Who’s winning the SMR race?

If friendly aliens landed on Earth in 2025, they’d be baffled.

You earthlings discovered nuclear power in 1938… yet you barely use it? Does not compute.

Nuclear is the cleanest, safest, densest, and most reliable energy source ever discovered. Yet instead of embracing it, we all but banned it in the West.

Thankfully we’ve snapped out of it, prompted by the need for a lot more clean energy to feed AI progress. Governments aren’t just allowing nuclear development again… they’re urgently pushing it, clearing regulatory roadblocks and widening loopholes for nuclear entrepreneurs to sprint through.

It’s the most exciting time for nuclear in at least 50 years. In this monthly Deep Dive - which you’ve received in lieu of our usual Rational Optimist Diary - we’ll focus on Small Modular Reactors (SMRs) and the crop of companies racing to deploy the first one in the US.

As you read, keep in mind 71 American nuclear-powered submarines have been cruising the oceans for some 60 years. They can go years without refueling. Sailors sleep feet from the reactors.

SMRs can bring this safe, portable, clean energy to everyone. Imagine dropping a self-contained, school-bus-sized SMR into a remote town to safely power it for 20 years.

We won’t have to imagine much longer.

I expect the first US SMR to switch on within 9 months, marking the start of the second atomic age. I met with most of the SMR “pure play” companies you’ll learn about below.

The founders driving these companies forward are some of the most determined entrepreneurs I’ve ever met in any sector.

First, a reminder that nuclear aces the four key energy tests:

No. 1: Clean. Nuclear is twice as clean as wind and 9 times cleaner than solar, when measured by CO-equivalent emissions.

No. 2: Safe. More men die working on oil rigs each year than have ever died from nuclear power. Chernobyl included. And this is with “old” nuclear technology. SMRs are even safer.

No. 3: Reliable. Nuclear reactors hum along 24/7 with a 92% uptime. That’s twice as reliable as coal and three times more reliable than solar. They can run for decades without refueling.

No. 4: Dense. A fingertip-sized fuel pellet packs the same energy as 1.3  tons of coal.

Nuclear’s Achilles’ heel has been cost.

America’s newest reactors, Vogtle 3 and 4 in Georgia, just finished seven years late and $20 billion over budget. That’s not because uranium or concrete is expensive. It’s because building a nuclear plant drags on for decades, allowing a mountain of interest on debt to pile up. In fact, assuming a 7% borrowing rate, 60-70% of nuclear’s cost isn’t the material, fuel, or labor. It’s the financing.

It wasn’t always this way. In the 1960s, the US was building reactors in four years flat for cheaper than the cost of a coal plant. France and South Korea still make nuclear plants efficiently. Here in my new home in the UAE, the Barakah nuclear plant, which opened in 2020, was built in just seven years thanks to standardized design and streamlined regulation.

But in America, layers of regulation first implemented in the 1970s strangled nuclear. Courts forced endless environmental reviews. The Nuclear Regulatory Commission (NRC) dragged its feet on approvals, then said “no” anyway to 90% of projects. Timelines ballooned from 4 years to 15, paperwork hit 100,000 pages, and costs skyrocketed.

The US has shut down more plants than it’s opened this century.

SMRs can solve the cost problem. They can be built fast in factories, turning nuclear into a product that can be quickly scaled and improved, like a TV or a smartphone.

Here’s old nuclear’s larger-than-life cooling towers:

Here’s a new nuclear microreactor at Aalo Atomics, small enough to fit in a backyard:

Let’s break down exactly what an SMR is:

S = small

For perspective, 94 “old school” nuclear reactors still power roughly 1 in 5 American homes today. On average, each one pumps out a little over 1,000 megawatts (1 gigawatt), enough to power around 850,000 homes.

SMRs generally produce under 300 megawatts and can be classified as follows.

Class 1: Microreactors

Think generators. Portable, truck-shippable, and designed for off-grid or defense. They can power a submarine, a military base, or a small town of about 5,000 homes.

Class 2: Campus-sized Reactors

Too big to ride on a truck, but small enough for a single site. Perfect for an industrial facility or small data center. Can power a big college campus or a small city of 50,000 homes.

Class 3: Utility Scale

The sweet spot utilities and huge data centers are excited about. Big enough to matter on the grid, small enough to replicate. Can power a large suburb or a mid-sized US city, about 200,000 homes.

Class 4: “LMRs” (Large modular reactors)

Still modular in approach, but as powerful as old massive nuclear plants. Can power a major metro area like Dallas, or roughly one million homes.

As we’ll see momentarily, most SMRs are made to be strung together to combine power. This blurs the lines between the classes. String 100 microreactors together and it’s a cluster that can power Zurich. Really, there’s no limit to how much energy an SMR supercluster can produce.

Unlike the concrete monstrosities of old nuclear, many SMRs are easy to transport. Imagine a self-contained power plant that runs for 5-10 years without refueling. No smoke, no emissions. Just clean, silent electricity.

Now imagine air-dropping that “power plant in a box” next to a data center in Texas, a remote mine in Alaska, or a village in Africa. Or even a frontline military outpost.

That’s what LA-based Radiant Nuclear is building. Its "Kaleidos" microreactor fits in a standard shipping container and can be hauled on a truck. More on Radiant below.

M = modular

Instead of building one giant, custom nuclear plant on-site, SMR parts are manufactured in a factory, shipped on a truck, then clicked together like LEGO bricks.

This is a big deal. Modularity allows for standardization. Most of America’s nuclear plants were built as one-off projects, each with a different design. It’s impossible to improve and build fast when the design is overhauled every time. “New nuclear” founders are choosing a few brilliant, safe, efficient designs and preparing to pump out as many as possible.

Making SMRs in a factory instead of on-site is also a game-changer. Anything built in a factory can be iterated on, optimized, and sped up. As I mentioned, 60-70% of the cost of old nuclear is financing expense, due in part to agonizingly long construction sagas. Nuclear’s Achilles’ heel is about to be a solved problem.

As a result, SMRs can be installed in months, not years. Aalo Atomics aims to deploy its microreactor in 60 days, factory to grid. Radiant is designing a reactor that can be delivered to a disaster zone and be running within a week.

If an SMR can be built and powered on within 1 year, it could easily cut the final cost of the electricity generated in half.  Imagine an early-adopter city whose electricity is 75% cheaper than its neighbor. People will flock to these energy-abundant oases... perhaps fleeing places like England, where skyrocketing energy prices have crushed industry and the middle class.

As I mentioned, many SMRs can be strung together to create a power source as big as needed. A growing city could start a few pods, then add more as demand grows. No more needing to commit $10 billion upfront.

R = reactor

All nuclear reactors make energy by splitting atoms (fission) to make heat, then using that heat to spin a turbine. What’s different with SMRs is the hardware around the heat.

Old reactors are like sprawling boiler rooms with miles of pipes, pumps, and valves spread out across a site the size of a football stadium. SMRs can shrink all that complexity into a small box. The reactor, coolant loops, and safety systems all live inside one thick steel vessel.

SMRs are even safer than old nuclear. In fact, most are "can't-fail, passively safe” systems for which meltdown is all but impossible.

One example: Many SMRs run on TRISO fuel. Instead of a fuel rod, the uranium is broken down into millions of tiny ceramic-armored grains. These stay intact up to 1,600 °C, far hotter than the reactor should ever get, because as the core heats up, the chain reaction automatically slows down.

Another: Some SMRs use molten salt fuel. The uranium fuel is dissolved into a liquid, which dumps into passively cooled tanks if it overheats.

In short, the safest SMR designs use the laws of physics to eliminate the classic meltdown scenario. If they get too hot, they’re automatically shut down and cooled without pumps, power, electricity, or operator intervention.

Two startups, Radiant and Valar Atomics, take this even further by having minimal or even no moving parts in their core reactor systems.

Nuclear’s ChatGPT moment

More progress has been made on nuclear regulation in the last year than in the previous 50 combined, both in the US and elsewhere.

Why?

Because big tech needs clean, reliable power, and it needs it yesterday.

One reason we turned our backs on nuclear is because we could. Thanks to efficiency gains, US energy demand hasn’t grown in 20 years. “Why build more if we’re using less?” was a semi-reasonable question.

That world is over. Every time you ask ChatGPT a question, you’re using an industrial-scale machine. Behind the scenes your words travel to cavernous data centers filled with rows of NVIDIA graphics chips. A single rack of NVIDIA H100s pulls more electricity than 50 American homes.

AI giants like Amazon, Microsoft, Google, and Meta have the GPU chips. What they lack is enough power to run them. In Virginia, the world’s largest data center hub, utilities warn AI could soak up half the state’s power by 2030. Ireland capped new builds because data centers were swallowing a fifth of national electricity demand!

When the internet scaled in the 2000s, we built server farms. To scale AI, we need power plants.

Nuclear has re-entered the chat because it’s the only energy source that can deliver enough clean, safe, round-the-clock electricity to feed AI. Nuclear is to AI what oil was to the Industrial Age. It’s the fuel for a new era of exponential progress.

Previously, utilities and governments were the only buyers of nuclear energy. They tend to be slow, constrained by regulation, and uninterested in innovation.

Big tech, on the other hand, is throwing big bucks at nuclear innovation. Microsoft inked a deal to help restart a reactor at the Three Mile Island plant in Pennsylvania. Amazon entered a 17-year contract with Talen Energy to power its data centers with nuclear. Meta inked a 20-year deal to buy power from Constellation Energy Corp.’s Clinton nuclear plant in Illinois.

Julia DeWahl, cofounder of nuclear startup Antares, told me tech companies are offering triple the going rate for guaranteed nuclear power. All of a sudden, nuclear has world-class, deep-pocketed customers in big tech.

Big tech also has lobbying power, which has helped blast open the regulatory logjam.

In the US, the prior administration was already pro-nuclear. The current one is even more so. In March, the President signed four executive orders aimed at freeing nuclear energy from its regulatory shackles. The five big changes:

1. Set a goal of quadrupling America’s nuclear reactors by 2050.
   

2. Speed up the development of “advanced nuclear” (read: SMRs) through pilot programs and streamlining environmental reviews. This orders the NRC to license new reactors within 18 months.
   

3. Orders the Department of Energy to approve at least three reactors by mid-2026. Trump wants three SMRs up and running for America’s 250th birthday – July 4, 2026.
   

4. Designates nuclear plants that power AI facilities as “defense-critical infrastructure,” opening a loophole to speedrun development for SMRs located next to AI data centers.
   

5. Lastly, but in our view most importantly, asks the NRC to reconsider its “As Low As Reasonably Achievable” (ALARA) regulation. You receive more radiation from eating one banana than you do from living next to a nuclear power plant for a year. Yet under the ALARA rule, even that isn’t safe enough.

Signing orders is step one. The hard part is turning ink into action inside the agencies.

I kept asking founders: “We’ve got the directives. When do the rules actually change?” 

That’s happening, but slowly. That’s why loopholes are our best bet for fast nuclear.

Big loophole No. 1: Department of Energy’s DOME program at Idaho National Lab: The DOE allows companies to build and operate demo reactors on federal land, bypassing years of permitting battles.

Big loophole No. 2: Department of Defense’s Project Pele: The Pentagon needs portable power for military bases. It’s funding microreactors small enough to fit on trucks and shipping containers.

While these and other loopholes - like building next to a data center or an existing nuclear plant - don’t fix the regulations, they do the next best thing: turn them into Swiss cheese so innovators can get through the holes.

The first American SMRs to turn on will be under the DOE or DOD programs. Weirdly enough, the fastest path to SMRs is through Washington DC – which has been nuclear’s worst enemy for 50 years.

Most importantly, these loopholes have reignited nuclear innovation.

My friend Eli Dourado of the Astera Institute asked me:

What’s the Starhopper of nuclear?

Starhopper was SpaceX’s stubby unimpressive test rocket. It was the rough draft for what eventually became the skyscraper-sized Starship.

Instead of taking ten years to design the “perfect” rocket on paper, SpaceX built something cheap and fast. They launched it, it blew up, they learned, they improved. That’s how breakthroughs really happen.

Nuclear never got its Starhopper. Imagine trying to innovate when you have to ask permission for nine years before you’re even allowed to test your idea?

Finally, real-life nuclear innovation has been reignited.

The rest of the world is also snapping out of its nuclear coma:

• The World Bank lifted its ban on nuclear financing after more than six decades.

• Germany walking away from its world-class nuclear fleet was one of the worst own goals in modern history. Prices spiked, factories shut down, and the country grew dangerously dependent on Russian gas. New Chancellor Friedrich Merz has admitted killing nuclear was a mistake. Will they restart the mothballed plants – at least the ones they didn’t demolish?

• UK Prime Minister Keir Starmer just announced a historic build-out of nuclear, including SMRs aimed at powering AI data centers. 

• Sweden committed to 2.5 gigawatts of new nuclear capacity by 2035. Denmark is in the process of ending a 40-year ban. Belgium scrapped its nuclear phase-out plan.

One big remaining roadblock for US nuclear is the National Environmental Policy Act (NEPA).

NEPA means years of lawsuits for anything needing federal approval. It covers not just the reactor but the surrounding area. You can design the safest reactor on Earth. But if every mile of power lines triggers a fresh round of bird-migration studies, you’ll never pour concrete.

For a true nuclear renaissance, we need to repeal or replace NEPA. Until that glorious day, loopholes are the way.

Meet the SMR contenders

The first company to turn on an SMR in the West will come from the following list. It’s roughly organized from “slow to fast.”

We’ll start by briefly covering the legacy majors. Then work our way down to deeper analysis of the startups run by innovators sleeping on mattresses in warehouses in order to get their SMRs up and running ASAP.

I expect one of the final three companies will win the SMR race.

Holtec

Piggybacking restarted reactors

Holtec is leading America’s first-ever nuclear restart. It wants to bolt two of its “SMR-300s” onto its soon-to-reopen Palisades nuclear plant in Michigan.

I like Holtec’s loophole of choice - bolting new reactors onto old mothballed sites. The tough part will be getting its SMR-300 built and licensed. It’s a “2030s” story.

Westinghouse

The Blue Chip of Nuclear

Westinghouse has built 67 currently operating reactors, the most worldwide. It’s now working on two SMRs: the Class 3 AP300 and the Class 1 eVinci microreactor, designed to power remote mines and military bases with minimal site preparation. eVinci was the first reactor to clear DOE’s preliminary safety review.

Westinghouse has been the fastest player in nuclear over the past 50 years, but that’s not saying much. I don’t think it has the DNA to truly move fast on SMRs.

GE Hitachi

Canada’s first SMR

GE Hitachi’s BWRX-300 Class 3 boiling water reactor is a slimmed-down version of its already NRC-certified design. It’s due to be built at Ontario Power Generation’s Darlington site in Canada by 2030. It will likely be the first contender to pour concrete later this year.

GE Hitachi is a dark horse to be the first Western company to deploy an SMR that supplies power to the grid. But its utility-scale size makes it a tortoise in a race against the cheetah-like innovators you’ll meet in a moment.

Rolls-Royce

British national champion with decades of expertise

Rolls-Royce has been making compact reactors to power Royal Navy nuclear submarines for decades. Now it’s jumping into the SMR race with a Class 4 reactor, the largest SMR covered here. It’s designed in five giant factory-made modules, which will be bolted together on site like a larger-than-life Lego set.

Clearly, Rolls-Royce can build reactors. But will it be allowed to? Regulations governing nuclear energy in most of Europe are even more nonsensical than in the US.

A glimmer of hope appeared recently: The UK government awarded Rolls a contract to build three SMRs as part of a £2.5 billion plan to kickstart nuclear there.

Other state champions

Because nuclear = national security

For the world’s most powerful countries, nuclear is too important to entrust to foreign companies. Nuclear is national security, energy, defense, and industrial policy rolled into one. That’s why most countries don’t leave nuclear decisions to the free market. They pick national champions.

Four national heavyweights are worth watching:

• Framatome (France)
  

• Rosatom (Russia)
  

• China National Nuclear Corporation (China)
  

• Korea Electric Power Corporation (Korea)

Of these, China’s CNNC is the leader in SMRs. Its Linglong One, located on Hainan Island, is the world’s first land-based SMR to begin construction. It is expected to achieve grid connection in 2026.

Countries with no nuclear history will buy the best foreign tech. For example, Korea’s KEPCO built four reactors in the UAE on time and on budget, a rarity in nuclear.

NuScale Power

Only SMR company with NRC blessing so far

NuScale is our first contender that doesn’t have decades of experience in old nuclear. It was founded in 2007 specifically to commercialize SMR technology.

NuScale holds the rarest card in nuclear: an NRC design certification, the only one of its kind. Its VOYGR “power module” is a Class 2 reactor designed to be passively safe and able to shut itself down without pumps. A dozen can be bolted together to reach nearly a gigawatt of output.

The world’s first SMR was supposed to be NuScale’s flagship project in Idaho. But it was terminated in 2023 after costs almost doubled to $9.3 billion. That’s after burning through two million engineering hours and $500 million just on licensing paperwork.

NuScale is a pioneer and was the first pure SMR company to go public in 2022. Its stock, which trades under the ticker “SMR,” has been red-hot. Unfortunately, NuScale hasn’t proven it can build at a competitive cost, or build at all. The best-case scenario is turning on its first reactor in 2030, and that’s if nothing else slips.

Nano Nuclear Energy

Seed-stage publicly-traded company

Nano Nuclear’s two Class 1 microreactor designs, ZEUS and ODIN, are each small enough to ride on a semitruck.

As far as I know, Nano is the only SMR maker that’s chosen uranium-hydride for its fuel core. This is a solid material that “breathes” by expanding as it heats. The reaction slows down as the core expands, meaning the physical properties of the core itself prevent dangerous runaway heat. It is inherently safer without needing to rely on complex cooling systems.

Nano gets a lot of attention because it is publicly traded. But it hasn’t achieved much of substance yet. It has no hardware in the ground and is still a bet on concept. Its 2030–31 deployment targets look optimistic.

TerraPower

The trailblazer

Founded by Bill Gates in 2008, TerraPower has been working on advanced reactors since some SMR founders were in elementary school.

TerraPower’s Natrium is a Class 3 sodium-cooled reactor paired with a molten-salt heat battery. The battery sets it apart. It can release enough stored heat to power half a million homes for more than five hours. That can vault Natrium into a Class 4 reactor in a pinch.

In 2021 TerraPower announced the first US SMR site at a retired coal plant in Wyoming. Its design has yet to be NRC-certified, but the company is building the rest of the plant while it waits for permission to build the nuclear part.

TerraPower will be remembered as the company that reopened the door to building advanced nuclear in America. But leaner startups will sprint through it first.

X-Energy

Bold bet on powering heavy industry

While most SMR companies are focused on generating electricity, X-energy is different: It aims to replace the smokestacks of heavy industry and defense.

Thirty percent of global energy use goes to “industrial heat.” Think steelmaking and other processes that require not just electricity but extremely intense and consistent heat. This is the hardest sector to decarbonize because solar and wind can’t generate industrial heat.

Most SMRs run too cool for industrial uses. X-Energy’s Xe-100 is a high-temperature gas reactor that pumps out steam at 560+ °C, almost twice as hot as most SMRs. That’s hot enough to facilitate jobs like chemical refining that are normally done by burning coal or natural gas.

Xe-100 is a Class 2 reactor, but can be bundled into a four-pack to upgrade to Class 3. Its first deployment is planned at Dow Chemical’s Seadrift plant in Texas in the early 2030s.

X-Energy is also building XENITH, a Class 1 SMR that the company believes will be capable of running for 20 years without refueling. It recently inked a deal with the Pentagon to deploy XENITH to military bases, taking advantage of the key DOD loophole that bypasses years of red tape.

X-Energy is one of the few contenders with real customers and a reactor design that solves a problem renewables can’t. It’s also one of only two nuclear startups backed by big tech, and is the only one Amazon has invested in.

If its Seadrift project in Texas and XENITH succeeds, X-energy will become America’s dual-use nuclear champion. A clear standout in the SMR race.

Natura Resources

The Permian’s nuclear moonshot

In 2016, a group of West Texas oil families donated $30 million to launch the Nuclear Energy eXperimental Testing (NEXT) Lab at Abilene Christian University (ACU).

Their goal: revive molten-salt technology that was abandoned in the 1960s. They created Natura Resources to turn the lab’s experiments into real reactors.

In 2022, Natura became the first U.S. team in decades to win federal approval to build a liquid-fueled molten-salt reactor. Molten-salt designs run at lower pressure and higher heat than today’s reactors, making them potentially safer and more efficient.

Construction is now beginning at ACU, where the campus lab will double as a live nuclear test site. What Natura learns there will feed into a larger commercial reactor planned for Texas A&M’s RELLIS campus by 2030.

Natura has a rare combination of advantages: it is one of only five startups to receive fuel allocation from the DOE, and holds a prized spot in the DOE’s Nuclear Reactor Pilot Program. It has a real shot at leading the molten-salt race.

Oklo

Buzziest, turns waste into fuel      

Oklo is probably the best-known pure SMR company. Sam Altman, founder of OpenAI, is a large shareholder.

Oklo’s Aurora Class 1 SMR can run on a type of fuel called HALEU (High-assay Low-Enriched Uranium). Many upcoming SMRs will run on HALEU, but only Oklo plans to source HALEU by recycling material from old spent nuclear fuel.

“What about the waste?” is the No. 1 pushback we hear from nuclear skeptics. Safely dealing with waste is a problem with many good solutions, as we wrote about here. Oklo plans to tackle it by turning nuclear waste into fuel.

Oklo’s business strategy is also unique. It doesn’t sell reactors. Instead it will own and operate its own plants and sell electricity to customers like data centers.

Oklo has been on a roll lately, achieving important milestones:

• The DOE approved the safety design for Oklo’s fuel‑fab facility at Idaho National Lab.
  

• Signed a letter of intent with data center provider Equinix for 500 megawatts of future power.
  

• Signed a 20-year power purchase agreement with Diamondback Energy near Midland, Texas.
  

• Awarded a contract to operate a reactor at Alaska’s Eielson Air Force Base.

For what it’s worth, several nuclear founders told me Oklo’s strengths are marketing and funding, not technology. We’ll find out soon enough. In 2022, the NRC terminated its review of Oklo’s Aurora due to “information gaps.” Oklo is now preparing its resubmission. This upcoming NRC resubmission, plus the upcoming pilot deployment of its microreactor for the Air Force in Alaska, is crucial. If those break its way, Oklo will be in pole position in the SMR race.

Oklo is the strongest and fastest-moving publicly traded pure SMR company, and it’s not particularly close (the others are NuScale and Nano). Its stock has rocketed over 1,000% in the last year.

Last Energy

Ruthlessly pragmatic, productizing proven design

Last Energy is building a school-bus-sized Class 1 pressurized-water reactor. This is the same technology that’s been used for 60+ years by old nuclear plants, but vastly shrunk down to fit in an SMR. Each unit is encased in 500 tons of steel shielding and buried in a shallow pit, making it invisible once installed.

When I met founder Bret Kugelmass in San Francisco, he summed up Last’s strategy bluntly: “No new reactor magic. Just take proven technology and manufacture it like a product.”

Dozens of Last Energy’s Class 1 reactors can be bolted together. It has signed a deal to stack 30 of them on a 200-acre site in Texas, to feed power directly to data centers. An NRC Early Site Permit is in motion.

Like Oklo, Last Energy plans to own its reactors and sell long-term power to customers. This strategy has helped it secure 55 deals already… because it’s easier to sell clean energy than a nuclear reactor.

This company lacks the hype of “new tech.” But that’s its edge. With 55+ units already contracted, it’s quietly at the top of the pack in actual deals. It could surprise everyone and win the SMR race, and has the potential to become a new global nuclear utility.

Kairos Power

First civilian reactor?

In the 50s and 60s, US national labs tested dozens of exotic reactor types. That work was “top secret” until it was declassified in the early 2000s, seeding today’s wave of nuclear startups.

Kairos is reviving one of the most promising designs: molten-salt cooling. Unlike water, molten salts do not expand when heated. That means the risk of a steam explosion is off the table, eliminating the need for a giant pressurized containment vessel.  Notably, this is different from nuclear startup Natura's reactor, which not only uses molten salts as coolant, but also to circulate fuel through its core.

In August Kairos announced its Class 2 Hermes reactor will plug directly into Tennessee’s grid by 2030. This is a major milestone, marking the first US utility to agree to buy power from an SMR. It also inked a deal to power Google’s datacenters in the region, making it one of only two contenders to have signed a deal with big tech.

Kairos has real hardware in the ground. Its Class 2 demo reactor is under construction in Oak Ridge, Tennessee, scheduled to be completed in 2027. It also has test rigs at Idaho National Lab and the Texas A&M RELLIS campus. This alone makes Kairos a clear leader in the SMR race, as most SMR companies haven’t installed anything yet. All reactor designs look good on paper. It’s only when you actually build that you figure out what works and what doesn’t.

Several nuclear insiders told me Kairos has what it takes to win. It will likely be the first NRC-licensed “civilian” reactor to power up in America. But it will trail the companies targeting DOE and DOD loopholes by 1-2 years.

Antares Industries

Blazingly fast new kid in nuclear

Founded in 2023 in Redondo Beach, Antares is barely two years old… yet it has already built a prototype of its reactor.

The company is going after what founder Jordan Bramble calls “strategic energy.” Its portable Class 1 SMRs are meant to power military bases, missile silos and remote mines.

Antares’ R1 is the smallest design of all SMRs covered in this Deep Dive, which should help it iterate fast. It is also designed to run itself through automated controls. Did someone say self-driving nuclear?

Antares is focused squarely on the DOE/DOD loopholes, which means its customers prioritize reliability, not cheap electricity. It was selected for the White House’s Nuclear Reactor Pilot Program, joining nine other startups in the race to power up an SMR by July 4, 2026.

In the SMR race, hungry, focused teams with no legacy baggage will win. Antares must be taken seriously, given how much it’s accomplished in just two years.

Valar Atomics

Clustering thousands of SMRs on a gigasite.

When you walk into Valar’s SMR prototype in El Segundo, you don’t feel like you’re inside a reactor. It’s more like a high-end video gaming studio, with neon yellow lights and a polished control center. Founder Isaiah Taylor insists on making his SMRs beautiful to give people the simple gut reaction: “I want one.”

We snapped this picture in front of Valar’s test reactor (Ward 0) when we visited recently:

Nuclear is in Isaiah’s blood; his grandfather worked on the Manhattan Project. He calls Valar the “anti-SMR, SMR company.” Instead of having hundreds of its SMRs dotted around the world, his vision is to build “gigasites,” with 1,000+ microreactors clustered together, pumping out far more power than any existing plant.

Isaiah’s gigasite concept is regulatory arbitrage. In nuclear, most of the regulatory cost and delay isn’t from certifying new designs. Instead, it’s from licensing the site, including dealing with dreaded NEPA.

Since it takes the same mountain of permits to license one reactor vs. thousands on the same piece of land…. why not cluster thousands there?  

If Valar’s vision is realized, these gigasites could produce more energy than the hungriest AI data center can consume or the most powerful battery can store. To store the extra energy, Isaiah plans to use cheap nuclear energy to make synthetic fuels.

In short, he plans to use ~900 °C nuclear heat to split water into hydrogen, then combine that hydrogen with captured CO₂. The result will be pulling carbon out of the air to synthesize gasoline and jet fuel, powered entirely by clean nuclear heat.

Talk about bold!

Valar is bypassing NRC licensing and exploiting the new DOE testbed carve-outs enabled by the recent executive orders. It also signed a deal with Utah to turn on what could be America’s SMR at the San Rafael Energy Research Center by July 4, 2026.

Valar built a non-nuclear version of Ward 0 in about 10 months. Now the goal is to build a new vessel every 10 weeks, which is about as fast as a reactor can be welded together.

Having met Isaiah several times, it’s clear Valar is a startup worth betting on. If Valar executes on gigasites, it’ll be the one to make nuclear cheap again.

Aalo Atomics

Building the gigafactory of nuclear

We’ve visited Aalo’s 40,000 sq. ft factory in Austin twice. The first time it was an empty shell, with nothing but a giant star-spangled banner hanging from the roof.

Nine months later we returned to find a full-scale unfueled prototype and a humming factory floor.

Aalo’s goal is to build a nuclear “gigafactory.” Think Tesla, but instead of cars rolling off the line, one Class 1 sodium reactor will be pumped out per week.

Every decision Founder Matt Loszak makes has one goal: turn on a reactor as fast as possible. The first time we met him, Aalo planned to power its reactors with a special TRIGA fuel. But Matt pivoted to using an “off-the-shelf” fuel, which is easier to obtain.

Aalo’s sodium reactor design is inspired by the MARVEL microreactor at Idaho National Labs – part of a DOE program that regulators have closely followed. Yasir Arafat, MARVEL’s chief designer, is Aalo’s cofounder & CTO. This pedigree + regulatory familiarity + fuel choice puts Aalo in the fast lane.

Five of Aalo’s Class 1 reactors can be wired together into the “Aalo Pod,” capable of powering roughly 50,000 homes or one AI data center. Aalo is looking at dropping Pods on federal land to power AI clusters under the recent executive orders. It’s also one of the ten startups selected as part of the White House’s plan to turn on several SMRs by July 4, 2026.

Its first reactor is due to power up under the DOME test program at Idaho National Lab next year. 

Among all SMR contenders, Aalo is the purest expression of the new nuclear: fast, modular, factory-built, and customer-obsessed. I believe it will meet its goal of turning on a reactor before July 4 next year.

Radiant Industries

Replacing diesel generators

Radiant is building Kaleidos, a Class 1 nuclear generator in a shipping container. Kaleidos is a sealed box that can be dropped almost anywhere by crane, run for five years without refueling, then be lifted out like a spent printer cartridge.

By staying Class 1 size and choosing air cooling, Radiant avoids many of the problems that delay larger SMRs. As CCO Ray Wert told me in Detroit: “Go big and you have to dig giant holes in the ground and install water pipes for cooling. You can lift our small, air-cooled reactors out like nothing was ever there.”

Radiant’s sales pitch is premium power. It’s not trying to sell low-cost energy, but reliability. Ray says: “Anyone saying SMRs will deliver the cheapest electricity is lying to you.” Hmm. Isaiah of Valar strongly disagrees, as we discussed on a recent podcast.

Ray also predicts:

“Kaleidos will be the first new reactor design to achieve a fueled test in over 50 years at the DOE’s testbed in Idaho next summer… and you can quote me on that.”

Radiant is also part of the Pentagon’s Project Pele and has a deal to deliver portable reactors to Air Force bases within three years.

Radiant’s ultimate goal is to replace diesel generators. I asked Ray to sum up Kaleidos in one word. He said: “portable.”

I think Kaleidos will be SMR’s first breakout hit.

Winner?

It’s clear that startups taking the DOE/DOD loopholes will be first to power up their SMRs.

I see a three-way tie: Valar, Aalo and Radiant.

These three have the perfect blend of moving fast, picking the right regulatory lanes, and plenty of capital to win the great American nuclear race.

I believe all three will turn on a reactor within the next 9-12 months.

If you’re as excited about SMRs as I am, here’s what to watch for next:

No. 1: DOE Pilot Program. Ten startups were selected. At least three should turn on by July 4 2026.

No. 2: Big tech nuclear deals. Building reactors to directly feed AI data centers is a fast path to deployment.

No. 3: Fuel allocations. If a startup wins an HALEU fuel allocation from the DOE, that’s a signal it’s hitting the right milestones.

Five startups (Radiant, Antares, Kairos, TerraPower, and Natura Resources) have already won an allocation. The process is ongoing, and the DOE plans to make 21 metric tons available by June 30, 2026.

Final word

Energy is the bedrock of innovation and civilization.

There is no such thing as a low-energy rich country:

Source: International Energy Agency 

I want my country, and yours, to keep moving up and to the right.

We can do it… but only with a lot more clean, cheap energy.

Every leap in living standards has come from climbing to a denser, cheaper, more controllable source of energy:

Fire → coal → oil → electricity → nuclear

Here’s to Atomic Age 2.0!