NuScale Power Corporation [SMR] — Company Research

Task 1 of the investment-memo workflow · Prepared 2026-07-27 · Analyst: Omar Ahmady Price at writing: $8.29 (Alpaca SIP close, 2026-07-27) · Market cap ≈ $3.03bn · EV ≈ $2.02bn

Sourcing convention used throughout: every figure is either (a) cited to SEC EDGAR XBRL, Alpaca, Alpha Vantage or a named public aggregator, or (b) explicitly labelled ESTIMATE / FLAGGED. Nothing is presented as fact that is not sourced. Where a disclosure does not exist (product-level revenue, geographic revenue, gross profit per module), that absence is stated rather than papered over.

0. Why this name is in front of us

SMR ranked #1 of 27 short candidates on the 2026-07-27 idea-screener run (reports/screens/Screen_2026-07-27.md), composite score +5.43 — the widest margin on the board. Every factor family pointed the same way: 12-1 momentum −80.5%, price at 15.2% of its 52-week high, −52.4% vs. the 200-day moving average, Piotroski F-score 0/9, accruals +0.106, gross profitability 0.008, asset growth +159%.

A screen is a candidate generator, not a verdict. This memo underwrites the name independently against the six hard gates in references/trade-construction.md. The screen's job was to find it; the gates' job is to decide whether a position exists.

1. What the company actually is

NuScale Power Corporation designs the NuScale Power Module (NPM) — a light-water small modular reactor delivered as a factory-built, transportable pressure vessel. Six or twelve modules combine into a plant. The uprated 77 MWe module received NRC Standard Design Approval in May 2025, making NuScale the only company in the United States with a small modular reactor design approved by the Nuclear Regulatory Commission. That is a genuine, hard-won regulatory asset and the memo does not dispute it.

What NuScale is not, on the evidence of its own filings:

The company's only prior firm order — the Carbon Free Power Project with Utah Associated Municipal Power Systems — was cancelled in November 2023 after the target cost rose to roughly $9,300/kW. That is the base rate the bull case has to argue against, and it is NuScale's own history, not a generic prior.

Corporate structure

NuScale is an "Up-C": NuScale Power Corporation (NYSE: SMR, Class A) sits above NuScale Power, LLC, in which other holders (historically Fluor) held exchangeable units paired with non-economic Class B shares. The economic split shows up in the loss allocation: FY2025 total loss $664.5m split $355.8m to Class A and $308.7m to the non-controlling interest (EDGAR). By Q1-2026 the NCI share of the loss had collapsed to $2.7m of $46.7m (5.7%) — the arithmetic signature of Class B holders having exchanged into Class A and sold. Weighted-average basic shares went from 163.7m (FY2025) to 319.7m (Q1-2026).

2. The ENTRA1 arrangement — the single most important fact in this file

In August 2025 NuScale entered a Partnership Milestones Agreement with ENTRA1 Energy LLC. ENTRA1 — not NuScale — is the commercialisation, distribution and deployment vehicle for the NuScale Power Module. ENTRA1 signs the customer; NuScale supplies the hardware and pays ENTRA1 for the privilege.

v1.4.0 revision (2026-07-27). The prior version of this file sourced the per-reactor fee from press and litigation reporting and flagged that "the single most important thing to verify" was whether the remaining milestones are cash- or stock-payable, and on what schedule. That question is now answered from primary documents. The Partnership Milestones Agreement and its Guaranty were located by EDGAR full-text search and read in full:

Local copies: _source_data/entra1/.

2.1 What the filed contract establishes (no longer press-sourced)

2.2 The per-module economics, derived rather than reported

The dollar amount per NPM is redacted in the exhibit under Reg S-K 601(b)(10)(iv). It is nevertheless recoverable by arithmetic from two filed numbers, and the two routes agree:

So the total ENTRA1 contribution is ≈$46m per 77 MWe module on a 2025 pricing basis — roughly $610 per kW of nameplate capacity, paid before NuScale delivers anything. The prior memo's "$16m per reactor" was correct but was only the 35% Milestone 2 tranche; it understated the full obligation by about two thirds. And unlike the press figure, $16m/NPM is in a filed document — the FY2025 10-K risk factors — which the prior version did not establish.

Escalated at the contractual 5% floor:

2.3 The cash that has actually moved — fully reconciled

The 30/20/50 split in PMA §1(d) reconciles to the cash: 50% ($247.5m) in 2025, 50% ($259.9m) in January 2026. The $507.4m is fully paid. It is gone.

2.4 Four things this changes, two of which cut against the bear case

Against the bear case — stated first, because it is the correction the prior memo needed:

1. A binding TVA PPA does not crystallise $1.25bn in one go. The prior memo's "$1,246.8m = 123% of the balance sheet" is right as a gross figure (72 × $17.27m at 2026 pricing = $1,243m) but wrong as a near-term cash call. Milestone 2 applies to a maximum of 48 NPMs at any one time ($829m at 2026 pricing), and PMA §1(a)(ii) further restricts invoicing: only a redacted number of NPMs may be invoiced before a redacted month in 2027, with the balance payable within 12 months of invoice thereafter. The obligation is real; the timing is contractually staged, and the prior memo's implied funding shock in the quarter of signature overstates it.
2. Because the funding shock is smaller and later, the squeeze risk on a PPA announcement is larger, not smaller. Less offsetting bad news arrives with the good news. This makes Gate 4 harder to pass, not easier.

For the bear case:

3. The milestones are cash, at the listed-parent level, on 60-day terms, with no set-off, guaranteed as a primary obligation payable within five days of an email. The prior memo's explicit invalidation trigger — "any 8-K disclosing that the ENTRA1 milestone obligations are payable in equity rather than cash" — can no longer fire. That trigger is dead, and the thesis is correspondingly tighter.
4. The $507m is not one-time in substance. The 10-K risk factors say it plainly: when a project advances from Milestone 1 to Milestone 2, "the limits for those Milestone Contributions will reset", so NuScale "could be obligated to make additional Milestone Contribution 1 payments without any assurances that any revenue generating contract will be entered into." A binding TVA PPA therefore clears all 72 Milestone 1 slots and re-arms roughly $533m (2026 pricing) of fresh Milestone 1 capacity, chargeable on another set of non-binding letters. The charge NuScale booked as one-time is structurally repeatable, indefinitely, to 2045.

2.5 The cash-out-to-revenue-in ratio at the PPA event — quantified from both sides

Management's own guidance for what a TVA PPA is worth to NuScale's revenue (Q1-2026 call, CFO Ramsey Hamady):

"should TVA come across the line with, for example, PPA, we anticipate that we would have site-specific services. So pre-OEM services. If we look at RoPower as an example… all in with RoPower, we realized about $8 million worth of revenue… over 2024 and 2025… We would anticipate something potentially in that scale once we get to a PPA."

Set that against the filed contract. A single VOYGR-12 plant is 12 NPMs; Milestone 2 at 2026 pricing is $17.27m per NPM:

That single ratio is the mechanism of this memo, and both sides of it are now primary-sourced: the numerator from a filed exhibit plus a filed 10-K figure, the denominator from management's own words.

The counterparty

• ENTRA1 is a subsidiary of the Habboush Group. NuScale's CFO, Robert Ramsey Hamady, was previously the Habboush Group's Chief Investment Officer (public biographies; noted in the Iceberg Research report of 2025-11-14).
• Iceberg Research (2025-11-14) reported that ENTRA1's listed office is a WeWork in Houston at the same address as NuScale's Houston office, and that the suite number was subsequently removed from NuScale's website. Iceberg questioned whether the agreement was negotiated at arm's length.
• A securities-fraud class action is pending: Truedson v. NuScale Power Corporation, et al., D. Or., No. 3:26-cv-00328, lead-plaintiff deadline 2026-04-20. The complaint alleges NuScale misrepresented ENTRA1's project experience — specifically that ENTRA1 "had never built, financed, or operated any significant projects in the field of nuclear power generation" and that the qualifications attributed to it belonged to Habboush Group principals — and failed to disclose the scale of the milestone-payment exposure.

Analyst note on weight: allegations in a complaint are allegations. What is not an allegation, because it is in the audited filings, is the $507.4m expense, the ~$495m of cash out the door, the $1.4m of deferred revenue, and the fact that ENTRA1 rather than NuScale is the counterparty to TVA. The thesis in this memo rests on the filed facts, not on the litigation succeeding.

3. Financial history (SEC EDGAR XBRL, CIK 0001822966)

All figures grouped by each fact's own period-end date, full-year durations only, latest filing winning on restatement — per references/edgar-pipeline.md.

Quarterly revenue tells the story more sharply than the annual series:

The FY2024→FY2025 revenue decline is not the headline; the intra-year collapse is. RoPower Doicesti FEED Phase 2 engineering carried revenue through mid-2025 and then ran off. Nothing has replaced it.

Balance sheet — the honest good news

This matters and the memo will not hide it. NuScale raised $1.30bn of equity in FY2025 at prices between roughly $17 and $45, and again $37m in Q1-2026. Management timed that issuance well. Underlying cash burn — cash opex less SBC, plus capex, less $10.8m/quarter of interest income — is ≈$43m/quarter, ≈$172m/year.

⇒ ≈5.9 years of runway with no further ENTRA1 milestone payments.

Any short thesis built on "this company runs out of money" is wrong. The bear case here is dilution and value-transfer, not insolvency. (The Q1-2026 headline CFO of −$314.7m looks catastrophic but is 84% working-capital: −$264.2m of it was paying down the ENTRA1 accrual raised in Q3-2025.)

4. Ownership, insiders and the sponsor's exit

Institutional ownership

• 45.4% institutional, 11.3% insiders, 43.4% retail (wallstreetzen, 2026-07-27). The retail share is unusually high and is directly relevant to the squeeze analysis in Task 5.
• 507 institutional filers holding 181.3m shares (fintel).
• Largest holders: Van Eck Associates 20.03m (12.2%), Vanguard 15.55m (10.5%), BlackRock 12.85m (7.9%) (marketbeat/wallstreetzen, 2026-07-27).
• Read honestly: Van Eck's position is almost certainly the nuclear/uranium thematic ETF complex, and Vanguard/BlackRock are index-mandated. The top three holders are passive or thematic-mechanical capital, not active conviction. "Institutions own 45%" carries very little directional information here.
• Limitation flagged: EDGAR has no single endpoint returning all 13F holders of a ticker (references/edgar-pipeline.md §4). These figures come from aggregators, not a primary pull.

Insider activity — SEC Form 4, parsed 2026-07-27 (CIK 0001822966)

Applying the empirical asymmetry (Cohen, Malloy & Pomorski 2012 — buys are the signal, sales are mostly noise):

1. The officer sales are code-S disposals immediately following code-M option exercises. That is routine exercise-and-sell behaviour and carries little signal. The memo does not present it as bearish evidence.
2. Fluor's exit is a different category. Fluor was NuScale's founding sponsor, majority owner and EPC partner. Between 2026-04-09 and 2026-04-21 it sold 39.94m shares at $11.63-$12.07 and went to zero. Reported total proceeds across the full 2025-26 sell-down: ~$2.4bn. A strategic partner liquidating its entire position at prices 78% below the peak is not tax-driven diversification.
3. The absence of buying is itself the datum. After an 84% drawdown, with the CEO, CFO, COO, CTO, CCO and nine directors all holding stock, not one open-market purchase was made in twelve months. Insiders' only cash transactions were sales.

4A. Scoping what is NOT published (v1.4.0 — required)

Filings are what everyone has. This section records what was looked for outside them, which corpus was used, and what it returned — including where the answer is genuinely not knowable.

4A.1 Transcript mention-frequency — run FIRST, generatively, before revising any view

Window: 16 consecutive quarters, Q2-2022 through Q1-2026 — the company's entire post-SPAC history. Source: stockanalysis.com transcript archive (Alpha Vantage's EARNINGS_CALL_TRANSCRIPT quota was exhausted before this task began; the substitution is disclosed rather than worked around). Cached to data/transcripts/; counts in data/mention_freq_SMR.json; chart chart_36_mention_frequency.png.

Source-integrity note, because it changed a conclusion. An initial pass used InsiderMonkey transcripts. For Q3-2025 that source returned zero mentions of "ENTRA1" while returning 43 for "TVA"; the stockanalysis.com transcript of the same call returns 78 and 43. One of the two transcriptions is defective. A striking "management never once named the counterparty" finding was drafted off the first source and discarded once the second was obtained. The counts below are from a single, consistent source, and the episode is the reason this file states the source per name rather than treating a transcript as a transcript.

Every emerging and decaying term, listed as an open question with no interpretation attached (the method in references/mention-frequency.md), then investigated:

Hypotheses generated by this process (not carried in from the prior memo):

• G-1. The vocabulary substitution. "Order", "customer" and "contract" — the words of a seller — have been replaced in prepared remarks by "TVA", "PPA" and "term sheet" — the words of a supplier to an intermediary. The change is dated precisely to Q3-2025, the quarter the PMA was signed. Independent corroboration: the PMA itself, which makes ENTRA1 (not NuScale) the counterparty to every customer document, and the Q1-2026 call where the CFO says "once ENTRA1 gets a PPA". This hypothesis came from the transcript counts and was then confirmed against the contract — not the other way round.
• G-2. The OEM asymmetry. OEM contract mentions are entirely analyst-driven (22 total mentions in Q1-2026, zero in prepared remarks in any of 16 quarters). The OEM agreement is Milestone 3 — a 50% tranche, ≈$23-25m per NPM in cash to ENTRA1. Management's silence on the instrument that triggers the largest single payment, while answering questions about it, is the disclosure asymmetry worth watching.
• G-3. The datacenter narrative is being abandoned, not pursued. "data center" went 7 (Q1-2024) → 0 in each of the last four calls, and "hyperscaler(s)" from 3 to ~1. Against a sector in which every competitor is loudly courting AI load, this is a decaying term where the market assumes a rising one. Corroboration attempted: NRC dockets show no NuScale-technology behind-the-meter project; no hyperscaler counterparty has been named in any filing.
• G-4. "Liquidity" emerging while "dilution" stays at zero across 16 quarters is a disclosure-posture observation, not an economic one, but it is the correct place to look when Milestone 2 becomes payable.

4A.2 Sector corpora (energy/utilities: NRC dockets, TVA records, DOE, state commissions)

Per references/unpublished-scoping.md, healthcare corpora do not apply. What was attempted and what it returned:

4A.3 What remains genuinely unknowable — with the corpora that failed to answer it

4A.4 Bottom-up TAM and the required implied-penetration statement

Built from units per references/tam-sizing.md, not from an industry-report headline.

Penetration-path sensitivity — GW that must eventually be delivered to justify today's $2,021.6m EV (plus $590.4m PV of burn to a 2031 first delivery, discounted at 14%):

Time to revenue: first NPM delivery is not before 2031 on the company's own 1-2 years pre-construction plus 3-4 years construction (SAA Recital C), and that clock has not started because no site is docketed with the NRC. What would falsify this section: a binding TVA PPA with a named site and an NRC construction-permit application docketed inside twelve months.

Required implied-penetration statement. At today's $8.29 and a $2,021.6m enterprise value, and using the full ENTRA1 contribution now readable from the filed agreement rather than the $16m Milestone 2 tranche alone, NuScale must eventually deliver and be paid for roughly 29.8 GW — 5.0× the entire announced TVA programme and 387 modules — against 0.0 GW of binding orders, 0 NRC-docketed sites and $1.4m of deferred revenue. On the Street's average $14.57 target the requirement is roughly 56 GW, or 9.3× the TVA programme. The prior memo's 8.8 GW / 1.47× figure was computed net of a fee one-third the true size; the correction moves the implied requirement from "demanding" to "arithmetically implausible at the estimated gross margin."

Honest counter, retained: the answer is highly sensitive to the undisclosed gross profit per module. At $150m per NPM — which cannot be ruled out because the NPM price is redacted — the requirement falls to 3.8 GW, which one TVA programme covers. The bear case rests on the fee being large relative to the margin, and the margin is the input that is not knowable.

5. Industry and competitive position

The theme is real; the question is who captures it

Global electricity demand from AI data centres, plus decarbonisation mandates, has produced the strongest policy tailwind nuclear has seen in forty years — DOE loan programmes, EXIM and DFC letters of interest, and utility procurement interest. Bank of America has publicly framed nuclear as a ~$10trn opportunity. The memo accepts the theme. What it disputes is the claim that NuScale equity is the way to own it.

Peer set (stockanalysis.com, 2026-07-27; SMR EV from EDGAR net cash)

The strongest argument against this memo's own direction, stated up front: on enterprise value, SMR at $2.02bn is now materially cheaper than Oklo at $4.80bn, despite being the only company in the group with an NRC-certified design and despite holding $1.01bn of net cash against Oklo's $2.21bn. Within the theme, SMR is no longer the expensive name — it is arguably the cheapest credible one. Task 5 treats this as a live counterargument, not a straw man.

Competitive landscape

• Oklo — fast reactor, no NRC approval, but a stronger narrative and a data-centre-adjacent customer story.
• X-energy, TerraPower, Holtec, Rolls-Royce SMR, GE Hitachi BWRX-300 — private or subsidiary; the BWRX-300 in particular has firm utility commitments (OPG Darlington) that NuScale does not.
• BWXT — the profitable comparison: a real revenue base and real margins in nuclear components.
• NuScale's differentiator is regulatory (design certification and standard design approval). Its disadvantage is commercial: it has ceded the customer relationship to ENTRA1.

TAM — treated sceptically

The addressable market for SMRs is genuinely large in a 2035-2050 frame. It is not the constraint on NuScale's value. The constraint is the share of any project's economics that reaches SMR's Class A shareholders after ENTRA1's per-reactor fee, and the timing (first delivery 2031 at the earliest). A large TAM does not answer either question, and the memo declines to use TAM as an argument in either direction.

6. Management

Capital allocation is the management issue, and it is severe. Paying $495m in cash — half the balance sheet at the time — on the trigger of a non-binding letter of intent is the single hardest fact in this file to reconcile with a shareholder-value frame. The CFO's prior employment by the counterparty's parent is disclosed and is a governance flag; it is not, on the public record, evidence of wrongdoing, and the memo does not treat it as such.

Offsetting credit where due: management raised $1.30bn of equity in FY2025 at $17-$45 per share. That was excellent market timing and is the reason the company has 5.9 years of runway today.

7. Risk register

Risks to a bearish view (i.e. what could make the stock go up violently)

1. A binding TVA power purchase agreement. Management has guided that a definitive PPA "could be signed by the end of 2026"; the CFO said publicly, "we're hopeful that TVA can come across the line at some point later this year." ENTRA1 has said discussions are "advancing well." Sell-side commentary has explicitly argued the stock could double on signature.
2. Positioning. 70.5m shares short = 20.8% of float, days-to-cover 2.28, 43% retail ownership. The fuel and the ignition source are both present.
3. Realised volatility 99.5%; 1-year daily beta to SPY 3.91. Ten separate days of +20% or more since 2022; a +109.4% rally in twenty sessions between 2025-04-28 and 2025-05-27.
4. The balance sheet. $1.01bn net cash and no debt removes the financing-distress path entirely.
5. Relative value. SMR is cheaper on EV than Oklo. A sector re-rating lifts the cheapest credible name most.
6. Government support. DOE loan-programme signalling, EXIM ($3bn LOI) and DFC ($1bn LOI) support for the Romanian project could be announced at any time.
7. Acquisition. A debt-free company with the only NRC-certified SMR design and $1bn of cash at a $2.0bn EV is a plausible strategic target.

Risks to the company (i.e. what supports a bearish view)

1. Revenue has gone to ~zero and there is no contracted backlog ($1.4m of deferred revenue).
2. Contingent ENTRA1 obligations of ~$3.4bn implied in total, ~$1.25bn on a binding TVA deal — greater than the entire balance sheet.
3. Structural: NuScale does not own the customer relationship. Commercial success accrues first to ENTRA1.
4. Securities litigation (D. Or. 3:26-cv-00328) with an unquantified damages exposure.
5. First-of-a-kind execution. Vogtle ran seven years late and ~$17bn over. NuScale's own CFPP was cancelled.
6. Dilution. Share count is up 6.3x since FY2022 and the shelf remains active (S-3ASR filed 2025-08-11; 424B5 prospectuses filed 2025-11-07 and 2026-02-26).
7. Quality factors: Piotroski F-score 0/9, accruals +0.106, gross profitability 0.008 — the empirical profile in which shorts work.

8. What would have to be true for the bulls

Stated as a live case, not a straw man. The bull is betting that: - The 6 GW TVA programme converts to a binding PPA within 12-18 months, and the ~$1.25bn ENTRA1 fee is either smaller than reported, staged over many years, payable in stock, or simply swamped by the value of a confirmed 78-module order book; - Being the only NRC-certified SMR design becomes decisive as utilities move from studies to procurement; - Government financing (DOE/EXIM/DFC) de-risks first-of-a-kind capital cost; - $1.01bn of net cash and no debt buys enough time for the 2031-2033 delivery window to arrive.

Does this memo have evidence against that case, or only scepticism? Honestly: partial evidence. There is hard evidence on the structure (the $507m expense, the $16m/reactor fee, the $1.4m of deferred revenue, the 95.8% revenue collapse — all from filings). There is no evidence that the TVA PPA will fail, and none that the technology will fail. The variant is about who captures the value and at what dilution, not about whether SMRs get built. That distinction is carried explicitly into Gates 2 and 4.

Prepared under the investment-memo skill (template v1.3.0). Data: SEC EDGAR XBRL (CIK 0001822966, pulled 2026-07-27), Alpaca Markets SIP daily bars (2026-07-27), Alpha Vantage EARNINGS_ESTIMATES (pulled 2026-07-27), Alpaca options snapshots (2026-07-27), SEC Form 4 filings, and named public aggregators as cited.