Phase Space AI

NuScale Trade Construction 2026-07-27

NuScale Power [SMR]

NuScale Power [SMR] — Trade Construction & Risk Management

⚠️ SUPERSEDED IN PART — 2026-07-29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The Gate block and the Gate 4 expected-return arithmetic below are also superseded, by the named Criteria (each with a type: BINDING or MEASURED, returning PASS / FAIL / INDETERMINATE), the reverse-DCF implied-path test, and a separate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 5 · 2026-07-27 · Spot $8.29

---

POSITION VERDICT: RETIRED 2026-07-29 — the memo outputs an analysis, not a position.
                  See NuScale_Criteria_and_Valuation_2026-07-29.md for the current Criteria block.
The gate lines below are the 2026-07-27 record and are superseded.
GATES: 1(Mechanism): PASS (strengthened)   2A(Estimate variant): PASS (qualified)   2B(Duration): n/a — short
       3(Catalyst): PASS   4(Expected Return): FAIL   5(Feasibility): FAIL   6(Momentum/Tape): PASS
ENTRY: no entry authorised at $8.29  |  TARGET (research view): $6.00 (was $6.75)  |  INVALIDATION: $12.50
TIME HORIZON: n/a — no position. Re-underwrite after the TVA PPA question resolves (guided year-end 2026).
SCENARIO-WEIGHTED E[R] (outright short, 12m): GROSS +2.3% → NET −0.2% after borrow and slippage
   (v1.3.0: gross −1.5% / net −4.0%. The thesis got materially better AND the right tail got fatter;
    the two corrections nearly cancel, and the gate still fails by 4.9pp against the 4.7% cash hurdle.)
   Simple framing: risking ~117% to make ~28%.
SIZING: Conviction Medium-Low (composite 4.5/10) | Volatility HIGH (99.5% realised, beta 3.91)
   → sizing grid output 1% → below minimum viable size for the convexity → rounds down to Watchlist
VEHICLE: none. Both the equity short and every defined-risk options structure priced today have negative E[R].
INVALIDATION / CONVERSION TRIGGERS: see §8

---

1. The four conclusions, kept separate

The failure mode this section exists to prevent is reasoning as one narrative ("the company is bad AND expensive AND therefore a short"). The four conclusions are stated independently.

1.1 Fundamental conclusion

Deteriorating, and the deterioration is observed rather than forecast. Revenue fell 95.8% YoY to $0.565m in Q1-2026; deferred revenue is $1.409m; there are zero binding orders. The company expensed $507.4m and paid ~$495m in cash to ENTRA1 — triggered by a non-binding term sheet. Piotroski F-score 0/9, accruals +0.106, gross profitability 0.008. But: $1,013.9m of cash and investments, zero borrowings, and 5.9 years of runway. This is a business with no revenue and no solvency problem.

1.2 Expectations conclusion

ΔE is large on revenue and on the price target; it is zero-to-negative on EPS.

The variant is real but it is specific: it concerns the revenue trajectory, the ENTRA1 cash-obligation structure, and the implied module count — not the P&L loss, where the Street is already more bearish. Critically, the house target of $6.75 sits between Goldman's $6.00 and Citi's $7.50 — at the bearish edge of the Street, not beyond it.

1.3 Valuation conclusion

Under the differentiated forecast, fair value is $6.75 (−18.6%), with a bear at $4.50 and a net-cash floor of $2.76. Today's $2,021.6m EV implies 8.8 GW = 1.47× the entire announced TVA programme must eventually be delivered, against 0.0 GW of binding orders. The stock is demanding relative to contracted reality.

But the probability-weighted target is $8.41 against a spot of $8.29. An 84% drawdown has already discounted most of what this analysis identifies. What is left to capture is roughly 19% of modal downside against a 105% right tail — and that ratio, not the fundamental view, is what decides this memo.

1.4 Portfolio conclusion

No position. The research conclusion (bearish) and the portfolio action (pass) are different things, and here they diverge. Detail in §3-§7.

---

2. The six hard gates

Gate 1 — Causal mechanism · PASS

A specific, named, evidenced mechanism, not a valuation observation:

NuScale has contracted away the commercialisation of its own product to 2045. ENTRA1 — not NuScale — is the counterparty to every customer document. Under the filed Partnership Milestones Agreement NuScale pays ENTRA1 in cash as commercial progress occurs: ≈$46.98m per 77 MWe module in total (15% on a non-binding letter, 35% on ENTRA1's binding PPA, 50% on the OEM agreement), escalating at the greater of 5% or CPI every 1 January to ≈$63m by a 2031 delivery, guaranteed by the listed parent as primary obligor and payable within 60 days of invoice with no set-off. $507.4m was expensed and fully paid on a non-binding MOU alone. Simultaneously the revenue that funded the P&L (RoPower Doicesti FEED Phase 2) has run off, leaving $0.565m of quarterly revenue and $1.4m of deferred revenue — and management's own guidance for what a TVA PPA is worth in revenue is "about $8 million" on the RoPower analogue, against $207m of Milestone 2 cash per twelve-module plant. A ratio of roughly 26 : 1, cash out to revenue in, at the moment the good news arrives.

Evidence that it is underway, not hypothetical — all from filings:

Quantitative corroboration (the standard references/trade-construction.md requires): accruals +0.106 (Sloan 1996 — poor earnings quality), F-score 0/9 (Piotroski 2000 — worst possible), gross profitability 0.008 (Novy-Marx 2013), asset growth +159% (Cooper/Gulen/Schill 2008). This is exactly the empirical profile in which shorts work; the scorecard does not dissent anywhere on fundamentals. Gate 1 passes on its strongest possible footing.

Gate 2 — Negative variant vs. consensus · PASS (qualified)

Consensus is not already fully bearish, which is the specific risk this gate exists to catch on a −80% momentum name: - 6 of 18 analysts are still Buy or Strong Buy; - the average target is $14.57 = +75.8%; - consensus FY2026 revenue is $53.5m against a $2.3m annualised run-rate — the Street requires a 31× step-up across Q2-Q4 with no announced contract to produce it; - the Q1-2026 print missed consensus revenue by 89.9% and the estimates for Q2 ($11.9m) and Q3 ($16.6m) were only partially cut.

Why consensus is wrong, specifically: the Street is modelling revenue as if NuScale sells reactors. Under the ENTRA1 structure NuScale's near-term revenue is bespoke engineering services to one Romanian project, and that work has run off. The Street is also valuing a successful TVA programme as pure upside while the same event triggers a cash obligation to ENTRA1 that management has never quantified on a call.

v1.4.0 correction, made against the memo's own interest. The prior version put that obligation at $1,246.8m = 123% of the balance sheet, at signature. The filed agreement says otherwise, and the correction must be stated plainly:

Net effect on the short case: the mechanism is stronger and more permanent, but the near-term funding shock is smaller and later than the prior memo assumed. That is bad for a short, because the funding shock was the only thing that would have offset a PPA announcement's squeeze. It is carried straight into Gate 4 rather than being buried here.

The qualification, stated plainly, not buried: on EPS the Street's FY2027 number (−$0.734) is worse than the house number (−$0.53). And the house price target of $6.75 is not below the Street low of $6.00. This memo is at the bearish edge of published opinion, not beyond it. That materially weakens the "variant" claim relative to a case where the house is alone. Gate 2 passes on revenue and structure; it would fail if the claim were about the loss line.

Gate 3 — Catalyst that resolves this disagreement · PASS

These are not "scheduled events near the position" — the Q2 print tests the exact metric Gate 1's mechanism moves, and the PPA tests the exact structural claim. Gate 3 passes cleanly.

Note the asymmetry, because it matters for Gate 4: the TVA catalyst is bidirectional and violent. It is the single strongest catalyst for the short and the single largest squeeze risk.

Gate 4 — Positive scenario-weighted expected return, net of costs · FAIL

Scenarios over 12 months to July 2027 (probabilities logged for later calibration scoring). Both the targets and the right tail move under v1.4.0, in opposite directions:

Gate 4 still fails, by 4.9 percentage points against the 4.7% cash hurdle — but the margin has narrowed from 8.7pp to 4.9pp, and it is worth being precise about why, because the two effects are independent:

• the thesis improved materially (E[R] +3.8pp from the corrected fee flowing through Bear and Base targets);
• the convexity worsened (E[R] −0.5pp from a higher bull target, because the funding shock that would have blunted a PPA rally turns out to be capped and deferred).

The probability-weighted target is now $8.10 against a spot of $8.29 — a 2.3% gap, versus a negative gap in v1.3.0. There is, for the first time, a sliver of edge. It is nowhere near enough.

What it would take. At p_bull = 0.20 (residual to Bear) the same targets give gross +10.9%, net +8.4%, clearing the hurdle by 3.7pp. That is the honest statement of how close this is: the entire decision now rests on one number — the probability of a binding TVA PPA inside twelve months — and management guides to "later this year." Gate 4's own language still governs ("a mildly positive E[R] with severe right-tail risk can still fail this gate"), and the right tail here is a documented +109% in 20 sessions.

Would a defined-risk options structure rescue it? Tested, and no.

Per references/alpaca-options.md, the default is a defined-risk spread, not naked premium. Every candidate was priced from live Alpaca snapshots on 2026-07-27 and every one has negative E[R].

(a) Structural: 15-Jan-2027 $8 / $5 put spread (172 days; spans Q2, Q3 and the guided PPA window) - Buy SMR270115P00008000 at $2.00 ask (IV 101.3%, δ −0.334) · Sell SMR270115P00005000 at $0.49 bid (IV 103.3%, δ −0.136) - Net debit $1.51 · width $3.00 · max gain $1.49 · max loss $1.51 · breakeven $6.49 (−21.7%) - E[R] against 6-month interpolated scenario prices ($5.00 / $7.25 / $15.00 at 40/35/25): 0.40 × (+98.7%) + 0.35 × (−50.3%) + 0.25 × (−100%) = −3.1%

(b) Tactical: 21-Aug-2026 $8 / $6 put spread (25 days; 16 days past the 5-Aug print) - Buy SMR260821P00008000 at $0.73 ask (IV 104.3%, δ −0.386, OI 6,553) · Sell SMR260821P00006000 at $0.10 bid (IV 112.1%, δ −0.102, OI 4,047) - Net debit $0.63 · width $2.00 · max gain $1.37 · max loss $0.63 · breakeven $7.37 (−11.1%) - Requires an 11.1% fall in 25 days merely to break even. The option-implied move is ±22.7%; the house expected move on a Q2 revenue miss is −10%. The market already prices more downside than the house forecasts — the explicit Gate 5 options test, failed. - E[R] across a four-point event distribution ($5.50 / $7.00 / $8.30 / $10.00 at 10/35/35/20): −12.7%

(c) Bear call spread (sell the expensive IV instead of buying it): 21-Aug $10 call at $0.34 bid / $13 call at $0.12 ask ⇒ credit $0.11-0.22 against a max loss of $2.78-2.89. A ~13:1 loss-to-gain ratio on a name that has printed ten +20% single days. Rejected on inspection.

Root cause, stated once: at ~100-112% implied volatility across the surface, the options market has already priced the two-sided distribution correctly. There is no mispriced expression of this view available today.

Gate 5 — Implementation feasibility · FAIL

Gate 5 fails, and v1.4.0 does not rescue it — nor could it. As the brief anticipated: a 20.8%-of-float short at 99.5% realised volatility is implementation-constrained regardless of how good the thesis becomes. Better information changes the numerator of the trade; it does not change the borrow crowd, the retail register, the option-implied move, or the book's −2%-of-book hard stop. Nothing in the filed PMA touches any of those.

The failure is not borrow — borrow is cheap and plentiful, which was worth establishing rather than assuming. It is that the maximum plausible loss on any size large enough to matter breaches the book's pre-committed position hard stop, on a catalyst that is scheduled to occur inside the horizon.

Gate 6 — Don't fight the tape · PASS (with the crash-risk caveat named)

12-1 momentum −76.7%; price at 15.5% of the 52-week high; −51.5% vs. the 200-day MA. A short would be with the prevailing momentum, not against it. Gate passes.

The caveat that must be named (Daniel & Moskowitz 2016): momentum crashes hardest precisely in sharp reversals off panicked bottoms in high-volatility, high-short-interest names. SMR — 99.5% realised vol, 20.8% of float short, 43% retail, sitting on a binary catalyst — is close to the canonical momentum-crash setup. So the gate that passes most easily is also the one most likely to reverse violently, which is a reason to weight it less, not more.

---

3. Portfolio context — the portfolio-book contract

Read from /Users/oahmady/Desktop/Projects/Investing Hub/portfolio_book.json (as_of 2026-07-27).

Current state: zero positions; 100% cash; one watchlist entry (NET, short-if-converted).

1. Marginal correlation — no positions exist, so pairwise correlation is undefined. For the record, SMR's 1-year daily correlation to SPY is 0.50 with a beta of 3.91; against the NET watchlist entry the relevant future check would be a shared "high-multiple story-stock de-rating" factor rather than a sector overlap.
2. Exposure math — gross 0% / net 0% before; unchanged after, since the decision is no position. No breach.
3. Sector concentration — advanced nuclear / power infrastructure: 0% before, 0% after. Limit 25%.
4. Capital competition — the operative test on an empty book. The candidate's net E[R] is −4.0%; the cash hurdle is +4.7%. Cash wins by 8.7 percentage points. This is not "better than nothing but not compelling" — it is materially worse than the risk-free alternative. No position competes into the book.
5. Book update — a Watchlist entry with an explicit conversion trigger is written back (see the PENDING file and §8). Per the skill, a Watchlist decision is still a decision and still writes to the book.

Dynamic risk protocol note. Had the gates passed, this position would have lived under the −2%-of-book hard stop. At 99.5% realised volatility, the arithmetic is: a 2% short position reaches a −2%-of-book loss on a +100% underlying move; a 1% position on +200%. The name printed +109% in 20 sessions in 2025. The protocol is therefore not a theoretical backstop here — it would very plausibly be triggered, forcing a mechanical cut to half size and a full re-underwrite at the worst possible moment. Sizing a position that the book's own wartime rules would predictably stop out is not risk management, it is a scheduled loss.

---

4. Sizing — computed even though the answer is zero, so the arithmetic is auditable

Weighted conviction composite per references/trade-construction.md:

Volatility tier: HIGH — 99.5% realised (threshold >45%); beta 3.91 (threshold >1.5). Not a guess; computed from Alpaca daily bars.

Sizing grid (house max single-name 5%):

Medium-Low conviction × High volatility ⇒ 1% or below ⇒ rounds down to Watchlist, which is where the hard-gate rule had already put it. The two routes agree.

Consensus positioning, stated rather than absorbed into the grid. This would be a position against a Street whose average target implies +75.8% upside — but that consensus is weakening fast (Goldman $9→$6 on 2026-07-15; Barclays $15→$11 on 2026-07-22), and the house target is not below the Street low. So this is not a lonely contrarian call with unquantified crowding risk; it is a call that agrees with the two most bearish houses on the Street. That is a reason not to shrink size for contrarian risk — and it is also a reason to doubt that much excess return remains, since the bear case is already published.

---

5. Vehicle selection — for completeness, since the answer is "none"

If the gates were passed at a later date, the vehicle of choice would be the Jan-2027 $8/$5 put spread (or its then-equivalent) — defined risk is mandatory on a name with this right tail, and an outright short would be inappropriate at any size regardless of conviction.

IV-crush disclosure (required whenever a catalyst-dated options trade is discussed, even a rejected one): implied volatility of 104-112% into the 5-Aug print will collapse immediately afterwards regardless of direction. A correct directional call that produces a move smaller than ±22.7% loses money on the August structures. This is a further reason the tactical trade is rejected, not a footnote to it.

---

6. The seven questions every Long/Short must answer — asked, and answered honestly

1. What is the market wrong about? That NuScale's revenue re-accelerates to $53.5m in FY2026 (vs. a $2.3m run-rate), and that a binding TVA PPA is unambiguously good news for SMR equity when it triggers a $1,246.8m payment to ENTRA1 — 123% of the balance sheet.
2. What evidence supports it? Filed facts: $0.565m Q1 revenue, $1.4m deferred revenue, $507.4m expensed to ENTRA1, ~$495m paid, the reported $16m/reactor fee, Fluor's complete exit, zero insider buys, F-score 0/9.
3. Why does it resolve now? Q2 results on 2026-08-05; TVA PPA guided by year-end 2026.
4. What prevents the company growing into the valuation? Nothing structural — with $1.01bn of net cash and 5.9 years of runway it has plenty of time. This is the honest weakness in the bear case.
5. What would invalidate it? §8.
6. What is the adverse scenario? A binding 6 GW PPA into a 20.8%-of-float short base with 43% retail ownership and 99.5% realised volatility. The stock was $40 on the non-binding MOU. Modelled at +105%; plausibly worse.
7. Why is taking the position superior to not owning it? It is not. Net E[R] −4.0% against a 4.7% cash hurdle. That is the whole answer, and it is why this memo ends in a Watchlist.

---

7. What is actually left to capture — the question a −80% momentum name demands

The stock is down 83.8% over twelve months and 84.5% from its high. It is fair to ask what remains.

Ratio: risking ~117% to make ~28%. That is the memo in one line, and it is a better ratio than v1.3.0's 105:19 — but the direction of the improvement matters less than the fact that it is still the wrong side of 1:1 on a name whose catalyst is bidirectional and whose float is a fifth short.

---

8. Watchlist entry — conversion and invalidation triggers

Written to portfolio_book.watchlist (via PENDING_book_and_ledger_updates.json).

Converts to SHORT if all of the following hold

1. The TVA squeeze catalyst resolves or expires. Either (a) 2026-12-31 passes with no binding PPA signed and no equivalent firm order, collapsing the near-term right tail; or (b) a binding PPA is signed and the associated 8-K discloses a milestone obligation ≥$800m payable in cash, at which point the funding hole is confirmed and can be shorted into strength rather than ahead of it.
2. Short interest falls below 12% of float (from 20.8%) and days-to-cover rises above 4 — i.e. the crowd has left and the convex tail has thinned.
3. The revenue variant is confirmed by a print: reported quarterly revenue below $3m for two consecutive quarters with Street FY-forward revenue still more than 3× the annualised run-rate (the gap must survive, not just exist).
4. A defined-risk structure with positive E[R] is available: a put spread whose scenario-weighted E[R] exceeds the 4.7% cash hurdle net of costs — which requires implied volatility below roughly 70% or a materially higher entry price.
5. Entry only on strength: $11.00+, or after a failed rally that leaves a lower high below the 200-day MA.

Converts to LONG if

1. A binding TVA PPA is signed and the 8-K discloses that the number of NPMs invoiceable for Milestone 2 before the redacted 2027 date is small (say ≤12, i.e. ≤$207m of near-term cash) — the one materially decision-relevant fact still redacted from the filed agreement; and
2. NuScale-recognised revenue or firm backlog exceeds $150m; and
3. The stock is still below roughly 1.5× net cash (~$4.15) at that point. (Low probability, explicitly retained so the watchlist is not a one-way ratchet.)

Research-view invalidation (the level at which the bearish view itself is wrong, not merely untradeable)

• $12.50 — a sustained close above this reclaims the 50-day MA and implies the market is pricing a PPA as probable; the modal path in this memo would be wrong.
• Q2-2026 (5 Aug) revenue above $8m — the run-off thesis is wrong.
• ~~Any 8-K disclosing that the ENTRA1 milestone obligations are payable in equity rather than cash~~ — RETIRED under v1.4.0. This trigger can no longer fire. PMA §1(e) requires USD wire transfer within 60 days, "absolute, unconditional… free and clear from any set-off", "final, irrevocable, non-cancelable and non-refundable"; the Guaranty makes NuScale Power Corporation a primary obligor payable within five days of an email. Only a renegotiation or termination of the PMA — which requires ENTRA1's written consent and would itself be an 8-K event — now defuses the mechanism.
• A firm, binding reactor order placed with NuScale directly rather than through ENTRA1 — the structural claim is wrong.
• Total cash & investments above $1.1bn at any quarter-end with revenue above $25m annualised — the burn thesis is wrong.

---

9. Recommendation-ledger note (Task 9)

No entry is appended to trade_recommendations.jsonl. Per references/recommendation-ledger.md, the ledger records specific recommendations made; this memo makes none. Every structure priced today was rejected on negative expected return.

The scenario probabilities (bear 0.40 / base 0.35 / bull 0.25, unchanged) and the revised computed E[R] (gross +2.3%, net −0.2%; v1.3.0 was gross −1.5% / net −4.0%) are recorded here, in the manifest, and in the watchlist entry so that the decision not to trade can be scored later against what actually happened — which is the calibration-relevant artefact even though no capital was committed.

---

Task 5 executed under investment-memo v1.4.0 (re-run 2026-07-27; supersedes the v1.3.0 pass) and the portfolio-book contract. Options quotes: Alpaca v1beta1/options/snapshots, 2026-07-27. Price/vol/beta: Alpaca SIP daily bars. Consensus: Alpha Vantage EARNINGS_ESTIMATES (cached) and stockanalysis.com, 2026-07-27. ENTRA1 terms: SEC EDGAR filed exhibits 0001822966-25-000144 Ex. 10.1/10.2 and 0001822966-25-000088 Ex. 10.1. Transcript corpus: 16 quarters, Q2-2022–Q1-2026, stockanalysis.com.