NuScale Power [SMR] — Valuation Analysis

⚠️ SUPERSEDED IN PART — 2026-07-29

The position verdict in this document is retired. Under the current framework (references/criteria.md, 2026-07-29) the memo outputs an analysis, not a position. Whether an analysis justifies a position is a question about a particular book, and two books answer it differently.

The Gate block and the Gate 4 expected-return arithmetic below are also superseded, by the named Criteria (each with a type: BINDING or MEASURED, returning PASS / FAIL / INDETERMINATE), the reverse-DCF implied-path test, and a separate 12-month target.

→ Current analysis: NuScale_Criteria_and_Valuation_2026-07-29.md

Everything else here — the research, the evidence, the mechanism work — stands. Residual references to "Watchlist" in the prose below are the historical record of the 2026-07-27 assessment and are left intact deliberately.

Task 3 · 2026-07-27 · Price $8.29 · Market cap $3,029.8m · Net cash $1,008.2m · EV $2,021.6m Model: NuScale_Financial_Model_2026-07-27.xlsx (all figures below are live formulas in that workbook, verified by opening it in Excel and reading back the computed cells).

1. Why there is no DCF in this file

A discounted cash flow model requires a cash flow to discount. NuScale's position:

• Q1-2026 revenue $0.565m; TTM revenue $22.7m and falling;
• Deferred revenue (contract liability) $1.409m;
• Zero binding reactor orders;
• First commercial module delivery 2031 at the earliest;
• Operating cash flow negative in every year of the company's existence.

Every line of a five- or ten-year DCF on this company would be an assumption. The output would move by a factor of five on inputs that are not disclosed anywhere (gross profit per module, module count, delivery schedule). That is not a valuation; it is a spreadsheet-shaped opinion with false precision, and building one would breach this skill's data-sourcing discipline.

Three tools are used instead, in descending order of reliability:

1. Liquidity-runway analysis — the only fully sourced valuation-relevant work available.
2. Implied-expectations (reverse) valuation — invert the question: what must be true for today's EV to be right? This is falsifiable in a way a forward DCF is not.
3. Scenario/probability weighting and relative value — how the price behaves across discrete outcomes.

2. Section A — Liquidity runway

Conclusion: NuScale is not going bankrupt. Any bear case built on insolvency, a going-concern qualification or a financing squeeze is factually wrong and this memo does not make one.

Then the same table with the bull catalyst applied

This is the central analytical claim of the memo. The event the Street is waiting for — a binding TVA power purchase agreement — is simultaneously the event that crystallises a cash obligation larger than NuScale's entire balance sheet, forcing a 12-25% equity issue. The good news and the funding hole arrive in the same 8-K.

v1.4.0 — the caveat that stood here is resolved. The prior version read: "the $16m/reactor figure is a reported number… not a figure this analyst has read in a filed agreement… If the fee is payable in stock rather than cash, or staged over the delivery period rather than at signature, the severity falls materially. That is the single most important thing to verify on the Q2 call." It has been verified, from the filed Partnership Milestones Agreement (8-K 2025-09-02, Ex. 10.1) and Guaranty (Ex. 10.2) — see Research Document §2:

• The milestones are cash. USD wire transfer, 60 days from invoice, no set-off, "final, irrevocable, non-cancelable and non-refundable" (PMA §1(e)) — and guaranteed by the listed parent as primary obligor, payable within five days of an email (Guaranty §2). The stock-settlement escape does not exist.
• They are staged, but not in the way that helps. Milestone 2 is capped at 48 NPMs at any one time ($829m at 2026 pricing, not $1,243m), and PMA §1(a)(ii) defers invoicing of the excess past a redacted month in 2027 with 12-month payment terms. So the timing is softer than the prior memo assumed — which makes the squeeze risk on a PPA announcement worse, because less offsetting bad news lands with the good news.
• The fee is ≈$46m per NPM, not $16m. $16m is the 35% Milestone 2 tranche. The total, derived from $507.393m ÷ 72 ÷ 15% and cross-checked against the 10-K's own "approximately $16 million per NPM" ÷ 35%, is $45.7–47.0m per 77 MWe module, escalating at the greater of 5% or CPI every 1 January to ≈$63m by a 2031 delivery.

Section B below is re-run on the corrected fee.

3. Section B — Implied-expectations (reverse) valuation

Question: how much NuScale capacity must eventually be delivered for today's $2,021.6m enterprise value to be fair?

(The prior version compared against a rounded 6.0 GW programme; 72 NPMs × 77 MWe is 5.54 GWe, so both the 1.47× and the 5.4× are stated against the same denominator here.)

Sensitivity — GW that must be delivered, on the corrected fee

How to read this, revised. With the true fee, any gross profit per module at or below ~$47m produces no positive value per module at all — the ENTRA1 contribution consumes the whole margin. At the memo's generous $60m estimate, 78% of gross profit per module is paid to ENTRA1 on a 2025 basis, and more than 100% by a 2031 delivery at the contractual 5% escalation floor. The requirement only becomes plausible at $150m of gross profit per module, i.e. roughly $1,950/kW of pure margin.

Restated falsifiable claim: the current price capitalises between five and nine successful, fully delivered TVA programmes — before a single binding order, a single named site, or a single NRC construction-permit application exists.

And the honest counter, unchanged in force but now correctly located: the binding uncertainty is no longer the fee — that is now read from a filed contract — but the gross profit per module, which is undisclosed and whose ceiling is contractually capped by a redacted NPM price (PMA §1(g)). The bear case is a claim about a ratio, and only one side of the ratio is knowable.

4. Section C — Relative value

Two readings, and both are true:

• Against SMR: at 89× TTM sales and 894× the Q1 annualised run-rate, the multiple is not a multiple.
• For SMR: it trades at 42% of Oklo's enterprise value with an approved design Oklo does not have, and 33% of its market capitalisation is cash. If the theme re-rates, this is the cheapest credible vehicle in it.

Per references/consensus-bridge.md: if the entire peer group is re-rating together, sell-side relative-value targets track the group's multiple rather than asking whether the group is rich. That is exactly what is happening here, and it is a sector-regime question, not a stock-specific short thesis. This memo does not claim to know whether the advanced-nuclear complex is correctly valued.

5. Section D — Consensus estimates and the house view (the Gate 2 evidence)

Source: Alpha Vantage EARNINGS_ESTIMATES, pulled 2026-07-27 (raw JSON cached at .data/av_earnings_estimates.json).

Revenue

Reality check: Q1-2026 actual revenue was $0.565m against a consensus of $5.57m — an 89.9% miss. To hit the FY2026 average of $53.5m, NuScale must produce $52.9m across Q2-Q4 against a Q1 run-rate of $0.565m — a 31× step-up, with $1.4m of deferred revenue on the balance sheet and no announced contract that would generate it.

The house forecast is simply the run-rate plus modest RoPower implementation work. It requires no bearish assumption at all; the Street number requires a large, specific and unannounced one.

Note also the FY2027 dispersion: $26.7m low against a $415.8m high — 15.6×. That is not a consensus. It is a small number of analysts modelling a TVA programme and a similar number modelling nothing. The "average" is an artefact.

EPS — where the house view is not more bearish, stated plainly

This must be disclosed and not buried. On the bottom line the Street is already at least as bearish as this memo — more so for FY2027. The reason is mechanical: NuScale's loss is driven by a ~$230m operating cost base, not by revenue, so an 88% revenue shortfall moves EPS by only a few cents.

Therefore the Gate 2 variant is specifically about revenue trajectory, the ENTRA1 cash-obligation structure, and the price target — not about the P&L loss. Presenting a "negative variant on earnings" here would be false.

Revision direction

FY2027 estimates have been cut 45% in ninety days — a clean revision tailwind for a bearish view (Chan/Jegadeesh/Lakonishok 1996). FY2026 has drifted marginally better on cost control. A split reading, disclosed as such.

6. Section E — The consensus bridge (required by references/consensus-bridge.md)

6.1 Where the Street actually is

Source: stockanalysis.com, 2026-07-27, 18 analysts (investing.com reports a 14-analyst set with a $15.36 average — the variance between aggregators is disclosed rather than hidden behind one false-precise number).

Recent target actions — direction matters more than level:

Is consensus already bearish? Partly — and this is the crux of Gate 2.

• On ratings: majority Hold with two Sells. Not bullish, but 6 of 18 are still Buy or Strong Buy.
• On targets: the average implies +75.8%. That is emphatically not a bearish Street.
• On direction: targets are being cut hard and fast (Goldman $9→$6, Barclays $15→$11 in the last two weeks). Consensus is weakening, not weakened — i.e. the de-rating is in progress, not complete.
• On the house target: at $6.75, this memo sits between Goldman's $6 and Citi's $7.50 — at the bearish edge of the Street, not beyond it. No claim to a unique insight is made. Stated plainly: if the house view is right, the payoff is that the average target converges toward where the two most bearish houses already are; there is no scenario in which this memo is right and the whole Street is wrong in a way nobody has spotted.

6.2 The required decomposition: numbers, or multiple?

The standard bridge asks whether a target gap is a disagreement about fundamentals or about the multiple. For SMR neither frame quite applies, because there is no meaningful revenue to multiply. The correct decomposition here is numbers vs. structure, and it resolves cleanly. Using a shared base:

Neither the house nor the Street target can be defended as a revenue multiple. What each actually embeds is an implied module count (Section B methodology, holding gross profit at $60m/module and the discount rate at 14%):

(Second figure in each cell re-runs the same methodology with the full $46.98m ENTRA1 contribution per NPM established from the filed agreement, rather than the $16m Milestone 2 tranche used in v1.3.0. Denominator is the 5.54 GWe announced TVA programme.)

The disagreement, stated as one testable sentence: the Street's average target requires NuScale to eventually deliver roughly three TVA programmes; the house target requires one. Twelve months from now either (a) binding orders exist that make three programmes credible, or (b) they do not and the multiple converges toward one. This is directly checkable later and is what will be logged for outcome-scoring.

6.3 Other honest reasons the Street might diverge — not cherry-picked

1. Sell-side targets skew structurally bullish industry-wide. A useful prior, insufficient alone.
2. Relative-value anchoring. With Oklo at $4.8bn EV, a $14.57 target on SMR (implying $4.3bn EV) is simply "SMR should trade like Oklo." That is internally coherent and may be right.
3. Optionality is genuinely hard to value. A binding 6 GW PPA would be transformative. Analysts may be probability-weighting it at 30-40% where this memo weights it at 25%. That is a difference of opinion about a probability, not about a fact — and this memo has no privileged information about the TVA negotiation.
4. Targets are being cut in real time. The average is a lagging aggregate of stale targets. Goldman at $6 and Citi at $7.50 are the current marks from firms that have updated post-Q1.

6.4 The real risk to the house view, named

If the advanced-nuclear re-rating is structural rather than a reversion candidate, an anchor on contracted-revenue reality will systematically miss for as long as that regime holds. SMR would then trade on narrative and relative value against Oklo, and the reverse-valuation arithmetic in Section B would be irrelevant to price formation for years. This is not a hypothetical: it is what has happened to the entire group since 2024, and it is why this memo's conclusion is a Watchlist rather than a Short.

7. Section F — Positioning and options market read

Short interest (aggregator/FINRA — lags; disclosed)

Sources: dailypolitical/fintel (FINRA bi-monthly settlement) and stockanalysis.com. 20.8% of float short with a 2.28-day cover is the definition of a crowded, fast-unwinding short. Read as signal (Asquith/Pathak/Ritter 2005), high and rising short interest predicts underperformance. Read as positioning, it is squeeze fuel.

Implied borrow cost — computed, not assumed

Put-call parity on the 15-Jan-2027 $8 strike (Alpaca options snapshots, 2026-07-27): C − P = S·e^(−qT) − K·e^(−rT), with C = $2.395, P = $1.975, S = $8.29, K = $8.00, r = 4.0%, T = 0.4712 yrs.

⇒ S·e^(−qT) = 0.420 + 7.851 = $8.271 ⇒ q ≈ 0.5% p.a. The $9 strike returns a slightly negative implied q (bid-ask noise, and the option quotes appear to reflect a spot marginally above the $8.29 official close).

Conclusion: the options market embeds essentially no borrow premium. With a 339m-share float, 27.0m shares of average daily volume and index/ETF holders (Vanguard, BlackRock, Van Eck) supplying lendable stock, borrow is available and cheap — call it 0-2% p.a. Borrow is not the binding constraint on this trade, and it would have been wrong to assume otherwise.

Implied volatility and the implied move (Alpaca options snapshots, 2026-07-27)

ATM straddle, 21-Aug-2026 (25 days, spanning the 5-Aug earnings print): $0.850 + $1.030 = $1.88 on an $8.29 underlying ⇒ option-implied move of ±22.7%.

House expected move on a Q2 revenue miss: −10% central. Precedent — the Q1 print missed revenue by 89.9% and the stock did not break down on it, because the market has already stopped underwriting the revenue line.

⇒ The implied move materially exceeds the house expected move. Under the variance-risk-premium rule in references/alpaca-options.md, buying premium here is a negative-expected-value structure. This is a Gate 5 input and it fails.

7A. Transcript Mention-Frequency table (REQUIRED — references/mention-frequency.md)

History window: 16 quarters, Q2-2022 through Q1-2026 — NuScale's entire post-SPAC history. Source: stockanalysis.com transcript archive, cached to data/transcripts/. (Alpha Vantage's EARNINGS_CALL_TRANSCRIPT endpoint was quota-exhausted; the substitution is disclosed, and a cross-source discrepancy that was caught and resolved is recorded in Research Document §4A.1.) Chart: chart_36_mention_frequency.png.

(Full 16-column series for 47 terms in data/mention_freq_SMR.json; abbreviated here for width.)

In prose. Emerging: TVA, PPA, OEM, binding, term sheet, sites, liquidity — the vocabulary of a supplier to an intermediary, all dated to Q3-2025, the quarter the PMA was signed. Decaying: CFPP, UAMPS, DOE, data center, customers, order/orders, ATM — the vocabulary of a seller, plus the entire federal and AI-datacenter narrative. Never present at all: backlog, dilution, cancel.

The honesty rules applied. A single quarter's spike is not a trend, so the reads above rest on the prepared-remarks series, not the totals: "orders" at 18 in Q2-2025 collapsing to 0 unprompted for three straight calls is a trend; TVA at 43 in one quarter is not, on its own, and is treated as a question rather than a finding. Every emerging term was checked against an independent corpus (Research Document §4A.2): TVA/PPA/OEM against the filed PMA and the NRC docket, "sites" against NRC licensing records (which show none), DOE against DOE award announcements (which show none for this programme).

8. Factor & Anomaly Scorecard (required)

Synthesis. Ten of fourteen readings are tailwinds for a short, and not one fundamental-quality factor dissents. This is a textbook high-accruals, zero-F-score, collapsing-revenue, negative-momentum profile — the screen was right to rank it #1 of 27, and Gate 1 is corroborated exactly as references/trade-construction.md requires.

Every dissent sits in the implementation block: 20.8%-of-float crowding, a 2.28-day cover, 99.5% realised volatility and a documented +109% twenty-session rally. That is precisely where Gates 4 and 5 fail. The research conclusion and the portfolio action are different things, and here they diverge.

9. Valuation conclusion

The probability-weighted target is $8.41 against a spot of $8.29. After all the work — a 0/9 F-score, a 96% revenue collapse, a half-billion-dollar payment on a non-binding letter, a full sponsor exit, and an implied expectation of 1.47 TVA programmes — the honest, probability-weighted answer is that the stock is approximately fairly priced for its distribution of outcomes. An 84% drawdown has already discounted a great deal.

That single number is why this memo does not end in a Short. It flows straight into Gate 4.

Model: NuScale_Financial_Model_2026-07-27.xlsx, verified in Excel. Charts: NuScale_Charts_2026-07-27/ (35 charts, 300 DPI). Sources: SEC EDGAR XBRL CIK 0001822966; Alpaca Markets SIP bars and options snapshots; Alpha Vantage EARNINGS_ESTIMATES; stockanalysis.com, fintel, marketbeat, wallstreetzen; SEC Form 4 filings — all pulled 2026-07-27.