NuScale Power
Pre-revenue technology developer + OEM/licensor: NRC-certified SMR design licensed and sold (modules + FEED/engineering services) through exclusive global commercializer ENTRA1 to utilities, governments, and AI-datacenter offtakers; future recurring revenue from module fabrication, fuel, and plant services. Today substantially milestone/grant-funded, not yet self-sustaining.
The thesis on this name
State of Data-Center Power
AVOID as a near-term datacenter-power thesis (not a high-conviction fundamental short). NuScale owns the only NRC-certified SMR design and ~$1.0B liquidity, but its flagship project's costs ran from $5.3B (2021) to $9.3B (2023) — ~$20,000/kW, roughly 3x Oklo's ~$7,000/kW target (fact) — and real US reactor revenue likely waits until the early 2030s with no contracted US construction timeline. The valuation prices a datacenter-SMR ramp the deployment schedule cannot deliver this decade. The thermal-and-power demand is real, but NuScale's economics and timeline make it the wrong vehicle; avoid the long, and treat strength as a fade rather than a thesis.
State of Nuclear Energy
The most over-valued way to own a real regulatory asset. NuScale has the only NRC-approved SMR design — a genuine moat — but Q1'26 revenue collapsed 95% to $0.6M, meaningful reactor revenue is early-2030s (BofA Neutral, $12 PT), and prior order cancellations (UAMPS/CFPP) show how fragile the pipeline is. The valuation discounts flawless multi-GW deployment a decade out; the design approval is priced as if plants were already shipping.
State of Nuclear Energy
Short the most over-valued way to own a real regulatory asset: only-NRC-approved SMR design, but Q1'26 revenue collapsed 95% to $0.6M and meaningful reactor revenue is early-2030s — paired against the cash-flowing producers, not naked.
Earnings, margins, COGS & capex
NuScale is a pre-commercial story stock: revenue collapsed to $0.565M in Q1 FY26 (vs $13.4M a year earlier) as the RoPower license and Fluor FEED engineering work wound down, while net loss widened to $46.7M and operating cash outflow hit $314.7M — including a one-time $259.9M Milestone Contribution to partner ENTRA1 (fact, Q1 FY26 10-Q). FY2025 GAAP 'net income' of ~$356M is an accounting artifact of non-cash warrant/fair-value gains, not operations; the underlying business loses money on tiny, lumpy milestone revenue. The thesis rests entirely on converting a ~$25B ENTRA1 framework and a 6 GW TVA MOU into firm, paid module orders — none of which were signed as of Q1 FY26. The $1.0B+ war chest (no debt) buys runway but the company guides to operating-cash-flow positive only by late 2026, contingent on a TVA PPA and OEM contract that may slip (estimate/management guidance).
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~75¢ is cost of goods and ~24¢ operating expense, leaving ~1¢ of operating profit (~1¢ net).
Revenue trend
Margins
down (FY24 COGS only $4.9M as Fluor FEED ramped costs)
worsening
underlying loss widening
worsening (ENTRA1 milestone + long-lead prepayments)
COGS structure
COGS is engineering/EPC services cost — primarily Fluor's FEED Phase 2 work for RoPower (drove FY25 cost of sales to $20.0M from $4.9M in FY24) and design-licensing delivery. There is essentially no manufacturing COGS yet because no module has been sold/built for commercial operation; future COGS will be dominated by long-lead forgings/reactor-vessel fabrication (Doosan Enerbility), fuel (Framatome), and balance-of-plant. Cost structure today is R&D + SG&A heavy, not unit-economics driven (fact).
Capex
Conventional PP&E capex is modest; the real cash use is (1) the $259.9M ENTRA1 Milestone Contribution and (2) prepayments for long-lead materials (heavy forgings, steam-generator tubes) to de-risk a 2030 first-plant delivery — together driving the $314.7M Q1 FY26 operating cash outflow. This funds supply-chain readiness ahead of firm orders, i.e. spending to be 'shovel-ready' before a customer has paid (fact + estimate).
Latest earnings
Miss: revenue $0.565M vs ~$4.99M consensus (~89% downside surprise); EPS -$0.14 vs -$0.13 consensus. Stock fell sharply post-print (-8.7% intraday; some trackers cited ~-21% over the window) (fact).
Management targets operating-cash-flow positivity by late 2026, explicitly contingent on executing the TVA PPA and subsequent OEM contract payments; acknowledged RoPower FID could slip to 2027 and TVA PPA timing is uncertain (management guidance — high execution risk).
- Liquidity
- $1.0B at 3/31/26 → ~$1.2B by early May (no debt)
- Op cash outflow
- $314.7M (incl. $259.9M ENTRA1 milestone payment)
- Modules in fabrication
- ~12 modules in manufacturing toward a 2030 first-plant target (mgmt)
- Firm paid module orders
- 0 — no binding offtake/PPA signed as of Q1 FY26
Growth drivers
- AI-datacenter baseload demand — ENTRA1's up-to-$25B agreement to deploy NuScale plants across the U.S., plus the Standard Power 24-module / 1,848 MWe Ohio+Pennsylvania datacenter projects (fact, announced 2025)
- TVA + ENTRA1 non-binding MOU to deploy up to 6 GW of NuScale SMR capacity across TVA's seven-state region — the largest U.S. SMR program announced (fact)
- Regulatory lead — 77 MWe (US460) Standard Design Approval via NRC FSER (May 28, 2025) — the only NRC-approved SMR, a multi-year moat over Oklo/X-energy/TerraPower (fact)
- RoPower / Romania VOYGR-6 (462 MWe, six 77 MWe modules) — FID approved by Nuclearelectrica shareholders; first international flagship (FID could slip to 2027) (fact/estimate)
- Module fabrication + recurring fuel and plant-services revenue once plants are built — the eventual annuity if FOAK plants are completed (estimate)
- U.S. policy tailwinds — federal nuclear loan/grant support, DOE backing, and AI-energy national-priority framing (fact)
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-02-26. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
NuScale is the only NRC-approved SMR sitting on $1B+ of net cash at the exact moment AI datacenters are scrambling for firm, carbon-free baseload power; if even a fraction of the ENTRA1 $25B framework / 6 GW TVA MOU / Standard Power datacenter pipeline converts to binding orders, the regulatory lead and first-mover supply chain make it the default Western SMR vendor and the stock re-rates off a near-zero revenue base.
- Regulatory moat is real and years deep: the 77 MWe US460 has NRC design approval (FSER May 28, 2025) while most rivals are pre-licensing — a customer who needs power this decade has few approved options (fact)
- Best-funded pure-play SMR: ~$1.0–1.2B liquidity, no debt — can fund long-lead fabrication and survive multiple slipped quarters without an emergency raise (fact)
- Demand narrative is structural, not cyclical: ENTRA1's $25B agreement, a 6 GW TVA MOU, and Standard Power's 24-module datacenter projects point at the most durable power-demand wave in decades (fact, mostly non-binding)
- Optionality on a single catalyst: one firm hyperscaler/utility PPA validates the model and likely drives a large re-rating given the tiny current revenue base (estimate)
- Supply chain already moving: ~12 modules in fabrication at Doosan, Framatome fuel agreement — de-risks the path to a 2030 first plant if a deal closes 'soon' (fact/mgmt)
This is an AVOID as a near-term datacenter-power thesis: NuScale has effectively no commercial revenue ($0.565M in Q1 FY26), is burning >$300M/quarter, has zero firm paid orders, and its only built-cost data point — the cancelled CFPP at ~$20,000/kW — is ~3x competitors' targets, so the market is paying a ~$3.3B+ market cap for a non-binding pipeline and a 2030+ first plant that keeps slipping.
- No firm orders: the entire bull case is non-binding MOUs/frameworks (ENTRA1, TVA); RoPower FID could slip to 2027 and TVA PPA timing is unknown — revenue inflection is perpetually 'next year' (fact)
- Uncompetitive cost: CFPP flagship went $5.3B (2021) → $9.3B (2023), ~$20,000/kW vs Oklo/peers targeting $3,000–6,000/kW — until NuScale proves a cheaper built cost, datacenter buyers have better options (fact)
- Cash burn is severe and partly going OUT the door to ENTRA1 ($259.9M milestone payment in Q1 FY26) — the $1B cushion is finite if conversion keeps slipping (fact)
- GAAP profitability is a non-cash fair-value illusion; underlying business loses ~$45M+/quarter on ~$0.5M revenue (estimate)
- Intensifying competition: X-energy IPO'd April 2026 ($1.02B), plus Oklo, TerraPower (Natrium), Westinghouse AP300, Rolls-Royce — several with hotter/cheaper designs and their own datacenter deals; NuScale's lead is licensing, not economics (fact)
What it is worth
Optionality / scenario valuation (not earnings-based). With ~$0.5M quarterly revenue and non-cash GAAP profit, P/E (~-5x reported) and P/S are meaningless. Value = $1B+ net cash floor + risk-weighted option value on converting the ENTRA1 $25B / TVA 6 GW / Standard Power pipeline into firm module orders. Anchor to net cash (~$3/sh on ~346M sh) plus a deal-conversion premium; cross-check vs sell-side targets.
~$3–7 (sell-side low $7
drift toward net-cash floor if orders keep slipping and burn forces dilution)
~$10–15 (sell-side mean ~$15.4 across ~14 analysts; 'Neutral' consensus — pipeline credible but unsigned)
$25 (sell-side high
multiple firm orders + RoPower FID confirm the model converts)
Stock trades on binary catalyst risk: a single binding hyperscaler/utility PPA could re-rate it sharply; continued order drought drifts it back toward the cash floor. AVOID for the near-term datacenter-power thesis — own the optionality only at/near the cash floor, not at a multi-billion pipeline premium.
SWOT
Strengths
- Only SMR with U.S. NRC design approval — both the original 50 MWe and the uprated 77 MWe (US460) design (SDA via FSER May 28, 2025); a regulatory moat competitors are years behind on (fact)
- Strong balance sheet — ~$1.0–1.2B liquidity and no debt (Q1 FY26) — multi-year runway to fund supply-chain readiness without an immediate raise (fact)
- Marquee commercial framework — ENTRA1's up-to-$25B deployment agreement plus a non-binding 6 GW TVA MOU — the largest announced U.S. SMR pipeline (fact, non-binding)
- Established supply chain for long-lead items — Doosan Enerbility forging reactor vessels, Framatome for fuel — ~12 modules already in fabrication (fact)
- Direct AI-datacenter exposure via Standard Power (24 modules / 1,848 MWe) — aligned with the highest-conviction power-demand narrative (fact)
Weaknesses
- Essentially no commercial revenue ($0.565M in Q1 FY26) and no firm, paid module orders — all pipeline is non-binding MOUs/frameworks (fact)
- Severe cash burn — $314.7M operating outflow in one quarter; even the $1B+ cushion is finite if orders keep slipping (fact)
- FOAK cost problem — the abandoned UAMPS/CFPP flagship ran from $5.3B (2021) to $9.3B (2023) — ~$20,000/kW, far above competitors' $3,000–6,000/kW targets — an unproven, uncompetitive cost basis (fact)
- GAAP 'profit' is a non-cash warrant/fair-value mirage — underlying economics are deeply loss-making — screens can mislead (estimate)
- Revenue is lumpy and milestone-driven (RoPower license + Fluor FEED) — collapses to near-zero between project phases (fact)
Opportunities
- AI/datacenter baseload power is the strongest demand thesis in a generation; a single firm hyperscaler-backed order would re-rate the stock (estimate)
- Convert TVA 6 GW MOU and ENTRA1 $25B framework into binding PPAs/OEM contracts — the inflection management is guiding to (estimate)
- International expansion — RoPower/Romania flagship plus interest across Europe/Asia where NRC approval is a credibility passport (fact)
- Series-production learning curve — if FOAK is built, nth-of-a-kind module costs should fall sharply toward competitive LCOE (estimate)
- U.S. federal nuclear policy tailwinds (DOE support, loan guarantees, AI-energy prioritization) could de-risk first projects (fact)
Threats
- Execution/timeline slippage — RoPower FID may slip to 2027, TVA PPA timing unknown — every delay extends cash burn and tests patience (fact)
- Cost competitiveness vs Oklo, X-energy (IPO'd April 2026, $1.02B raised), TerraPower, Westinghouse AP300, Rolls-Royce — several with cheaper $/kW targets (fact)
- FOAK risk repeating the CFPP cancellation — cost overruns or an offtaker walking could collapse the thesis (fact)
- Equity-funding dependence: continued burn likely forces dilutive raises if orders don't convert (estimate)
- Macro/rate + nuclear-construction risk (skilled labor, forging capacity, regulatory licensing of specific sites) can blow out schedules and budgets (estimate)
Moats, dependencies & bottlenecks
Moats
High near-term (3–5+ yr lead), erodes as rivals get licensed Only SMR with NRC approval — both 50 MWe and uprated 77 MWe US460 (FSER May 28, 2025). The single most defensible asset (fact).
~12 modules in fabrication at Doosan Enerbility + Framatome fuel deal; hard to replicate quickly but not exclusive (fact).
Medium (and double-edged) ENTRA1 holds the $25B framework + TVA 6 GW MOU, but NuScale also pays ENTRA1 large milestone fees ($259.9M in Q1 FY26) — channel control sits partly outside NuScale (fact).
Medium (depletes with burn) $1B+ net cash funds survival/optionality but is not a structural moat — it buys time, not pricing power (fact).
Opposite of a moat today — FOAK ~$20,000/kW is ~3x peers' targets; no demonstrated cost advantage (fact).
Dependencies
ENTRA1 datacenter customers, Standard Power) Entire revenue inflection depends on converting non-binding MOUs to firm paid orders; none signed as of Q1 FY26 (fact).
Sales channel is exclusive to ENTRA1; NuScale's commercial fate is tied to ENTRA1's deal-closing and pays it large milestone fees (fact).
Heavy-forging and nuclear-fuel capacity are narrow global bottlenecks; schedule slips here directly delay first plant (fact). Non-US suppliers (Korea/France) — analysis only.
Design approval secured, but each plant still needs combined operating license / site approval — schedule and cost risk (fact).
>$300M/qtr burn means a prolonged order drought likely forces a dilutive raise despite today's $1B cushion (estimate).
Advantages
- Sole NRC-approved SMR design (50 MWe + 77 MWe US460) — a credibility and time-to-market edge no Western rival currently matches (fact)
- Strongest balance sheet among pure-play SMRs: $1B+ net cash, no debt (fact)
- Largest announced pipeline by headline GW — ENTRA1 $25B framework + TVA 6 GW MOU + Standard Power 1,848 MWe datacenter projects (fact, non-binding)
- Real, moving supply chain — ~12 modules in fabrication; ahead of pre-fabrication peers (fact)
- Direct, named AI-datacenter exposure (Standard Power) — clean alignment with the dominant power-demand thesis (fact)
- International flagship in motion (RoPower/Romania VOYGR-6) gives a non-US demonstration path (fact)
Weaknesses
- Near-zero commercial revenue ($0.565M Q1 FY26) and zero firm paid orders — pipeline is all MOUs (fact)
- Severe, partly outbound cash burn ($314.7M Q1 FY26 incl. $259.9M to ENTRA1) (fact)
- Uncompetitive demonstrated cost (~$20,000/kW CFPP) vs $3,000–6,000/kW peer targets (fact)
- GAAP profit driven by non-cash fair-value/warrant gains masks deep operating losses (estimate)
- Timeline slippage risk endemic (RoPower FID → 2027 possible; TVA PPA timing unknown) (fact)
- Commercial dependence on a single exclusive channel (ENTRA1) it does not control and must pay (fact)
Bottlenecks
- No binding paid order — the gating item for everything; the company can fabricate modules but cannot recognize commercial revenue until a customer signs and pays (fact)
- Heavy-forging and large-component manufacturing capacity (Doosan) — a physical global chokepoint for any SMR reaching commercial scale (fact)
- FOAK construction cost/schedule — the unproven $/kW that must come down from ~$20,000/kW for datacenter buyers to choose NuScale over alternatives (fact)
- Site-specific NRC combined operating license (COL) per project — design approval ≠ permission to build a given plant (fact)
- Cash runway vs deal timing — burn rate compresses the window in which non-binding pipeline must become binding (estimate)
- Nuclear-skilled labor and EPC capacity (historically Fluor) — execution bandwidth for first builds (estimate)
Top signals & trends
Top signals
The single biggest catalyst — would validate the model and likely re-rate the stock off a ~$0.5M revenue base (estimate). None signed as of Q1 FY26.
Flagship international build; a firm FID is bullish, a slip extends burn and dents credibility (fact).
$314.7M outflow in Q1 FY26; watch whether burn normalizes after the one-time $259.9M ENTRA1 payment or stays elevated (fact).
Guidance is contingent on a TVA PPA and OEM payments that may not arrive on time — high miss risk (mgmt/estimate).
Any data point materially below the ~$20,000/kW CFPP figure would weaken the bear cost case; silence here is itself a tell (fact).
Rival design-approvals or datacenter deals erode NuScale's regulatory-lead premium (fact).
Trends
Structural driver behind ENTRA1, TVA, and Standard Power interest — the core reason the stock has a narrative (fact).
Lowers first-project risk and accelerates licensing/financing (fact).
Erodes NuScale's first-mover premium and pressures pricing/economics (fact).
CFPP $5.3B→$9.3B is the cautionary tale; until series production proves out, cost is the bear's strongest card (fact).
Headlines are large (GW, $25B) but non-binding; market increasingly demands signed paid contracts (estimate).
NRC approval as a credibility passport opens non-US demand (fact).
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Korea — forges reactor pressure vessels / heavy forgings; ~12 modules in fabrication (non-US; analysis only)
France (EDF-owned) — nuclear fuel supply agreement for NuScale plants (non-US; analysis only)
EPC / FEED engineering services (RoPower FEED Phase 2); former majority owner — fully exited its NuScale stake by Q2 2026
Sweden — software/services for the RoPower project (non-US; analysis only)
Broader U.S. nuclear-fuel-cycle supplier base relevant to advanced reactors
Exclusive global commercialization partner; holds the up-to-$25B deployment framework — NuScale's de facto sales channel (private)
U.S. federal utility — non-binding MOU (via ENTRA1) for up to 6 GW of NuScale SMR capacity
Datacenter infrastructure-as-a-service — 24 modules / 1,848 MWe across Ohio & Pennsylvania (private)
Romania — VOYGR-6 (462 MWe) flagship; Nuclearelectrica is the state nuclear operator (non-US; analysis only)
Former flagship U.S. municipal-utility customer — project cancelled 2023 after costs ran to ~$9.3B/$20,000/kW (cautionary reference)
Aurora fast reactor, upsized to ~75 MWe for datacenters; build-own-operate model and aggressive cost targets (~3x cheaper $/kW vs CFPP). Pre-NRC-approval but strong datacenter narrative.
Xe-100 high-temp gas reactor (750°C, process heat/H2); IPO'd Apr 2026 raising $1.02B — now well-funded. Amazon-backed datacenter ties.
Natrium sodium-cooled reactor with thermal storage (dispatchable); Wyoming demo under construction. Different use-case but competes for the same baseload/datacenter dollars.
300 MWe PWR leveraging proven AP1000; design approval targeted ~2027, construction-ready ~2030. Incumbent credibility.
BWRX-300 boiling-water SMR; Ontario (Darlington) first-mover build underway — arguably ahead on an actual commercial construction start.
470 MWe PWR, UK-anchored; larger unit size, strong European pipeline.