SNN — S.N. Nuclearelectrica S.A.

SNN — S.N. Nuclearelectrica S.A. (BVB) — Deep-value analysis (REFRESH)

Generated 2026-09-04. Reference price: … (tracker, 04.09.2026; yfinance intraday …) · Market cap: ~… · 301,643,894 shares · Exchange: Bucharest Stock Exchange.

REFRESH regime. This report updates the reference analysis from 2026-08-07 (verdict then: INTERESTING, GBL score …). What I explicitly inherit and do NOT re-derive: the business and moat description, the capital-allocation history under Cosmin Ghiță (CEO since September 2017), the April–May 2026 governance episode (the state voted against its own dividend proposal), the Thorndike grid applied to the CEO, and the shareholding structure. What I fully re-derive, on primary sources: H1/Q2 2026 results, the balance sheet, cash flows, the FCF bridge, all valuation models, and the verdict. Reason: the 14.08.2026 half-year report and a subsequent event that changes the thesis.

Symbol-collision note — and a correction to the input material. SNN on the NYSE is Smith & Nephew plc (medical devices, UK). Here I analyze S.N. Nuclearelectrica S.A., Yahoo symbol SNN.RO. The tracker confirms: exchange=BVB. The research brief prepared for this session (research-SNN-20260904.md) analyzes Smith & Nephew entirely — the wrong company — and was discarded in full; no figure from it appears in this report. For the same reason, the “9 new filings” that triggered the REFRESH triage (6-Ks from 12.08–02.09.2026, SEC-Filings\SNN\) are Smith & Nephew filings with the SEC — collision noise, not information about Nuclearelectrica. Nuclearelectrica doesn’t file with the SEC. The primary source used here is SNN’s integrated half-year report for H1 2026 (291 pages, consolidated and standalone statements, reviewed by PKF Finconta SRL, signed 12.08.2026 by Cosmin Ghiță and Daniel Adam), downloaded directly from nuclearelectrica.ro, plus current reports to the BVB and Romanian financial press for dating events. This is an analysis, not investment advice.

Executive summary

The thesis has flipped. The August analysis saw a company “cheap on paper, with weak 3-5-year visibility” and gave it a median margin of safety of …. Three facts that emerged since — all verified in the financial statements, not in the press — break the central pillar of that thesis.

First: earnings growth has already stopped, in the second quarter. Consolidated net profit for H1 2026 was 1,079.6mn RON, versus 826.9mn RON in H1 2025 (…). But Q1 2026 alone had brought in 851.5mn RON, so Q2 2026 produced only 228.1mn RON, versus 304.2mn RON in Q2 2025 — a …% year-over-year decline and a 73% decline from the prior quarter. The falsifier “earnings growth is real and sustained,” which the mechanical triage had flagged as unverifiable for lack of a second quarter, is now verifiable and has failed. Moreover, the entire half-year growth comes from the disappearance of the contribution to the Energy Transition Fund, which had cost 504.9mn RON in H1 2025 and was zero in H1 2026. H1 2025’s operating result, recalculated as if CFTE hadn’t existed, would have been 1,326.8mn RON; H1 2026’s is 1,120.7mn RON. Underlying, excluding the tax windfall, the operating result FELL by 15.5%.

Second: both reactors at Cernavodă are shut down, and the company is producing nothing. Unit 1 was shut down in a controlled manner and disconnected from the National Energy System on July 28, 2026, and Unit 2 on August 13, 2026, due to the unprecedented low level of the Danube, caused by severe drought. The auditor (PKF Finconta) included the fact in an emphasis-of-matter paragraph in the review report, and Note 30 treats it as a non-adjusting subsequent event under IAS 10. As of this report’s date, the reactors remain shut down: the Ministry of Energy announced on September 2 that the plant will not restart before September 10. This is the first time in the plant’s history that both units have been down simultaneously. The value-destruction mechanism is already visible in H1: having been sold forward on contracts, undelivered energy must be bought on the market. In H1 2026 SNN bought 474 thousand MWh (of which 469 thousand in Q2 alone) for 294.7mn RON — i.e. at …/MWh, above its own average selling price of …/MWh. A producer with near-zero marginal cost becomes, during the shutdown, a negative-margin buyer.

Third: the stock isn’t cheap versus its own history. The falsifier in the delta pack compared a P/E of 7.29x to a sector threshold of 11.4x and declared it “holds.” The relevant comparison is different. At … and a TTM EPS of …, SNN trades at P/E 7.56x and P/B 1.24x — both the highest values in the last five years (year-end P/E: 3.53x in 2022, 4.90x in 2023, 6.61x in 2024, 6.71x in 2025; P/B: 0.92 / 1.04 / 0.91 / 0.99). The price has grown faster than earnings, and the earnings the multiple is calculated on are the most inflated in company history and have just started contracting.

Estimated value. Normalized owner earnings (NOPAT at full capacity, post-refurbishment) that I declare as the base: …mn RON/year. The EPV model, adjusted for the present value of the production gap during Unit 1’s refurbishment period and for the net cash remaining after the July 13 dividend payment, gives …/share. The Monte Carlo simulation over 20,000 scenarios, with the same assumptions, gives a median intrinsic value of … and an undervaluation probability of …%. Triangulating five models gives a range from … (DCF bear) to … (Graham Number), with the median at …. The analyst average target implied by the tracker (disc_consens = −0.3108) implies ~… — practically identical to my EPV, which is the required external check: my figure isn’t an idiosyncratic outlier, it converges with what brokers publish.

Verdict: AVOID / MONITOR, downgraded from INTERESTING. Not because the business is bad — the asset is excellent and indispensable to Romania — but because today’s price embeds an earning power that (a) is the product of an already-capitalized tax windfall, (b) is about to be halved for nearly four years by Unit 1’s refurbishment, and (c) has proven vulnerable to a physical risk — the Danube’s flow — that neither the old thesis, nor the GBL score, nor the mechanical falsifiers had in their model. The re-evaluation catalyst isn’t a better price, it’s a number: the Q3 2026 result, to be published in November, which I estimate will be in operating-loss territory.