Management summary — latest findings
Consolidated read for the shareholders, current as of 10 Jul 2026. The tool above is the model; this is the decision. Every number here is either verified against primary data or run through the dispatch engine on the right. Backing detail lives in the vault research docs (feasibility, nuclear/grid regime, carbon). Not investment advice.
① The verdict
Calafat is an equipment-confirmed 4 MW solar + 16 MWh battery deal with an honest ~18% equity IRR energy-only, ~22% with ancillary on the promoter's low EPC cost. Its return is genuine — but its scarcest asset is the grid-connection right (ATR) that the Romanian market can no longer supply to new projects. Proceed toward the 30 Jul binding offer, close the open technical/legal items, and design the asset to be sellable.
② The deal & the now-locked configuration
- Structure: buy the two AMG SPVs (STAR INVEST subsidiaries), share deal. Entry €850k incl. land + all ATR grid rights + ETL. Binding offer due 30 Jul 2026; COD ~2027.
- Solar: 8.55 MWp AIKO 775 Wp bifacial (N-type ABC), behind a 4 MW export gate (2×2 MW SPVs, sub-5 MW by design).
- Battery: 16 MWh = 4×4 MWh containers, 0.25C → 4 MW (a 4-hour battery, power = gate). Round-trip 87.6% measured at the meter, cooling included. Turnkey AC (not a DC-block). Make: Huawei LUNA2000 vs Chint — to confirm.
- Connection is non-firm (curtailable) — get the expected lost-hours number.
③ The economics (locked config, post-2021 basis, EPC-low cost)
| Scenario | Equity IRR | Project IRR |
|---|
| Energy only | 17.7% | 13.3% |
| + Ancillary (aFRR/FCR) | 21.6% | 15.4% |
| + Bifacial upside (+6%) | 22.8% | 16.0% |
Build ~€7.5M. Drivers: ancillary ≈ +4 pts, grid-charging arbitrage ≈ +4 pts, battery degradation ≈ −1 pt, bifacial ≈ +1 pt (upside not in the promoter's yield). The promoter's headline "22.86%" is a levered, 0%-tax figure; on honest tax + our dispatch it is the ~18–22% above.
④ How it actually runs (verified on real hourly prices, 2021/23/24)
- ~1 cycle/day, never more — the round-trip toll + the 4 MW gate leave one profitable charge-cheap/sell-dear swing per day.
- Summer: the battery fills from the plant's own cheap/clipped midday solar. Winter: from cheap overnight grid. It picks whichever is cheapest, hour by hour.
- Grid-charging is optional — remove it and the deal still stands at ~15–18% (solar-only keeps 58–72% of battery revenue). The grid layer is the spread-regime-exposed kicker, not a crutch.
- The battery is armed ≥90% full on 81% of days to serve the evening peak. Degradation ≈ 83% capacity at year 10.
⑤ Why the prices should hold — the regime bet
Nuclear (near-term tailwind): Cernavodă Unit 1 (700 MW) is offline for refurbishment ~Dec 2026 → Dec 2028 — Romania runs on one reactor, tightening the market exactly in Calafat's early payback years. From ~2030, Units 3&4 (+1,440 MW) and the Doicești SMR (+462 MW) add baseload and normalise prices — but that pressures the price level (solar capture) more than the arbitrage spread (battery), which the deepening solar duck curve keeps alive. Use the "Nuclear-phased" spread path above to see this.
Grid (structural moat): Transelectrica's 2028–2037 capacity map shows 0 MW available in many zones — ~60 GW of approved projects chase ~11.5 GW of integrable grid room. A valid ATR is now a scarce, appreciating asset — and Calafat's SPVs already hold theirs. That is value the IRR doesn't capture: a moat plus a build-to-sell exit to buyers who can't get their own connection.
⑥ Checked and set aside
- Carbon / green certificates: the "€30/certificate" was the closed Romanian green-cert scheme (new plants ineligible); the real figure is €0.30 — a Guarantee of Origin. Realistic income ~€30–70k/yr from 2027. Negligible; not a pillar.
- Curtailment protection: the promoter's "15-year priority dispatch" reassurance was refuted — non-firm connection carries no compensation. Priced as a risk, not a comfort.
⑦ Risks & open items
- Regime risk — the whole return leans on post-2021 prices holding (nuclear timing + grid scarcity support this, but it's the core bet).
- Non-firm connection — get real expected lost-hours.
- Confirm with Erkan: battery make (Huawei/Chint) · RO 50 Hz/20 kV variant · bifacial in his yield · green-cert basis.
- Confirm ourselves: does the ATR permit 8.55 MWp DC (drives ~17% clipping)? · does the selective ≥5 MW allocation regime touch the 2×2 MW ATRs? · firm vs non-firm value in a zero-capacity zone.
⑧ Recommendation
Proceed to the binding offer. The equipment and economics hold at ~18–22% on the promoter's cost, the near-term regime is favourable, and the ATR scarcity is a genuine downside hedge and exit story the IRR understates. Close the technical/legal items first, price the non-firm curtailment honestly, and structure the asset (NL/Cyprus holding over an RO SPV) so it can be sold as easily as held.