The nuclear SMR race is down to the wire

Our final SMR power rankings

In today’s Diary:

• 34 days on the shot clock

• 5 steps to real electricity

• Reactors buried miles deep

• Bill Gates finally breaks ground

• Will China win?

Hey Rational Optimist,

Get excited. We’re 34 days out from July 4^th, and the race to turn on America’s first nuclear small modular reactor (SMR) is in the final stretch.

In January we published our first nuclear SMR Power Rankings. We ranked 16 companies that were racing to turn on their reactors by the DOE’s ambitious July 4^th 2026 deadline, America’s 250^th birthday.

Since then the race has crystallized. 3 companies have pulled away from the pack. I’m flying out to visit all 3 frontrunners in late June, days before the deadline. I’ll report back.

Today let’s update our SMR power rankings for the final stretch.

If you’re new to nuclear, read this to understand why it’s a rational optimist’s dream energy source. Nuclear is safe, clean and abundant. And it’s on its way to being cheap as regulators finally get out of the way and let innovators do their thing.

What exactly does it mean to “turn on” a reactor?

Understanding this will help you cut through the hype, of which there is plenty in the SMR space. “We turned on our reactor” can mean five completely different things.

Picture a ladder with five rungs:

Rung 1: The prop.

A full-size reactor with everything except the nuclear material. There’s a hole where the uranium is supposed to be. Engineers plug it with a “dummy” made of silicon carbide or plain steel.

It looks like a reactor but can’t start a chain reaction because there’s no nuclear fuel inside.

Rung 2: Critical.

Real uranium goes in, and the reactor achieves a self-sustaining chain reaction. This is what companies are targeting for the July 4^th deadline.

But critical does not mean powerful. At this point the reactor throws off about as much heat as a few light bulbs.

Rung 3: Turning up the dial.

After proving the chain reaction works, the crew now creeps the power upward in small deliberate steps to make sure everything works as intended.

Rung 4: Real electricity.

The reactor can finally heat fluid hot enough to spin a turbine and generate electricity. This is what most people imagine when they hear “we turned on a reactor.”

Rung 5: Paying customers.

The reactor sells power to the grid or a data center. This is the only rung that matters to your electric bill.

The frontrunners are racing for the second rung by July 4. Going critical is a huge milestone we should celebrate. It’s also only a few light bulbs’ worth of heat, not power on a grid.

Rung 4 is a 2027 story at the earliest. Selling real power to customers is likely another year beyond that.

It makes sense to separate the nuclear race in two:

A sprint and a marathon.

Which race a company is running comes down to which regulator it answers to. You can’t talk about nuclear without talking about regulation.

The traditional path runs entirely through the Nuclear Regulatory Commission (NRC). Historically that meant years and millions of dollars spent just on paperwork. NuScale spent $500 million and 2 million labor-hours just to get NRC design certification.

The fastest movers found a way around the NRC’s regulatory quicksand: the DOE Reactor Pilot Program. Run your reactor on federal land, and the clock drops from a decade to about a year. This is the only reason a July 4 deadline is even thinkable.

That fork in the road splits the race into two:

• The Sprint: the microreactor race to criticality this year through the DOE pilot.

• The Marathon: the race to build utility-scale reactors in the 2030s. These companies took the slow NRC road from day one.

For the sprint, the finish line is only 34 days out.

Back in January, ranking the 16 contenders was mostly an art. We’d visited and talked with most of the leaders. But there was no way to know who was truly ahead.

Now the leaderboard is clear thanks to one milestone with an ugly bureaucratic name.

After the DOE reviews a company’s complete safety case for a finished design, it approves the Documented Safety Analysis (DSA). This basically says, “We think your reactor is safe. You can put real uranium in now.”

So a company either has this government stamp or it doesn’t. The three leading contenders all hold a DSA.

Between holding a DSA and a reactor going critical, three things still must happen.

The big one is the Readiness Review. The DSA certifies the reactor; the Readiness Review certifies the team. The way I think about it: the DSA says the airplane is good. The Readiness Review greenlights the pilots.

Get that stamp, and the DOE says the magic words: start-up authorized.

Next is the part everyone’s excited for: real uranium goes in, and the crew slowly coaxes the reactor up to operational criticality (Rung 2).

The first company to achieve that wins the July 4^th race.

Title favorites: the sprint

Our top 3 all hold a DSA and are in the process of acquiring a Readiness Review. Here’s how they stack up.

#1: Valar Atomics (retains #1 spot)

Last November Valar became the first startup ever to split atoms. It is the only company in the race that’s achieved “cold criticality.” They did so on a small benchtop test core called NOVA at a national lab.

Valar’s Ward 250 reactor is sitting at the San Rafael Energy Lab in Utah with DOE inspectors walking the floor for its Readiness Review, having received DSA approval in April.

#2: Antares (jumps up from #5)

Antares was the first company to clear a full DSA, and it’s been grinding through the final Readiness Review since early April.

The fuel for its Mark-0 reactor is already being fabricated. Its factory is doubling in size, and it just won a slot to put a prototype on a Texas military base. Antares is the regulatory pacesetter.

#3: Aalo Atomics (holds #3)

Aalo earned its DSA in late April and has the most finished hardware of the three: a complete reactor building at Idaho National Laboratory (INL), and a reactor built at its Austin HQ to be shipped to INL.

What Aalo plans to take critical by July 4 is a separate, zero-power test reactor. The full Aalo-X plant that generates electricity is due online in 2027.

Strong contenders but won’t make July 4th

#4: Radiant

In April Radiant moved into DOME, the only purpose-built, fueled microreactor test building in America. Radiant is its first and so far only tenant. It also landed a huge deal with data center giant Equinix, which preordered 20 units.

Radiant has dropped out of the leading pack because it lacks DSA approval. But they did get preliminary DSA approval in February, making the only company with permission to conduct the first full power test at DOME. As a result, once they get their DSA, they could leapfrog ahead of competitors.

I think Radiant will be one of the first to turn on a real full-power SMR. Just not by July 4.

#5: Natura Resources

Natura is the only company in the field holding an NRC construction permit for the first-ever liquid-fueled reactor. But it recently pivoted from the slow NRC route to the DOE fast-track.

Its building at Abilene Christian University is finished. But its reactor isn’t fueled, and it still has to clear all the regulatory hurdles.

#6: Oklo

In May the NRC pre-approved one piece of Oklo’s safety paperwork. Real progress, but Oklo is still years from being allowed to break ground on a reactor.

The brightest spot for Oklo is its isotope-reactor subsidiary, Atomic Alchemy. Atomic isn’t trying to make electricity, but a 15-megawatt-thermal pool of water to produce medical isotopes. Isotopes are the radioactive working ingredient in cancer treatments, and are used to sterilize most medical equipment in hospitals.

Alchemy is working on this in one of my favorite places, Proto-Town, targeting July 4 criticality.

#7: Last Energy

Last Energy is building small versions of the same boring pressurized-water reactor that already powers two-thirds of America’s nuclear fleet.

Earlier this month it withdrew its application for a proposed reactor site in Bryan, Texas. Texas A&M says the main RELLIS pilot where Last Energy will operate is on schedule. A 2027 story at the earliest.

#8: Deep Fission

Deep Fission wants to bury a 15-megawatt reactor a mile underground in a 30-inch borehole and let the Earth be the safety barrier.

In March Deep Fission walked back its July 4 deadline and effectively dropped out of the sprint. It’s actively drilling 6,000 ft test wells at its demo site in Parsons, Kansas. It also filed to go public under the ticker FISN.

Title favorites: the marathon

These are the companies aiming to power America’s cities and data centers in the 2030s.

They need two main regulatory approvals:

• NRC construction permit, aka permission to build. This usually takes 3 to 5 years of safety review… and almost no one has one yet.

• NRC operating license, aka permission to run. Granted after the reactor is built and tested. Nobody has one of these golden tickets yet.

#1: TerraPower

TerraPower recently won the first construction permit ever issued by the NRC for a commercial non-light-water power reactor. Last month it broke ground on the nuclear part of its Natrium plant in Kemmerer, Wyoming, making it the only utility-scale reactor under construction in America.

Bill Gates founded TerraPower in 2008. It’s great to finally see the oldest SMR contender break through.

#2: GE Hitachi

In April GE’s Canadian partner Ontario Power Generation set the reactor base for the first SMR officially under construction anywhere in the G7.

The Tennessee Valley Authority’s NRC construction permit application for GE’s reactor is currently under review.

#3: X-energy

The money story of the year. In April the Amazon-backed company priced the largest nuclear IPO ever at roughly $1 billion.

In February its TRISO fuel subsidiary won the first new fuel-fab license in nearly 50 years. And its four-reactor project with Dow Chemical in Texas just cleared a key environmental review.

#4: Kairos Power

Kairos was a Title Favorite. But last month the NRC granted Kairos a 28-month extension on the construction deadline for its demo reactor, pushing first operation toward 2028. Kairos drops out of the 2026 conversation.

But it’s still a top marathon contender, having broken ground on the first power-producing SMR with an NRC construction permit. Kairos also holds the one thing other contenders would kill for: a power-purchase agreement with the Tennessee Valley Authority.

Wild card: BWX Technologies

For 70 years BWX Technologies (BWXT) has built the reactors that power every US Navy submarine and aircraft carrier. BWXT is now building a reactor to power US military bases.

BWXT’s Project Pele reactor is small enough to pack into four 20-foot shipping containers. BWXT is currently assembling it at Idaho National Lab, with first electricity production targeted for 2028. The clear frontrunner in the “defense” lane.

Stragglers

NuScale finally got its Romanian project funded in February, but the schedule slipped to 2033. Its US project died in 2023 because the math didn’t work.

Holtec is working to restart the shuttered Palisades nuclear plant in Michigan and then build two SMRs next door. The Palisades restart date slipped to mid-2026 with more delays expected. It will get over the line, but the SMRs are gated on getting Palisades running first.

Rolls-Royce had the strongest international run in the Marathon, inking contracts for SMRs at sites in Wales and the Czech Republic. But Europe remains the world’s hardest place to build big things. I’ll believe this is real when concrete is being poured.

Westinghouse’s eVinci was supposed to be the first reactor in the DOE’s DOME test bed in spring 2026. Instead Radiant took the DOME slot, and eVinci’s first experiment now slips to 2027.

3 SMR predictions as we come down to the wire:

1. Valar, Antares and/or Aalo will hit zero-power criticality (Rung 2) by July 4 or soon after. I’d rank them in that order.
   

   The most important thing to watch now is who passes Readiness Review first. Grid electricity (Rung 4) is a 2027 story.
   

2. The first land-based SMR to power a grid anywhere could be Chinese. CNNC’s Linglong One has been under construction since 2021 and is targeting commercial operation later this year.
   

3. The most immediate opportunity for nuclear to power more AI data centers (Rung 5) will come from restarting old plants, not new SMRs. Constellation’s Three Mile Island Unit 1 is being recommissioned to deliver power to Microsoft in 2027. Holtec’s Palisades should be next in line.

No matter who wins, we’re all winners in the nuclear revival. July 4^th is the starting pistol for a decade-long journey to deliver floods of clean, abundant, cheap nuclear power to the US grid.

We need it.

Thanks for reading and please send this to anyone who may be interested.

We need more rational optimists in the world!

Hit “like” and “restack” to get the word out.

—Stephen McBride