$OKLO, $XE, $SMR and $IMSR: Four Advanced Nuclear Business Models, Compared

The distinction that organises everything: who owns the plant

Advanced nuclear is usually sorted by coolant. Sodium, helium, molten salt, light water. It is the wrong first cut for an investor, because the coolant determines the engineering problem while the business model determines the cash flow, the capital requirement and the point at which a shareholder finds out whether any of it worked.

There are only two real answers to the question “what does this company sell”, and a third that sits between them.

The consequences are not subtle. Under Model B, a company can report tens of millions of dollars of revenue before a single reactor exists, because engineering hours are billable. Under Model A, a company reports essentially nothing until the first plant is energised, and then reports a stream. Comparing their income statements without holding this in mind produces conclusions that are exactly backwards.

The scorecard: six dimensions that actually separate them

Reactor specifications are widely published and rarely decisive. What separates these four is a shorter list: what they sell, what a regulator has actually granted them, what fuel they need and whether it exists, whether anyone has signed something binding, how much money they hold, and whether any hardware has ever gone critical.

Money: who has it, who is spending it, and who is being paid

Three charts settle most of the argument about relative scale. The first is what the market is paying. The second is what each company actually holds. The third is the one that surprises people.

X-Energy overtaking Oklo is the detail worth sitting with. The company with the opposite business model, listed for barely three months, is carrying the larger valuation. Two structurally different bets on the same electricity demand, priced almost identically.

Two further details deserve a mention because they distort any naive screen. X-Energy’s reported net loss for the quarter was $166.2 million, of which $109.0 million sat in other expense and was driven by a non-cash mark-to-market on warrants — an artefact of the pre-IPO capital structure rather than an operating event. And NuScale’s operating cash outflow of $372.9 million for the first half of 2026 dwarfs its $96.7 million net loss, because accounts payable and accrued liabilities were run down from $286.5 million to $19.8 million over the period. Neither number means what it appears to mean at first glance.

The capital expenditure row is the one to keep. Oklo spent roughly $94 million on physical assets in a single quarter and $126.9 million across the half, against $33.2 million in the whole of 2025. NuScale spent under half a million in the same quarter. That is the build-own-operate model and the licence-the-technology model showing up in cash, exactly as the theory predicts. Nothing about it tells you which is the better business. It tells you which one is currently converting shareholder money into steel.

The regulatory ladder: what has actually been granted

Press releases about regulatory progress are written to sound like approvals. Most of them are not. A useful hierarchy runs roughly like this, from weakest to strongest evidence that a plant will be permitted to operate.

Against that ladder, the ranking is not what market capitalisation would suggest.

Fuel: the constraint that quietly sets the schedule

A reactor design is worthless without qualified fuel in the right form at the right enrichment. This is where the four separate most cleanly, and it is the dimension most likely to move a first-power date by years.

The two companies above the line are exposed to a bottleneck that no amount of engineering talent can dissolve. United States commercial HALEU capacity is being built now, not decades ago, and it is small. Oklo has responded with a deliberately diversified fuel strategy: five metric tons of material recovered from previously irradiated fuel awarded from Idaho National Laboratory, a non-binding letter of intent with Centrus for domestically produced HALEU with deliveries expected to begin in 2029, advanced negotiations under the Department of Energy’s surplus plutonium programme, and a recycling facility in Tennessee on a roadmap of up to $1.68 billion. That is four routes because no single route is dependable.

X-Energy has taken the opposite approach and built the bottleneck itself. Its TX-1 facility at Oak Ridge, Tennessee received a special nuclear material licence from the NRC in February 2026 — described as the first for a Category II facility in the United States, with an initial forty-year term — and construction is targeted for completion in the first half of 2028. Owning the fuel plant converts a supply risk into an execution risk, which is a genuine strategic difference rather than a marketing one.

NuScale and Terrestrial Energy simply do not have this problem. Fuel under five percent enrichment is what the world’s existing fabricators already make. It removes a category of delay, and it also removes a moat: anyone can buy that fuel, so nobody is differentiated by having secured it. Terrestrial Energy adds a second wrinkle by replacing the entire core unit every seven years, which is the basis of its recurring revenue model and also a recurring logistics and regulatory obligation for whoever owns the plant.

How to rank a contract, because the headline number is almost never the commitment

Every company in this sector publishes a pipeline in gigawatts. Those numbers are not comparable to each other and are rarely comparable to anything. A workable hierarchy, strongest first:

1. A power purchase agreement with a financed project behind it. None of the four has publicly disclosed one for a commercial plant.
2. A prepayment. Money that has actually moved. Oklo’s January 2026 prepayment agreement with Meta, tied to a 1.2 gigawatt campus in Pike County, Ohio, is the clearest example in the group: the mechanism funds fuel procurement for the first phase.
3. A development agreement with a named site and a filed licence application. The Dow relationship behind X-Energy’s Seadrift project qualifies, with the caveat that Dow may terminate for convenience.
4. A framework or master agreement with a capacity number. Oklo’s 12 gigawatt master power agreement with Switch, running to 2044. Amazon’s options on more than 5 gigawatts of X-Energy projects through 2039. Large, real, and contingent.
5. A letter of intent or memorandum of understanding. Talen with X-Energy, Riot Platforms with Terrestrial Energy, Equinix and Prometheus with Oklo.
6. An exploratory collaboration. Press-release stage.

Applied honestly, the picture changes. X-Energy discloses an aggregate pipeline of 144 reactors and roughly 11.5 gigawatts across the United States and the United Kingdom as of March 31, 2026 — but that figure assumes every customer exercises every contingent right it holds. Oklo’s 12 gigawatt Switch relationship is a single non-binding framework with an eighteen-year horizon. Neither is a forecast. Both are a statement about how much interest exists at a price nobody has had to pay yet.

The one genuinely binding-looking item in the group is not a power contract at all. It is the notice of intent to award issued by the Defense Logistics Agency on behalf of the Department of the Air Force, tentatively selecting Oklo to supply electricity and at least 5 megawatts of power plus steam to Eielson Air Force Base in Alaska. Five megawatts is a rounding error against twelve gigawatts. It is also the closest thing in this comparison to a customer who has started a procurement process rather than issued a press release.

$OKLO — Oklo Inc.: the only one that wants to own the plant

Oklo is the outlier by construction. It intends to design, build, own and operate its Aurora powerhouses and sell electricity and heat under long-duration agreements, rather than license a design to a utility. Everything distinctive about the company follows from that choice, including the parts that are uncomfortable.

The upside case is straightforward. If the model works, one licensed design becomes a fleet of company-owned assets producing recurring revenue with fleet economics, and the company captures the operating margin rather than a one-off licence fee. The Aurora is a sodium-cooled fast reactor in the 15 to 75 megawatt electric range with scope for larger configurations, deliberately sized to sit next to a load rather than serve a regional grid.

The cost of that choice is capital, and the second quarter of 2026 is where it became visible. Capital expenditure was roughly $94.1 million in the quarter and $126.9 million across the first half, against $33.2 million in all of 2025. Property, plant and equipment on the balance sheet rose from $42.3 million at the end of December to $176.2 million at the end of June. Operating expenses reached $74.4 million in the quarter, 2.7 times the year-earlier figure, with research and development up 244.2% to $39.5 million.

The quarter also carried the first revenue in the company’s history, $1.210 million. It is worth being precise about what that is, because it will be widely misreported: the 10-Q states that revenue and cost of sales resulted principally from the businesses acquired during the half. Oklo bought ARMEC on June 4 for $20.5 million and Creative Engineers on June 15 for $12.9 million, for precision machining and sodium-systems chemical process engineering respectively. The revenue is their engineering work. It is not power, it is not isotopes, and it is not a run rate.

What Oklo has that the others do not

On August 5, 2026 the Groves Isotope Test Reactor achieved first criticality, announced the following day, less than a year after groundbreaking. The sequence is documented and unusually fast: a Department of Energy agreement executed on January 7, safety design agreement approved March 17, preliminary safety analysis May 27, final safety analysis June 30, startup authorization July 23, criticality August 5. Six authorisation steps in seven months, on private land, under the Reactor Pilot Program.

Groves is a low-power test reactor for radioisotope work. It does not feed a grid and it is not an Aurora. But among these four companies, Oklo is the only one that has taken any hardware critical, and the organisational capability that produces — nuclear safety analysis, quality assurance, commissioning, operator qualification, readiness review — does not transfer from a slide deck.

What to watch, and what would break the case

Three of the five Department of Energy steps for the Idaho Aurora remain. There is no accepted NRC combined licence application, which is what commercial non-federal sites require. Fuel depends on a five-ton allocation plus three routes that are all still non-binding. And the balance sheet was bought with equity: 23.1 million shares issued in the first half for $1,851.9 million net, taking the count from 160.5 million to 185.1 million, a 15.3% increase in six months. The average net price realised fell from about $95.50 in the first quarter to about $62.55 in the second.

$XE — X-Energy: revenue today, and a fuel plant of its own

X-Energy is the newest listing in the group and, on the August 6 close, the largest by market capitalization. It sold 50,892,857 Class A shares at $23.00 in an offering that closed on April 27, 2026, raising approximately $1.1 billion net of $67.3 million of underwriting discounts.

Its reactor is the Xe-100, a helium-cooled high-temperature gas reactor producing 80 megawatts electric or 200 megawatts thermal per module, configured as a standard four-module plant of 320 megawatts electric. It uses TRISO pebble fuel enriched to 15.5% and is designed to move between 40% and 100% output in minutes, with steam temperatures up to 565 degrees Celsius. The high-temperature output is the commercial point: it addresses industrial process heat, which is a market conventional light-water reactors serve poorly.

The business model is intellectual property and services. X-Energy charges a technology fee for the Xe-100 licence, then sells site characterisation, project planning, procurement and construction support, and operations and maintenance services across a plant life of sixty years or more. Separately, TRISO-X converts customer-supplied HALEU into pebble fuel as a fabrication service, without taking uranium inventory risk. The company does not build, own or operate the plants.

Why it has revenue when the others do not

Because engineering hours for Dow, Amazon and the Department of Energy are billable now. First-quarter services revenue was $39.9 million against $17.1 million a year earlier, plus $3.5 million of grant income. That growth is real. It is also, on current cost structure, unprofitable: direct costs of $65.4 million exceeded revenue, and total operating expenses of $109.5 million produced a $66.1 million operating loss. The headline $166.2 million net loss is inflated by a $109.0 million non-cash warrant mark-to-market tied to the pre-IPO structure, and should not be read as a cash figure.

The fuel plant is the real differentiator

TX-1 at Oak Ridge, Tennessee began construction in October 2024 and is targeted for completion in the first half of 2028, sized for the fuel requirements of the first eleven Xe-100 reactors at full rate. In February 2026 the NRC issued its special nuclear material licence, described as the first for a Category II facility in the United States, with an initial forty-year term. Among these four, X-Energy is the only one building its own fuel fabrication capacity at commercial scale.

What to watch, and what would break the case

The Dow construction permit is the gating event: the NRC review is expected to complete around the end of 2026 with the permit anticipated in the first quarter of 2027. Dow can terminate the underlying agreements for convenience, which makes the anchor customer relationship weaker than the headline implies. Amazon holds 65,836,948 Class A shares, roughly 22.9% of that class, and its options on more than 5 gigawatts through 2039 are contingent rights rather than orders. The IPO lock-up expires on September 1, 2026. And first commercial delivery of an Xe-100 is guided to the early 2030s, later than Oklo’s stated 2028 ambition.

$SMR — NuScale Power: the strongest regulatory position, and a cautionary tale

NuScale is the oldest story in the group and the one that has already been through a full cycle of expectation and disappointment. Its module is a light-water reactor using natural circulation with no primary pumps, originally rated at 160 megawatts thermal and 50 megawatts electric, uprated to 250 thermal and 77 electric. A twelve-module plant reaches roughly 924 megawatts electric.

Its regulatory achievement is genuinely unmatched here. The 50 megawatt design received a full NRC design certification through rulemaking, effective February 21, 2023, the first small modular reactor design ever certified in the United States. The uprated 77 megawatt design received a standard design approval announced on May 29, 2025 — a real but lesser milestone, and a step below full certification.

The commercial structure is unusual and matters

ENTRA1 Energy holds exclusive global rights to commercialise, distribute and deploy NuScale products. The plants are developed, owned and operated by ENTRA1, using NuScale modules. NuScale supplies technology and long-lead equipment. An investor buying $SMR is therefore buying a component and licensing business whose route to market runs through a single exclusive partner, not a company that sells directly to utilities.

The financials show what that looks like before deployment. Second-quarter revenue was $75 thousand and first-half revenue $640 thousand. Net loss including non-controlling interests was $50.1 million in the quarter and $96.7 million in the half. Capital expenditure was $1.95 million across the half, which is roughly two percent of what Oklo spent, exactly as an asset-light model implies. Liquidity of $1,893.0 million at June 30 is second-largest in the group and, against that burn, substantial.

One figure needs explaining before it misleads: operating cash outflow of $372.9 million for the half against a $96.7 million net loss. The gap is a working-capital unwind, with accounts payable and accrued liabilities falling from $286.5 million to $19.8 million. It is a settlement of obligations, not a step change in the burn rate.

The cautionary part

In November 2023 the Carbon Free Power Project with a Utah municipal power group — NuScale’s flagship domestic deployment — was terminated by mutual agreement, with the Department of Energy cost-share reduced from $232.8 million obligated to roughly $50 million through an asset transfer. A certified design did not survive contact with project economics. That episode is the single most instructive data point in this entire comparison, and it applies to all four companies rather than to NuScale alone.

Current activity centres on the RoPower project at Doicești in Romania, where the government approved a final investment decision on February 12, 2026, and on a framework between ENTRA1 and the Tennessee Valley Authority. Company guidance points to deployment by 2030.

$IMSR — Terrestrial Energy: the annuity model, at the earliest stage

Terrestrial Energy listed on Nasdaq on October 29, 2025 after completing a business combination with HCM II Acquisition Corp, raising more than $292 million gross — approximately $242.0 million from the trust plus a $50.0 million private placement — with only 7,390 shares redeemed. It is the smallest and the newest of the four as a public company, and the earliest in development.

The Integral Molten Salt Reactor is a graphite-moderated thermal-spectrum design using a eutectic fluoride salt with uranium tetrafluoride, explicitly without beryllium or isotopically enriched lithium. A plant is a twin unit: 822 megawatts thermal gross, 390 megawatts electric net. Fuel is standard-assay low-enriched uranium, below 3% for the initial load and below 5% for make-up, which places it firmly on the right side of the enrichment line.

The revenue model is the most interesting thing about it

Terrestrial explicitly rejects build-own-operate. It sells the plant design, the core unit and the fuel salt, plus engineering and operations services, to developers who own the plants. The core unit is replaced every seven years, and the company’s own filings put that recurring replacement business at more than half of the lifetime value of a plant across a 56-year operating life, with construction and initial core supply at roughly 23% and pre-construction services at about 4%.

That is a genuinely attractive structure — an installed base that must buy a large component on a fixed schedule — and it is entirely theoretical until plants exist. Full-year 2025 revenue was zero, against $248 thousand in 2024. First-quarter 2026 net loss was $10.5 million on operating expenses of $11.9 million, with $9.1 million of operating cash consumed. The burn is modest because the activity is modest.

Where it stands, honestly

The NRC has approved topical reports on principal design criteria and on postulated initiating events, the latter with a safety evaluation report in May 2026. Canada’s regulator completed a phase two vendor design review in April 2023 finding no fundamental barriers to licensing across nineteen areas. There is no docketed construction permit application. Ground leases and research agreements covering roughly 77 acres at the Texas A&M RELLIS campus were signed on June 18, 2026 to enable NRC site characterisation work — the precursor to an application, not the application.

Terrestrial was also selected for the Department of Energy Reactor Pilot Program and executed its agreement for Project TETRA on January 6, 2026, with the programme target of criticality by July 4, 2026. As of the research cut-off, no first criticality announcement for TETRA has been located on the company’s investor relations newsroom or in its SEC filings. Oklo’s Groves reached criticality on August 5, announced on August 6, and was described in Oklo’s own release as the first reactor under the programme to do so on private land. Readers should treat the TETRA schedule as unconfirmed rather than missed.

Timelines in company documents are not fully consistent: investor communications point to a first commercial plant in the early 2030s, while the annual report separately indicates 2034 with fleet deployment in the late 2030s. A Department of Energy loan guarantee application of up to $890 million is described as under review, with no disclosed outcome.

A reusable triage tool: six questions for any advanced nuclear developer

The specific names will change. The questions that separate a real programme from a well-funded presentation do not. This list works on any developer in the sector, listed or private, and answering it takes about twenty minutes with a filings archive.

Run against these four, question six is currently answered by one company and question two is answered best by a different one. That divergence is the whole reason the comparison is interesting, and the reason no single name dominates the scorecard.

A ticker to keep on the watchlist, and why: $NNE

NANO Nuclear Energy Inc. ($NNE) was deliberately left out of the main comparison, because at roughly $917 million it sits in a different maturity class from Oklo and X-Energy at eight billion, and placing it in the same table would suggest an equivalence that does not exist. It belongs on a watchlist for a specific reason rather than as a smaller version of the same trade.

Why it earns the slot. On May 20, 2026 the NRC formally accepted for review the construction permit application for the KRONOS micro modular reactor at the University of Illinois Urbana-Champaign, submitted by the university on March 31, 2026. That moves NANO Nuclear onto rung two of the regulatory ladder above — an application docketed with a formal clock running — which is further than Terrestrial Energy has reached and on the same rung as X-Energy, at a fraction of the market capitalization. The company estimates the review completes in 2027, opening the possibility of nuclear construction at the university site in the second half of 2027.

KRONOS is a stationary high-temperature gas-cooled microreactor aimed at data centres, industrial sites, remote communities, mining and military bases. The portfolio also includes ZEUS, a portable solid-core battery reactor, and LOKI, a space-focused portable unit. Two subsidiaries address the fuel chain: Advanced Fuel Transportation, exclusive licensee of a patented high-capacity HALEU transport basket developed by three national laboratories, and HALEU Energy Fuel, aimed at domestic fabrication.

What holds it back. The reactor is a university demonstration unit, not a commercial power plant, and the applicant is the university rather than the company. Liquidity of $568.7 million at March 31, 2026 is solid for its size but the share count rose from 41.7 million at September 30, 2025 to 52.1 million by March 31, 2026, roughly a quarter more shares in six months. The September fiscal year end makes every quarterly comparison against the other four an approximation. And the six-month net loss of $15.7 million is flattered by $9.8 million of interest income, so the operating burn is larger than the bottom line suggests.

The risks that apply to all four equally

There is also a shared demand risk that is rarely stated plainly. The entire investment case rests on data-centre and industrial buyers being willing to pay a premium for firm, colocated, carbon-free power on a ten-year view. Those buyers have alternatives: gas with firm fuel supply, restarting or uprating existing conventional reactors, renewables with storage, geothermal, and simply building where transmission already exists. They will choose on reliability, time to power and delivered cost. Reactor novelty does not appear on that list.

Primary sources

• Oklo Inc. Form 10-Q, quarter ended June 30, 2026 — filed August 7, 2026.
• Oklo: Groves Reactor Achieves First Criticality — announcement of August 6, 2026; the Form 10-Q dates the event to August 5, 2026.
• X-Energy, Inc. Form 10-Q, quarter ended March 31, 2026 — filed June 4, 2026, including IPO subsequent-events disclosure.
• X-energy announces pricing of upsized initial public offering.
• NuScale Power Corporation SEC filings — Form 10-Q for the quarter ended June 30, 2026, filed August 5, 2026.
• Federal Register: NuScale small modular reactor design certification — final rule published January 19, 2023, effective February 21, 2023.
• Terrestrial Energy Inc. SEC filings — Form 10-K for 2025 and Form 10-Q for the quarter ended March 31, 2026.
• Terrestrial Energy executes DOE agreement for Project TETRA — January 6, 2026.
• NANO Nuclear Energy: NRC formally accepts KRONOS construction permit application — May 20, 2026.
• U.S. Department of Energy Reactor Pilot Program.
• MerlinTrader: Oklo ($OKLO) stock hub — full company research hub, updated for Q2 2026.
• MerlinTrader: the HALEU enrichment bottleneck.