Cloud Computing
PrivateThe only NRC-approved SMR — but a pre-commercial story trading at 86x EV/sales while its entire $25B order book runs through ENTRA1, a 3-employee single-investor entity it has agreed to pay >$3B in milestone "consideration" for NON-binding deals; revenue just fell 95% and a securities-fraud class action is live. BEARISH/MEDIUM until a binding PPA converts paper to product.
Research
The verdict
The only NRC-approved SMR — but a pre-commercial story trading at 86x EV/sales while its entire $25B order book runs through ENTRA1, a 3-employee single-investor entity it has agreed to pay >$3B in milestone "consideration" for NON-binding deals; revenue just fell 95% and a securities-fraud class action is live. BEARISH/MEDIUM until a binding PPA converts paper to product.
Primary sources
Source documents — open to read in full
NuScale designs and is trying to commercialize a small modular nuclear reactor (SMR) — the NuScale Power Module (NPM), a 77 MWe (gross) integral light-water reactor that is factory-built, ships as a single vessel, and scales in 4 / 6 / 12-module plants (308 / 462 / 924 MWe) . The pitch: passive safety (no operator action, AC/DC power, or added water needed to keep it safe indefinitely), a small site-boundary emergency-planning zone, and — crucially — **regulatory certainty**. NuScale's design is *the only SMR ever to receive a Standard Design Approval (SDA) from the U.S. NRC*: a 50 MWe / 12-module design certified in 2020, and a **6-unit, 77 MWe design approved May 2025** .
The catch is that there is no product revenue yet. The company "operates in one reportable segment," and "all significant revenue generated to date arises from engineering and licensing fees and services provided to potential customers, with the end goal of selling NPMs" . FY2025 revenue was **$31.5M** , almost entirely engineering/licensing — i.e. NuScale is a pre-commercial company selling design work, not power plants. The first NPM is targeted for commercial operation ~2030 ``.
Customers / channel — the load-bearing fact. As of 2025, NuScale routes essentially its entire commercial future through ENTRA1 Energy LLC, its "exclusive global strategic partner for the commercialization and development of power plants utilizing NPMs," which is also its prospective customer (NuScale is "key supplier" to ENTRA1 projects) . ENTRA1 holds **global exclusive rights** to commercialize, distribute, deploy, finance, own and operate NuScale plants . The two also run a 50/50 JV, ENTRA1 NuScale LLC . The only *named direct* international customer is **RoPower** (Romania) . Domestically, the marquee is the TVA / ENTRA1 6-GW agreement (Sept 2025) and Standard Power (2 GW, Ohio + PA, data centers) ``.
Suppliers: Doosan Enerbility (Korea) fabricates NPM sub-assemblies and is an investor; long-lead material (heavy forgings, steam-generator tubes, weld material) is manufactured in South Korea ``. Fluor (NYSE: FLR) was the historical majority owner, EPC partner, and — through 2025 — the dominant revenue counterparty.
Contract structure: fixed-price engineering/licensing contracts recognized as performance obligations complete; heavily concentrated and lumpy. The NPM "sales" are not yet binding — they are non-binding agreements and milestone term sheets routed through ENTRA1 (see Lens 3 / Lens 10).
Upstream inputs → NuScale → end customer, with named stakeholders:
. LLM work-in-process on the balance sheet was **$63.8M** at 2025-12-31 .Chokepoints / single-source dependencies: (1) NRC licensing — every plant still needs a construction permit + operating/combined license per project even with the SDA done; (2) Korean heavy-forging supply for the pressure vessel — a single geographic concentration; (3) and most acutely, ENTRA1 as a single channel — NuScale has contractually handed exclusive global commercialization to one counterparty. This lens is not generic: the chain is real and named, but the demand end of it is a single, thinly-staffed intermediary (Lens 10).
Real moat #1 — regulatory. NuScale is the only SMR with NRC SDA (twice: 50 MWe in 2020, 77 MWe in May 2025). The first SDA took 41 months — the fastest NRC reactor approval ever — and cost ~$500M in testing/engineering plus ~$130M responding to NRC RAIs over 250,000 review hours . No other SMR/advanced-reactor developer has even *applied* for an SDA . This is a genuine, expensive, time-based barrier — a customer can reference the approved design and skip years of design-certification risk. It is the single best thing about the equity.
Real moat #2 — IP + light-water lineage. 513 patents issued + 268 pending (781 total); 428 employees, 22 PhDs ``. The reactor is built on 60+-year-old proven LWR physics and conventional LEU fuel — lower technology risk than fourth-gen competitors.
Where the moat is thinner than bulls claim: (a) The SDA is for the design; it is not an order book and not a deployed plant. No NuScale reactor is under construction anywhere as of mid-2026. (b) Bargaining power is inverted. NuScale needs ENTRA1 (its only channel) far more than ENTRA1 needs NuScale — ENTRA1 "retains sole and full discretion to select, contract with, or purchase from NuScale" , yet NuScale must pay ENTRA1 milestone "consideration" regardless. (c) Competitors are closing the regulatory gap fast: **GE Vernova Hitachi BWRX-300** is *physically under construction* at OPG Darlington (Canada) since May 2025, targeting 2029-30 — likely the first Western SMR to actually operate — and TVA's BWRX-300 at Clinch River is in NRC review; **Holtec SMR-300** had its construction-permit application accepted by the NRC (Feb 2026); **X-energy Xe-100** has Amazon backing and a Dow Seadrift project . NuScale's lead is in paper approval; rivals may win the race to concrete.
One reportable segment — SMR commercialization . Revenue disaggregation by service (FY, $000s) :
| Service line | FY2025 | FY2024 | FY2023 |
|---|---|---|---|
| Power Plant & NPM-related services | $30,077 | $36,161 | $21,120 |
| Energy Exploration Centers | $1,268 | $473 | $673 |
| Other | $134 | $411 | $1,017 |
| Total | $31,479 | $37,045 | $22,810 |
Geography: predominantly U.S. operations; LLM manufacturing in South Korea; the only international customer revenue is Romania (RoPower) via the Fluor subcontract ``.
Customer concentration is extreme. Revenue from Fluor (related party) was 76.0% of total in FY2025 (11.4% FY24, 74.1% FY23) . The trend is *decelerating and now collapsing*: FY25 revenue fell 15% YoY, and **Q1 2026 revenue was $0.565M vs $13.4M in Q1 2025 — down 95.8%** , with $0 from Fluor as the Doicesti FEED Phase 2 contract finished. The "segment" is, in effect, a project-services shop that just ran out of project. There is no recurring revenue base — only the promise of future NPM sales.
. EPS **−$0.14** vs −$0.13 expected .. The cash drop is largely the **$259.9M ENTRA1 Milestone 1 payment settled in Q1 2026** (the second tranche of a $507.4M obligation) , plus rotation of cash into investments.Flag vs own history: the FY2025 net loss of $664.5M is dominated by a one-time **$507.4M ENTRA1 Milestone 1 expense** booked in G&A (which is why FY25 G&A ballooned to $609.8M from $75.9M). Strip that out and the underlying operating loss is ~$155-180M — still large relative to ~$31M revenue, but the headline loss is a milestone-accounting event, not a cash incinerator in the normal sense. The cash side, however, is real: that $507.4M is being paid ($247.5M in 2025 + $259.9M in Q1 2026) to ENTRA1.
No transcripts on the shelf (transcripts/ empty); sentiment from web ``.
Pre-revenue SMR developers do not support P/E or EV/EBIT — they have no E. The honest peer frame is valuation vs (near-)zero sales and regulatory/deployment stage. Multiples `` or n/a.
| Company | Ticker | Mkt cap | Stage / NRC status | EV/Sales | P/E | Notes |
|---|---|---|---|---|---|---|
| NuScale | SMR | ~$3.73B `` | Only NRC SDA (77 MWe, May-25); first power ~2030 | ~86x `` | n/a (loss) | Channel = ENTRA1 |
| Oklo | OKLO | n/a | NRC cert expected ~2027-28; data-center + fuel recycling | n/a | n/a (loss) | Main rival; also pre-revenue `` |
| Nano Nuclear | NNE | n/a | Earlier-stage microreactor | n/a | n/a (loss) | `` |
| Centrus Energy | LEU | n/a | HALEU/LEU enrichment — has revenue | n/a | positive | Fuel-cycle, not a reactor dev `` |
| GE Vernova (BWRX-300) | GEV | n/a | Under construction, Darlington; ~2029-30 | n/a | positive | Embedded in a profitable parent `` |
Pattern: this name trades on (1) NRC milestones and (2) headline "order" announcements via ENTRA1, and it falls hardest on anything that questions ENTRA1's substance or the cash cost of those orders. The market reacts to narrative legitimacy, not to financials — there are no financials to react to. Net: from $53.43 (Oct 2025) to ~$9.88 (Jun 2026), an ~82% drawdown ``.
. Track record: he steered NuScale through *both* NRC approvals and the 2022 SPAC listing (via Spring Valley) — a genuine regulatory achievement. But the signature commercial event on his watch, the **2023 cancellation of the UAMPS/Carbon Free Power Project** (the flagship 6-module Idaho plant, killed on cost escalation), is the original sin: it forced the pivot from a real first customer to the ENTRA1 channel. The wind-down still echoes (the $5.1M restricted-cash letter of credit and the LLM settlement) .. CODMs are CEO/CCO/CFO managing to "Net loss" .. Weighted Class A shares went **73.4M (2023) → 93.2M (2024) → 163.7M (2025)**; shares outstanding reached **318.5M** by Q1 2026 against **662M authorized** (raised from 332M in Dec 2025) . They have ~$962M of ATM headroom left — i.e. capacity to issue ~97M more shares at current prices.Accounting: The books are clean in the GAAP sense (no SEC LR/AAER — see below), but two structures demand scrutiny:
. NuScale determined M1 **does not meet ASC 606 capitalization criteria → expensed, not an asset** . Translation: NuScale paid out half a billion dollars in cash for a non-binding term sheet, and its own auditors agree it created no asset. Milestone 2 (35%, on a binding PPA/off-take) and Milestone 3 (50%, on a binding OEM agreement) remain — implying a total milestone obligation of ~$3.38B, with ~$2.87B still to come ``. That remaining exposure is ~3x current total liquidity and would have to be funded by issuing stock.Cash vs earnings: the FY25 headline loss overstates cash burn (the $507.4M M1 is partly non-cash timing), but the structural operating loss (~$155-180M ) and the milestone cash outflows are real. Minimal capex — this is a labor + IP business (labor $77.2M FY25) .
Regulatory findings (required):
. The core allegation: NuScale misrepresented ENTRA1's qualifications and **failed to disclose the scale of its milestone-payment exposure** (the $495M Q3-25 payment + >$3B potential). The catalyst was a **Guggenheim Securities report (Nov 2025)** calling ENTRA1 *"a 3-year-old company that has never built, financed or operated anything,"* with *"3 employees and 1 investor,"* better described as *"an entity supporting the activities of a single individual, Mr. Habboush"* . Lead-plaintiff deadline was April 20, 2026. This is the single most important risk in the file — it goes to whether the entire commercial channel is real.A bottom-up EPS model here is close to meaningless — NuScale has no revenue base to grow and its loss is dominated by milestone timing it cannot precisely forecast. So the honest projection is a scenario on cash and dilution, not a point EPS ``:
; structural operating loss ~$150-200M/yr ex-milestones ; no Milestone 2/3 triggered in FY26 (no binding PPA signed yet) → net loss ~$160-200M, EPS ~−$0.50 to −$0.65 on ~320M+ shares . Runway: ~$1B liquidity ÷ ~$200M structural burn ≈ **~4-5 years** *if no further milestone payments* . First NPM power ~2030 → first NPM-sale revenue is ~4 years away.I am not logging a Brier forecast (forecast.ts create skipped per --watchlist rules — no genuine committed base case on an EPS line for a pre-revenue name). The scoreable binary, if/when promoted to a thesis, is: "NuScale signs a binding PPA or OEM agreement (Milestone 2 trigger) by 2027-12-31" — that is the event that converts the whole story from paper to product.
Bull case. NuScale owns the one thing nobody else has: a completed, twice-approved NRC design. In a world where AI data centers, reshoring and electrification are driving the largest power-demand surge in decades, a certified SMR is a call option on the cleanest, most-grid-friendly form of always-on baseload. The TVA 6-GW program (72 NPMs) and Standard Power (2 GW) and the $25B ENTRA1 agreement are exactly the kind of demand that, if any single one converts to a binding PPA, re-rates the equity violently. The balance sheet (~$1B, no debt) buys ~4-5 years of runway to get there. With the stock down 82% and Dale Klein (ex-NRC chair) on the board, the risk/reward improves the closer it gets to a binding order.
Bear case (the 2-3 things that can permanently impair it).
Pre-mortem (18 months on, thesis broke): No binding PPA was signed; the Truedson securities case survived a motion to dismiss and forced restated disclosure on ENTRA1; the stock drifted to $5; NuScale kept the lights on via dilutive ATM issuance, doubling the share count again; the "order book" was quietly reframed as "framework agreements." The certified design is now a licensing asset others build around, not a product NuScale sells.
Are multiples too high? Yes — unambiguously, on any near-term basis. ~86x EV/Sales on shrinking revenue is a 2030+ optionality price, not a fundamentals price. The only justification is the embedded call option on certification + the demand wave.
Contrarian view (what the market may be refusing to see, both ways): Bears assume ENTRA1 is fatal; but the structure (an external developer that finances/owns/operates the plants) is arguably the right capital-light model for a company that cannot itself fund $20B of plants — it offloads the balance sheet that killed CFPP. The market may be conflating "ENTRA1 looks flimsy today" with "the channel model is wrong." If ENTRA1 (or a designated affiliate / strategic) lands one binding utility PPA, the same structure the bears hate becomes the bull thesis. The honest read: this is binary and event-driven, and you do not have to own it before the binding-order event.
Dismantling the bull case:
The default arms dealer of the AI buildout — a real moat compounding a $15B backlog into 30% organic growth, but priced at 82x for perfection while insiders sell 65:0 and EMEA orders are already cracking.
A 109-year-old radiator maker that 80/20'd itself into a data-center liquid-cooling growth story and got re-rated 14.7x in five years — the asset is real and a single hyperscaler has pre-committed >$4B of cooling for CY27–29, but at ~38x forward / ~133x trailing the price already pays for flawless execution while Q4 just showed the margin and supply-chain cracks that flawless execution doesn't have. Own the post-RMT pure-play on a data-center capex scare or a second margin miss — not here.
A quietly excellent RF-and-photonics compounder that the market has finally noticed — fab-lite 56% gross margins, three secular legs (defense, AI data-center optics, telecom), a net-cash balance sheet after killing its 2026 converts, but now priced at ~54× forward with an AI-optical multiple it must keep compounding into; great business, demanding entry.