For the first time in nearly half a century, a brand-new nuclear reactor is about to go critical on American soil at Idaho National Laboratory. The last one — NRAD, the Neutron Radiography Reactor — fired up in 1977. Jimmy Carter was president. Star Wars was in theaters. And then, for the next 49 years, essentially nothing.
The reasons are well-rehearsed. Three Mile Island in 1979 froze public opinion. Carter’s 1977 ban on commercial reprocessing kneecapped the fuel cycle economics that fast reactors depended on. The Atomic Energy Commission was split into ERDA and the NRC, and the new regulator’s “as low as reasonably achievable” doctrine layered cost on cost on cost. By the late 1980s, building a reactor in America had become an exercise in legal warfare and bespoke on-site construction. The industry essentially stopped certifying new designs. Vogtle 3 and 4, completed in 2023 and 2024, came in at roughly $35 billion against an original budget of $14 billion — and they’re the only new reactors America has built this century. We’ve been living off Eisenhower-era engineering ever since.
That era is ending. And the race to be the 53rd reactor at Idaho Falls — the first new criticality there since NRAD in 1977 — is the cleanest signal institutional investors are going to get about who the serious players actually are.
The deadline that changed everything
On May 23, 2025, President Trump signed Executive Order 14301, directing the Department of Energy to authorize at least three advanced reactors and bring them to criticality by July 4, 2026 — the United States’ semiquincentennial. The mechanism is novel: pilot reactors operate under DOE’s Atomic Energy Act authority rather than the NRC’s notoriously glacial commercial licensing pathway. To my thinking, this is the government’s way of saving face after correctly recognizing that NRC has been the primary impediment to progress for the last half centruy. And to be sure, under Chris Wright’s leadership, DOE has been upstaging NRC left, right and center.
DOE selected eleven projects from ten companies in August 2025. No federal construction money was offered. Companies self-finance, and the prize is intangible — HALEU fuel allocations, a regulatory concierge team, accelerated NRC commercial licensing afterwards, and the marketing halo of crossing the line first.
The nuclear renaissance is on, and it’s on strong. More than $1.5 billion in private capital has flowed into these eleven projects in the last ten months, including some of my own. The EBR-II containment dome — a piece of nuclear history — has been repurposed into the world’s first dedicated advanced-reactor test bed (DOME) and opened nearly a year ahead of schedule. Antares Nuclear cleared a full DOE Documented Safety Analysis on April 6, 2026 in roughly ten months — a milestone DOE itself called “equivalent to receiving a license from the NRC.” For context, that process used to take half a decade. And to my thinking, Antares and DOE just proved that there was never any good reason for it to take so long.
Why this matters more than which company finishes first
Here’s what investors should internalize: it doesn’t matter whether Aalo, Antares, or Radiant wins the photo op by being first to criticality on or before July 4. What matters is the shortlist of companies who are still in the race at all.
Anyone with a marketing department can pitch a beautiful vision of the future of nuclear energy. Slide decks are cheap. What’s expensive is fabricating fuel, completing safety analyses, assembling a reactor, and convincing DOE to let you turn it on. Notice who isn’t in this race. NuScale — once the darling of the SMR sector, with the only NRC-certified small modular reactor design in America — sat this one out entirely. So did Holtec, Westinghouse (with eVinci, despite holding a DOME slot), Kairos, X-energy, and TerraPower. Some of those companies have legitimate strategic reasons. Some of them know better than to enter a race they can’t finish.
The race is separating the men from the boys. The frontrunners as of late April are:
Aalo Atomics holds the hardware lead. Its Critical Test Reactor (CTR) at INL was assembled in 40 days, fuel is en route, and CEO Matt Loszak says criticality is “weeks” away. PDSA approved March 4, 2026.
Antares Nuclear holds the regulatory lead — the only company with a full DSA in hand, now in final Readiness Review. CEO Jordan Bramble has stated flatly: “We will achieve k=1 before July 4, 2026.”
Radiant Industries is the only company attempting full-power operation rather than zero-power criticality. Its Kaleidos reactor is the inaugural tenant of DOME.
Behind them, Oklo’s 75 MWe Aurora-INL — the largest pilot project — is unlikely to make July 4 despite breaking ground last September. CEO Jacob DeWitte’s commercial timeline is late 2027. Last Energy, Deep Fission, Natura Resources, and Terrestrial Energy round out the cohort but are well behind on physical progress. Valar Atomics achieved a contested zero-power criticality at Los Alamos in November 2025, but its actual pilot site in Utah is unlikely to operate by July.
Einstein, factory production, and the $20/MWh dream
Now let me make the macro argument that should actually drive your investment thesis.
Our first and foremost goal should be to make nuclear cheaper than coal and gas. We’re never going to get there by doing the exact same thing (light water reactors) that we’ve been doing for the last sixty years, and expecting different results. Einstein was as clear on that point as he was in laying out the equation that explains how nuclear energy is possible to start with.
Bespoke on-site construction — every plant a custom megaproject with site-specific civil works, custom-fabricated pressure vessels, and ten thousand bespoke piping welds inspected by ten thousand bespoke regulators — is exactly why Vogtle cost $35 billion. It’s why Hinkley Point C in the UK is north of $50 billion and counting. It’s why the LCOE for a new gigawatt-scale reactor in the West has trended up over fifty years while every other technology — solar, wind, batteries, gas turbines — has come down.
The only economic model that fixes this is factory mass production. Standardized reactors, identical units, assembly-line manufacturing, certified once and replicated—not by the dozens or even hundreds. But by the thousands. This is how Boeing builds 737s. It’s how Toyota builds Camrys. It’s how every successful complex technology in the modern world has been scaled. And it’s a model that requires advanced reactor designs — small enough, modular enough, and standardized enough to leave the factory complete or near-complete, ship to site, and bolt down. The on-site work has to become trivial; the factory work has to become repetitive. We need to design nuclear energy that gets installed and configured on-site, not custom-built on-site.
This is the framework you need to evaluate every company in the race.
Aalo: factory-built powerplants for the mass market
Aalo Atomics is ultimately going after mainstream electricity. Their business plan begins with targetting data centers, because that’s where the money is right now. The Aalo-X is a 10 MWe sodium-cooled reactor using ordinary low-enriched uranium oxide fuel. But the strategic genius isn’t the reactor itself. It’s the manufacturing model. Aalo’s commercial product is the Aalo Pod: an entire 50 MWe nuclear power plant built in a factory and shipped anywhere it’s needed. Their initial target is the data center market, where hyperscalers will pay almost any price for firm clean power. Their long-term target is $20/MWh LCOE — competitive with combined-cycle gas and well below new-build solar plus battery storage.
That number is achievable only one way: factory mass production at scale. Aalo gets this. It’s why their entire business architecture — five Aalo-X reactors integrated into one Aalo Pod, designed for assembly-line fabrication, shipped on standard transport — is structured around manufacturing economics from day one rather than reactor economics. They’re not trying to build a better reactor. They’re trying to build a reactor that can be built better — at volume.
Antares: killing the diesel generator
Antares Nuclear is solving a completely different problem. The Mark-0 is a 500 kWt sodium heat-pipe-cooled microreactor — a transportable unit optimized for remote deployment. Antares is going after the diesel generator market: remote mining operations, military forward operating bases, isolated villages, off-grid industrial sites. These customers currently pay $0.50–$2.00 per kWh for diesel that has to be trucked, barged, or flown in.
Antares doesn’t have to beat the grid; it has to beat diesel — and that’s a much easier hurdle. Their Air Force ANPI selection at Joint Base San Antonio reflects exactly this thesis. It’s a smaller TAM than Aalo’s, but it’s a real, profitable, and underserved market.
Both companies could succeed. They are not competitors. They’re solving fundamentally different problems with appropriately scaled solutions. Aalo wants to power Phoenix. Antares wants to power the Red Dog mine in Alaska and Marine forward operating bases in the Pacific. Both visions are important. Neither is a substitute for the other. But the fuel choice that underlies them tells you something critical about each company’s path to commercial economics.
The TRISO problem: a $20,000 fuel pellet looking for a market
Here’s the part of the story that almost nobody talks about, and that institutional investors need to internalize.
Most advanced reactor designs — Antares Mark-0, Radiant Kaleidos, Oklo Aurora variants, Kairos Hermes, X-energy Xe-100, Project Pele, BWXT’s microreactors — use TRISO fuel: tiny uranium kernels coated in three ceramic layers, embedded in a graphite or silicon carbide matrix. TRISO is genuinely impressive engineering. Each particle is its own miniature pressure vessel. They survive temperatures that would melt conventional fuel. They’re nearly impossible to weaponize. The proliferation story is excellent.
The problem is the price tag.
Conventional LEU UO₂ fuel — the kind in every commercial light-water reactor in America, and the kind Aalo is using — runs roughly $2,500–$3,500 per kilogram of finished fuel, all-in: yellowcake, conversion, enrichment to 3-5% U-235, fabrication into pellets, and assembly into fuel rods. Mature, commoditized supply chain. Multiple competing fabricators (GE/GNF, Westinghouse, Framatome). The fuel cost contribution to LCOE for a conventional reactor is roughly $5–$8 per MWh — a small fraction of total levelized cost.
HALEU TRISO fuel today costs $30,000 to $45,000 per kilogram of finished fuel. Read that again. A kilogram of finished TRISO costs roughly 10–15x what a kilogram of finished conventional LEU fuel costs. The Nuclear Innovation Alliance estimates HALEU UF₆ alone — before any fabrication into a usable fuel form — at around $19,000/kg. Converting that into TRISO particles, with their four-layer ceramic coatings and stringent quality requirements, adds another $10,000–$25,000/kg on top. Even the most aggressive long-term industry projections — assuming high-volume production at facilities that don’t yet exist — put mature TRISO costs at maybe $5,000–$7,000/kg, still a 2–3x premium over conventional LEU.
Why is TRISO so expensive? Three compounding factors. First, the HALEU enrichment premium: 19.75% U-235 enrichment requires roughly six to seven times more separative work units than 5% LEU enrichment, and each SWU in a HALEU cascade is itself more expensive than an LEU SWU because of criticality safety constraints in the centrifuge cascades. The Nuclear Innovation Alliance estimates LEU SWUs at roughly $150 versus HALEU SWUs at roughly $1,000. Second, the fabrication process: each TRISO particle must be coated with four distinct ceramic layers in fluidized-bed furnaces, with quality assurance on billions of particles per fuel load. There are essentially no commercial-scale TRISO production lines in the world. Third, the supply chain: until recently the only operating commercial HALEU producer was Russia, and Centrus’s Piketon facility produces a few hundred kilograms per year against a commercial-scale demand that will need thousands of tons.
For a microreactor like Antares’ Mark-0 — small core, long refueling cycle, premium customer paying to displace $1.50/kWh diesel — TRISO economics work fine. The fuel is a small fraction of the cost of replacing a diesel genset that has to be airlifted to a Pacific island. For a remote mining village that currently spends $50 million a year on diesel logistics, paying a premium for a fuel form that can sit untouched in a sealed core for ten years is a feature, not a bug.
But for a reactor trying to compete head-to-head with combined-cycle gas at $20/MWh on the open electricity market? TRISO is a non-starter. A 10–15x fuel cost premium pushes the fuel component of LCOE from roughly $5/MWh to potentially $50–$75/MWh — by itself more than the entire delivered cost of natural-gas-fired electricity in most US markets. You cannot get to $20/MWh with $30,000/kg fuel. The math simply does not close.
This is why Aalo’s choice of conventional LEU UO₂ fuel matters more than most people realize. It is the choice of a company that has done the LCOE arithmetic and concluded that mainstream-market nuclear has to use mainstream-market fuel. Conventional LEU is supplied by half a dozen vendors in a mature global market. There is no enrichment bottleneck. There is no fabrication bottleneck. The fuel cost is a rounding error in the LCOE stack. And the entire commercial fuel cycle infrastructure — transportation casks, fabrication plants, regulatory frameworks, spent fuel handling — already exists and has existed for sixty years.
This is not to say TRISO is bad technology. It is excellent technology for the right application. But the right application is not “powering modern civilization at competitive cost.” It’s powering applications where the alternative is much worse, and customers will pay accordingly. Investors should be very clear about which market a given TRISO-based reactor is really addressing — and skeptical of any pitch that combines TRISO fuel with claims about beating gas-fired generation on price.
What to watch
Marketing departments talk. Reactors going critical walk. The next ten weeks will tell us which advanced nuclear companies have built the engineering, regulatory, and supply-chain machinery to actually deliver — and which ones have only built pitch decks. The frontrunners are close enough that the order of finish is genuinely uncertain. Watch the DOE Reactor Pilot Program weekly updates, watch fuel delivery announcements, and watch which companies clear their DSA reviews.
But when you build your shortlist, evaluate each company on three axes, in this order: Are they actually delivering hardware? What market are they really addressing? And does their fuel choice match their stated cost target? A TRISO-fueled reactor pitching $30/MWh power to data centers is not a serious commercial proposition. A TRISO-fueled reactor pitching diesel-displacement at $0.30/kWh to remote operators absolutely is. An LEU-fueled reactor pitching $20/MWh to hyperscalers might just rebuild the American power sector.
The 53rd reactor at Idaho Falls is coming very soon. The 54th, 55th, and 56th matter even more. The investable thesis is the shortlist of companies still in the race when the dust settles, with the right fuel for the right market — because those are the ones with a real shot at building the second atomic age.
Spot‑on. No notes.
Turbines seem to be the major chock point. GE Vernova for example has a back log of years. SMRs are useless without the ability to turn heat into electricity.