo Atomics' YC profile lists him as co-founder and COO, with prior experience co-founding Blue Innovations Group, an electric boat startup, and service in the White House. That is a strange-looking resume until you remember what nuclear actually requires.

Nuclear startups do not just need technical credibility. They need policy fluency, customer development, capital formation, government relationships, site strategy, public trust, industrial partnerships and the ability to sell something terrifyingly complex to people who are both desperate for power and allergic to embarrassment.

A nuclear company with only academics can get stuck in the lab. A nuclear company with only salespeople can become a PowerPoint reactor. The Apollo Atomics pairing is interesting because Halimi gives the company technical legitimacy while Walker gives it an operator who can live in the mess of customers, regulators, investors and public-sector stakeholders.

That is not a guarantee. It is just the right shape.

The heat around the company

Apollo Atomics is already backed by Y Combinator, which gives the company instant visibility inside the current AI-infrastructure frenzy. That matters more than usual because this is not a normal software YC story.

YC companies usually sell speed. Apollo Atomics is selling speed in a category famous for not having any.

According to confidential sources, Apollo Atomics is also backed by Pioneer Fund, an exclusive venture capital fund that is backed by over 500 Y Combinator alumni limited partners. Public posts also show Spot VC and other top-tier seed funds signaling involvement around the company.

The more interesting signal is what sources describe around the company, without getting into sensitive specifics: early customer pull from prospective power buyers, continued technical progress around the demonstrator program, and non-dilutive support that suggests the company is not merely telling a story into the AI power panic.

None of that means Apollo Atomics has solved nuclear deployment.

Early customer interest is not revenue. It is not a binding power purchase agreement. It is not NRC approval. It is not a factory. It does not mean the company has magically turned nuclear into a shippable appliance.

The public roadmap should be read the same way. A slide that puts A-0 in 2026, A-1 in 2027, A-10 for data centers in 2028, A-50 in 2029 and A-300 in 2031 is an important signal of ambition and sequencing. It is not the same thing as a licensed operating reactor, a signed customer contract or a completed manufacturing base.

But in this market, the direction of demand matters. Buyers are shopping for firm power years before many of these reactors can exist. That is the market pull investors are underwriting.

The public record adds another important piece. Apollo Atomics announced a research collaboration with MIT's Department of Nuclear Science and Engineering in April 2026 to expand experimental infrastructure around its reactor demonstrator. The company said the collaboration would test primary and secondary reactor loops under prototypic reactor conditions and generate data needed to validate its computational models and advance its NRC licensing pathway.

The company also said in that release that it had submitted its regulatory engagement plan to the Nuclear Regulatory Commission and was targeting a construction permit application in 2028.

That is where the story gets real.

A nuclear startup can make a video. It can win Demo Day. It can collect soft demand from customers terrified of the grid. None of that matters unless the company can turn its physics and manufacturing story into a licensing story.

Apollo Atomics appears to understand that. It is not waiting until the end to discover the NRC exists. It is trying to build the regulatory case while the hardware story is still young.

Good.

That is table stakes, but a shocking number of hard-tech companies treat table stakes like optional furniture.

Why this is exciting

Apollo Atomics sits at the intersection of three massive shifts.

First, AI is turning electricity into a strategic bottleneck. The market spent the last two years talking about GPUs like they were the scarce resource. They still are, but the next shortage is increasingly obvious: powered sites. Data centers do not just need chips. They need land, water, interconnection, cooling, backup power and enormous amounts of reliable electricity.

Second, nuclear is having a credibility comeback. Nuclear is back because the alternatives are not enough on their own for customers who need firm, clean, always-on power. Solar and wind are essential, but intermittency and transmission constraints are real. Gas is fast but carbon-heavy and politically exposed. Batteries help, but they are not a full substitute for 24/7 generation at industrial scale.

Third, the advanced nuclear market is crowded with ideas that are brilliant, ambitious and hard to underwrite. The world may absolutely need new reactor architectures. But customers trying to power a data center before 2035 are not always looking for the coolest reactor. They are looking for the least risky path to electrons.

That is Apollo Atomics' opening.

The company is not saying: believe in our completely new version of nuclear.

It is saying: believe that we can shrink the component that makes the old version too big, too slow and too expensive.

That is a cleaner story.

It also gives the startup a sharper wedge with customers. If you are a hyperscaler, industrial operator or utility, you do not want to be the science experiment. You want the upside of SMRs without becoming the guinea pig for a fragile new fuel supply chain or a decade of regulatory novelty.

Apollo Atomics' pitch is basically: we kept the boring parts boring.

That may be the most exciting sentence in nuclear.

What has to go right

The hard part is that nuclear does not care about startup narrative.

A smaller steam generator does not magically solve licensing. It does not build a supply chain. It does not certify manufacturing quality. It does not secure sites. It does not train operators. It does not make local politics disappear. It does not turn early interest into signed power contracts. It does not remove the need for detailed safety analysis, environmental review and years of regulatory work.

Apollo Atomics has to prove several things at once.

It has to prove the Compact Steam Generator works not just in a demo, but under the brutal expectations of long-duration reactor operation.

It has to prove the plant architecture remains safe and licensable when compressed into a radically smaller footprint.

It has to prove the economics survive real manufacturing, nuclear-grade quality assurance, fuel procurement, installation, security, maintenance and decommissioning assumptions.

It has to prove customers will sign binding agreements at prices that support the factory model.

It has to prove the NRC will move on a timeline that matches the market story.

It has to prove a startup-sized team can become a nuclear-industrial company without getting crushed by the weight of its own category.

That is a lot.

But that is why the company is interesting. Easy nuclear startups are fake. Hard nuclear startups are the only kind worth covering.

The key question

The key question for Apollo Atomics is whether its Compact Steam Generator is a component breakthrough or a company breakthrough.

Those are different things.

A component breakthrough can be technically impressive and still fail to change the deployment model. A company breakthrough means the component unlocks a system: manufacturing, transport, licensing, contracts, financing, installation and repeatability.

The company's thesis depends on the second version.

If the steam generator really lets Apollo Atomics produce compact PWRs at predictable cost and ship them inside a 24-month deployment window, the upside is enormous. Not nice YC company enormous. Real energy infrastructure enormous.

A 10 MWe reactor for behind-the-meter power and data centers. A 50 MWe reactor for grids and industrial demand. A 300 MWe reactor for larger grids. A-100C systems for multi-unit campuses. All based on familiar nuclear materials and existing supply chains. That is the kind of product ladder that could map to actual customer segments instead of a single moonshot plant that takes forever to arrive.

But if the compact steam generator only solves part of the size problem while the rest of nuclear's institutional drag remains intact, Apollo Atomics becomes another smart team fighting the same ancient beast.

That is the tension.

RuntimeWire's read

Apollo Atomics is exactly the kind of startup this moment is going to produce: wildly ambitious, deeply technical, AI-adjacent without being an AI wrapper, and aimed at an infrastructure bottleneck that is no longer theoretical.

It is also the kind of startup that deserves skepticism proportional to the size of the claim.

The exciting part is not that the company says it can make nuclear smaller. Every SMR startup says some version of that. The exciting part is that Apollo Atomics' wedge is specific, legible and tied to the most proven reactor family in commercial use.

The new public roadmap makes the story sharper. Apollo Atomics is not merely saying compact reactors someday. It is describing a path from Compact Steam Generator demonstration to megawatt testing to data-center, grid and campus reactor systems.

That pathway still has to survive contact with licensing, manufacturing, financing, customers and the NRC. The slide is a signal, not proof.

But in a market full of advanced nuclear companies trying to persuade the world to adopt something new, Apollo Atomics is trying to make something old behave like a product.

That might be the better trick.

There is a version of the next decade where the winners in AI infrastructure are not just the companies with the best models or the cheapest GPUs. They are the companies that control the power stack beneath the intelligence layer.

Apollo Atomics is not there yet. It still has to license, manufacture, finance, sell and operate nuclear machines in a country where nuclear optimism has been humbled many times before.

But the shape of the company is right. The market pain is obvious. The team has real nuclear credibility. The investor heat is there. Sources point to customer pull and early technical progress without yet answering the only question that matters: can this become infrastructure, not just a compelling nuclear story?

A lot of startups are trying to make nuclear exciting again.

Apollo Atomics is trying to make it shippable.

That is a much better story.