ECONOMIE
ALFRED at Mioveni — what Romania is buying with its lead-cooled reactor and what alternatives it has for the 2030s
ALFRED, the lead-cooled reactor Romania wants to host at Mioveni, is feasible as a research programme and as an industrial bet, but not as a source of power before 2040, and the money being spent now buys data and skills, not megawatts.
The analysis follows five threads: what ALFRED is and why lead; what is being built at Mioveni now and with what money; how the schedule has slipped since 2013; what makes the project technically and financially hard; and what alternatives Romania has for nuclear power in the 2030s, in an autumn when the Cernavodă plant stands idle for lack of water. We end with three scenarios for 2030, the forecast table and the thresholds at which we would consider ourselves wrong.
The testable thesis: the question “ALFRED or the American small reactors?” is the wrong one. The two do not compete for the same decade. ALFRED competes with other routes into fast reactors and should be judged by a single test: whether the test facilities produce, by 2028, data that nobody else in Europe would otherwise have.
What ALFRED is, and why lead
ALFRED (Advanced Lead Fast Reactor European Demonstrator) is a demonstration fast-neutron reactor cooled by molten lead. The final design has 300 thermal megawatts and about 125 electrical megawatts, according to the consortium's presentation to the Generation IV International Forum in 2018. The planned fuel is MOX, a mix of uranium and plutonium oxides.
Lead has three advantages over water. It boils only at about 1,750 degrees, so the primary circuit runs at atmospheric pressure and cannot lose its coolant by boiling. It does not react violently with water or air, as the sodium of other fast reactors does. And lead does not slow neutrons down, so the reactor can burn part of the long-lived waste. The price is just as clear: lead freezes at about 327 degrees, corrodes steels at high temperature and is very heavy, which complicates earthquake protection.
That is why the design was rewritten in stages. Ansaldo Nucleare's presentation at the International Atomic Energy Agency course held at RATEN ICN itself on 18 February 2026 shows how the reactor would grow step by step, as materials are qualified:
| Stage | Thermal power | Core outlet temperature | What it demonstrates |
|---|---|---|---|
| 0 (commissioning) | almost zero | 390 °C | the systems, without power |
| 1 | 100 MWt | 430 °C | existing technology: austenitic steels, oxygen control |
| 2 | 200 MWt | 480 °C | coatings and new materials |
| 3 | 300 MWt | 520 °C | the performance targeted for commercial reactors |
The same presentation states openly that “there is a gap between current technology and commercial target” and that a first-of-a-kind remains a first-of-a-kind, with all its risks. In the cold lead pool the temperature is capped at 400 degrees precisely so that today's materials survive. Put differently, ALFRED's first stage is a 100-thermal-megawatt test reactor, not a power plant.
What is being built at Mioveni now
At Mioveni no reactor concrete is being poured. What is being built are the laboratories that must produce the safety data required for a future licence. The first piece is ATHENA, a 2.21-thermal-megawatt pool-type facility holding 880 tonnes of liquid lead in a vessel 3 metres wide and 10 metres high. The contract, worth about €20 million, was won in October 2021 by Ansaldo Nucleare and Reinvent Energy, with an announced duration of about two years.
The second piece is the 4ALFRED project, with a total value of 561.5 million lei, of which 438.3 million comes from the European Regional Development Fund, through the Smart Growth, Digitalisation and Financial Instruments programme. The financing contract was signed on 7 March 2025 and the project was launched on 10 July 2025. RATEN presents it as more than €112 million for six facilities: ATHENA, ChemLab, HELENA-2, ELF, HandsOn and Meltin’Pot.
The big 2026 contracts went to the same firm. On 2 March, SRS (Servizi di Ricerche e Sviluppo), part of the newcleo group since 2023, won a €36 million contract for HELENA-2, ELF and the modules that study how lead interacts with fuel in an accident, for delivery by the end of 2028. In May the same firm won the HandsOn facilities, for 54.5 million lei. On 21 October RATEN launched an 11-million-lei purchase of high-purity lead, reported in the press as about 1,000 tonnes.
The infrastructure has real value beyond the reactor. Lead loops and pools are rare worldwide, and their data are needed by every designer of a lead-cooled reactor, including those who will never build in Romania. The Director-General of the OECD Nuclear Energy Agency visited ATHENA and ChemLab in November 2025.
The schedule that slipped
The FALCON consortium (Ansaldo Nucleare, ENEA and RATEN ICN) was formed on 18 December 2013. The plan at the time foresaw construction from 2017 and reactor start-up in 2025. In 2014 Romania decided on the Mioveni site and committed to cover between 5% and 20% of the cost, and in May 2017 it took on the 20% share. The total estimate was about €1 billion, a figure not publicly revised since.
In 2023 the partners signed a memorandum on small lead-cooled reactors with the Belgian research centre SCK CEN and with Westinghouse. In June 2025 RATEN, ENEA, Ansaldo Nucleare and SCK CEN formed the EAGLES consortium, which aims at a 300-electrical-megawatt commercial reactor, EAGLES-300, for 2039. Westinghouse no longer appears among the EAGLES partners. The order of steps has changed: the first demonstrator will be LEANDREA, a low-power reactor at Mol, in Belgium, targeted to operate in 2034 and paid for from the Belgian budget. ALFRED is step two. In February 2026, EAGLES and newcleo agreed to work together on LEANDREA.
The move has a technical logic, because learning on a small reactor is cheaper. It also has a consequence for Romania: Europe's first lead-cooled reactor will not be at Mioveni, and SCK CEN's pages give no date for ALFRED.
What makes the project hard
There are five known problems, and the consortium acknowledges all of them in its own documents.
- Materials. At high temperatures lead attacks steels; that is why the first stage stops at 430 degrees and the cold lead in the vessel at 400. Stages 2 and 3 depend on coatings and materials not yet qualified. If they fail the tests, ALFRED stays at 100 thermal megawatts and the economic case for the commercial reactor weakens.
- Freezing. Lead solidifies at 327 degrees. The 2026 presentation sets the accident minimum at 330 degrees, a thin margin between overcooling and a blocked circuit.
- Earthquakes. The large mass of lead requires two-directional seismic isolators and a hanging vessel. Vrancea earthquakes are felt in Argeș too, and the licence will require the demonstration on site.
- Fuel. At Mioveni Romania makes natural-uranium fuel for CANDU reactors, not MOX. ALFRED's fuel would have to come from abroad. An example of how young the supply chain is: newcleo is only now licensing its fast-reactor MOX plant in France, and the French safety authority's opinion on its safety approach came on 13 July 2026.
- Licensing and money. Romania's nuclear regulator, CNCAN, has never licensed a fast reactor, and its dialogue with FALCON has been about codes and standards, not about an application. There is no financing decision for building the reactor: today's EU money covers only the laboratories.
The alternatives: what Romania gets, and when
A fair comparison has two levels. For power in the 2030s, what counts are reactors that can be licensed on known technology. For Romania's place in the next generation, what counts are the other routes into fast reactors.
| Option | What it brings | Announced schedule | Known cost | Main risk now |
|---|---|---|---|---|
| Refurbishment of Cernavodă unit 1 | a longer life for an existing reactor | preparatory works started in 2025 | €1.9bn contract excluding VAT | the unit is off the grid during the works |
| Cernavodă units 3 and 4 (CANDU) | two large reactors on a technology operated since 1996 | 2030 and 2031, when the contract was signed in 2024 | EPCM contract of about €3.2bn, project of about €7bn | no final investment decision yet; Danube water |
| Doicești, NuScale (6 modules of 77 MW) | a 462 MW modular plant | stalled in October 2026 | $6.5bn estimate in December 2025 | more than 80% of the investment-decision conditions unmet |
| Another water-cooled small reactor (e.g. BWRX-300) | a technology with a US construction permit (September 2026) | commercial operation targeted in the US for 2033 | unknown for Romania | a new project from scratch, with no site chosen |
| ALFRED at Mioveni | data and skills in fast reactors; 125 MWe only at the final stage | after LEANDREA (2034), no date | about €1bn (old estimate), 20% pledged by Romania | materials, fuel, no financing for the reactor |
Outside Romania, the lead-cooled race has three markers. Russia is building the BREST-OD-300 reactor at Seversk: at the start of 2026 its body was about 70% assembled, with completion scheduled for the end of the year. Rosatom had earlier announced physical start-up in 2026. In France, newcleo wants a 30-electrical-megawatt reactor, LFR-AS-30, near Chinon, commissioned in 2031. Its public debate opened on 2 April 2026. In Belgium, LEANDREA is being prepared, with RATEN as a partner.
The European Commission's small-reactor strategy of 10 March 2026 wants the first units running in the early 2030s but puts little money on the table: a temporary €200 million top-up through InvestEU. For a demonstrator like ALFRED, that money does not change the arithmetic.
What the Cernavodă drought says
The Cernavodă plant has been fully shut since 13 August. The Danube's flow where it enters Romania was about 1,300 cubic metres per second in the last week of October, against a long-term October average of 3,900, according to the forecast of the National Institute of Hydrology. The government is preparing to extend the state of alert in the energy sector, and a restart is now being discussed for December. The inter-ministerial committee on cooling water recommended that the design of units 3 and 4 take this year's situation into account.
For ALFRED the lesson is indirect. A lead-cooled reactor still needs a cold source, but it turns heat into power more efficiently: 125 electrical megawatts out of 300 thermal, about 42%, against roughly a third for today's water-cooled reactors. That means less heat to reject per megawatt-hour. It does not mean ALFRED solves Cernavodă's problem: it comes too late and is too small. It means water has to become an explicit criterion for every new reactor, including small-reactor sites.
Links 1, 4 and 5 are documented by RATEN, SCK CEN and the consortium's announcements. Link 2 is documented in Ansaldo Nucleare's design presentations of 2018 and 2026, and link 3 in the 2026 stage table. Link 6 is our finding: no public source gives a date or a source of financing for the reactor. Link 7 is documented in the press and in Nuclearelectrica's documents, but its connection to ALFRED is one of timing only: both leave the 2030s gap uncovered.
Three scenarios for 31 December 2030
The variable is the licensing status of the ALFRED reactor (not of the laboratories) at the end of 2030. The three scenarios cover every case and exclude one another.
- A · Construction licence issued (5%). CNCAN licenses construction of the ALFRED reactor. This would require financing to be decided in 2027 or 2028 and the Mioveni data to arrive on time. The EAGLES sequence, with LEANDREA in 2034, makes the scenario highly unlikely.
- B · Application filed, no licence (25%). RATEN or a project company files with CNCAN an application for a siting or construction licence, but no construction licence is issued by the deadline. It would show that Romania wants to keep ALFRED close behind LEANDREA.
- C · No application (70%). The laboratories run at Mioveni, while the reactor stays on the drawing board or waits for the results from Mol. This is the scenario in which Romania has bought data and skills, but not yet a reactor.
What it means for Romania
The money for the laboratories is well spent if it is judged by what it produces: published data, facilities used by other designers, testing contracts and Romanian specialists who stay in the country. It is badly spent if it serves as an energy promise. Energy planning should not count on ALFRED before 2040, and a decision to build the reactor should be tied to three verifiable gates: LEANDREA's results, qualification of materials for stages 2 and 3, and a contracted source of MOX fuel.
For the 2030s the order of priorities is different: the refurbishment of unit 1, the decision on units 3 and 4, with cooling designed for a poorer Danube, and a clear decision at Doicești. The Niculescu cabinet's government programme no longer mentions the small reactors at Doicești and promises, for the first half of 2027, a portfolio of joint energy projects with the United States, “including nuclear”, with no new budget allocation.
Forecast table
| Forecast | Horizon | Probability | How it is checked | Verification source |
|---|---|---|---|---|
| Nuclearelectrica or RoPower announces the termination or suspension of the NuScale-based Doicești project | by 31.03.2027 | 50–60% | A current report or shareholder resolution that terminates or suspends the agreements with NuScale or the project in its NuScale form | nuclearelectrica.ro, bvb.ro |
| Nuclearelectrica takes the final investment decision on Cernavodă units 3 and 4 | by 31.12.2027 | 30–40% | The shareholder resolution approving the final investment decision | nuclearelectrica.ro, bvb.ro |
| Russia's BREST-OD-300 reactor reaches first criticality (physical start-up) | by 31.12.2027 | 35–45% | Announcement by Rosatom or the operator of first criticality | rosatom.ru, world-nuclear-news.org |
| The HELENA-2 and ELF facilities at Mioveni are commissioned | by 31.12.2028 | 35–45% | Announcement by RATEN or SRS of the commissioning of both facilities | raten.ro, world-nuclear-news.org |
| Financing for building the ALFRED reactor (not the laboratories) is signed | by 31.12.2027 | 5–15% | A financing contract, government decision or budget law with a sum dedicated to building the reactor | monitoruloficial.ro, raten.ro |
| Scenario A: CNCAN issues the construction licence for ALFRED | 31.12.2030 | 0–10% | The licensing status of the ALFRED reactor | cncan.ro, raten.ro |
| Scenario B: licence application filed with CNCAN, no construction licence | 31.12.2030 | 20–30% | Same status | cncan.ro, raten.ro |
| Scenario C: no licence application for the ALFRED reactor | 31.12.2030 | 65–75% | Same status | cncan.ro, raten.ro |
When we would consider the analysis wrong
We set the thresholds in advance, so as not to interpret them after the fact.
- Our caution is wrong if, by 31 December 2027, the Romanian state or an investor signs financing for building ALFRED and RATEN files a siting licence application with CNCAN.
- The thesis on the laboratories' value is wrong if HELENA-2 and ELF are not commissioned by the end of 2029 or if, once running, they publish no results used by LEANDREA or other designers.
- The technical reading must be revised if BREST-OD-300 runs at rated power by the end of 2028 with no reported corrosion problems, because the technology gap the consortium cites would then be smaller than it says.
Main sources: RATEN's events timeline, Ansaldo Nucleare's presentation of the ALFRED design (IAEA, 18.02.2026), the ALFRED presentation to the Generation IV Forum (2018), SCK CEN's page on LEANDREA and EAGLES, the SRS contract (World Nuclear News), the 2021 ATHENA contract, the 2013 schedule (IPFM), the lead purchase (Economedia), the HandsOn contract (Economedia), the Doicești conditions (Curs de guvernare), the Doicești cost estimate (Agerpres), the EPCM contract for units 3 and 4, the refurbishment of unit 1 (HotNews), the Cernavodă shutdown (Ziarul Financiar), the Danube's flow (Profit.ro, citing INHGA), the restart discussed for December (Euronews Romania), the government programme (Profit.ro), the EU small-reactor strategy, the BWRX-300 construction permit, BREST-OD-300's progress (Nuclear Engineering International), newcleo's public debate (Légifrance).
.png?width=1200)
