|
|
| Fig. 1: Comparison of projected Angra 3 generation tariffs (November 2025 BNDES scenarios) with current Angra 1 & 2 tariffs and recent renewable auction prices in Brazil. [1,6,7] (Image source: F. Teixeira) |
Brazil's Angra 3 nuclear reactor, a 1,405 MWe pressurized water reactor, has been under some form of construction since 1984 and remains approximately 65% complete after a total of R$12 billion in investment. [1] According to the November 2025 BNDES (National Bank for Economic and Social Development) feasibility study, completing the reactor will cost an additional R$23.9 billion (~US$4.1 billion), while abandoning it will cost north of R$21.9–26.0 billion, which makes the financial outcome nearly identical regardless of the decision. [1] Meanwhile, the project burns through approximately R$1 billion per year in carrying costs while it sits idle: R$800 million in debt servicing to BNDES and Caixa Econômica Federal, plus R$200 million in equipment maintenance. [1]
Angra 3 is a Pre-Konvoi pressurized water reactor, designed as a twin of Angra 2, which has operated since 2001 at the same site in Angra dos Reis, Rio de Janeiro. Table 1 summarizes all three units at the Central Nuclear Almirante Álvaro Alberto, using data from the IAEA Reference Data Series and Eletronuclear. [2,3]
|
||||||||||||||||||||||||
| Table 1: Design and operational parameters of the three reactor units at the Central Nuclear Almirante Álvaro Alberto in Angra dos Reis, Brazil. [2,3] |
Angra 2's operational record is the most relevant benchmark for what Angra 3 could deliver. According to IAEA data, Angra 2 has maintained a lifetime load factor of approximately 84% and generated roughly 10 TWh per year in recent years. [2]
For context, nuclear energy provided in total 15.8 TWh of Brazil's 751.3 TWh total electricity generation in 2024, or approximately 2% of the mix. [4] At comparable efficiency levels to its predecessors, Angra 3 would thus nearly double this nuclear output.
The projection of the cost it would take to finalize construction has tripled over the past 17 years. Table 2 presents the escalation drawn from BNDES studies and official communications. [1,5]
|
|||||||||
| Table 2: Escalation of Angra 3's estimated completion cost from 2008 to 2025. [1,5] |
In addition to the R$23.9 billion additional necessary investment, approximately R$12 billion has already been spent, including financing costs. This results in an estimated R$36 billion total cost for Angra 3. [1]
The estimated cost of abandonment ranges from R$21.9 billion to R$26.0 billion according to the November 2025 BNDES update, encompassing loan repayment penalties, contract termination costs, tax incentive returns, site demobilization, and opportunity cost of invested capital. These obligations would need to be settled in the near term, while the R$12 billion already invested would be lost entirely. [1]
The November 2025 BNDES study projected Angra 3’s generation tariff at R$778.86–817.27/MWh, depending on the financing scenario. [1]
To put this in perspective: Angra 1 and Angra 2 currently sell power at approximately R$308.41/MWh. [6] In Brazil’s regulated electricity auctions, solar photovoltaic projects were contracted at an average of R$178.24/MWh in the most recent comparable renewable tender (A-5, 2022), and wind at R$160.36/MWh (A-5, 2021). [7]
Angra 3's projected tariff of R$778–817/MWh is therefore roughly 4.3–4.6 times the auction price of new wind or solar, and about 2.5 times the tariff of the existing Angra 1 and 2 reactors (Fig. 1).
Eletronuclear argues that the comparison to renewables is misleading because nuclear provides firm, dispatchable baseload power while wind and solar are intermittent. This is a legitimate technical distinction, but the price gap is large.
Had Angra 3 been completed on a schedule comparable to its German Pre-Konvoi siblings, which took 7–10 years from construction start to commercial operation, the reactor could have entered service around 1992–1994. Using Angra 2’s 84% lifetime load factor as a reference:
| Annual generation at 84% capacity factor: | 1,340 MWe × 8,760 hours × 0.84 = 9,857 GWh/year | |
| Cumulative generation (1993 - 2025, 33 years): | 9.86 TWh y-1× 33 y = 325 TWh |
For perspective, 325 TWh is equivalent to approximately 22 years of Brazil’s current total nuclear output (Angra 1 + 2 produced 15.8 TWh in 2024). [4] In other units: 325 TWh × 3.6 × 1015 J/TWh = 1.17 × 1018 J, or 1.17 exajoules of electricity that was never generated.
This is, of course, a counterfactual, as the German construction timelines reflected different institutional and economic conditions. But the calculation illustrates the scale of opportunity cost from four decades of delay.
The project has been suspended three times.
1984 - 1986: Site work began and approximately 70% of heavy equipment was purchased for roughly US$327 million. Brazil’s debt crisis and hyperinflation halted all work. Equipment entered long-term storage.
2010 - 2015: The Lula government revived the project. CNEN issued the construction license on May 31, 2010, and first concrete was poured on June 1, 2010, with a target of late 2015. Construction advanced to approximately 65% before being frozen by Operation Lava Jato (Car Wash). The 16th phase of the investigation, known as Operation Radioactivity, directly targeted corruption in Angra 3 contracts in July 2015. The president of Eletronuclear, Admiral Othon Pinheiro da Silva, was convicted in 2016 of receiving R$4.5 million in kickbacks from construction firms. [8] All electromechanical contracts were annulled in January 2017.
2022 - 2023: Construction resumed in November 2022 under a new consortium. Five months later, in April 2023, the municipality of Angra dos Reis halted work over unpaid socio-environmental compensation. The embargo was lifted by court order in June 2024, but Eletronuclear simultaneously terminated the construction contractor.
As of February 2026, no EPC (engineering, procurement, and construction) contract has been awarded for completion.
The TCU (Federal Audit Court), through its Fiscobras 2025 audit program, identified what it called “relevant failures” in the reference budget for Angra 3’s planned EPC tender. The audit found that cost tables from 2008 and 2013 were being used with only inflation adjustments, that BDI (overhead and profit) rates exceeded market norms, and that an unjustified 5% linear markup had been applied to the reference price—together representing potential overcharges of R$1.35 billion. [9] The TCU congressional presentation further noted that the Comissão Mista de Orçamento had formally warned the President that continuation costs were approaching abandonment costs. [10]
Three numbers define Angra 3’s situation: R$23.9 billion to complete, R$21.9–26.0 billion to abandon, and R$1 billion per year for every year of continued indecision. The most recent BNDES study finds completion marginally cheaper than cancellation, but Angra 3’s projected electricity tariff of R$778–817/MWh is roughly four times the price of new wind and solar capacity in Brazil. [1]
The reactor, if completed by 2033, would enter service nearly 50 years after its equipment was purchased. Whether that constitutes a viable investment depends on the value one assigns to firm, dispatchable, carbon-free baseload generation in a grid that derives 88% of its electricity from renewables but remains vulnerable to drought-induced hydroelectric shortfalls. [4] The numbers alone do not resolve that question, but they define the boundaries within which any honest answer must fall.
© Felipe Leite Teixeira. The author warrants that the work is the author’s own and that Stanford University provided no input other than typesetting and referencing guidelines. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
[1] "Estudo do BNDES Aumenta Expectativa Sobre Decisão Favorável à Angra 3," Eletronuclear, November 2025. [BNDES Study Raises Expectations for Favorable Decision on Angra 3].
[2] "Nuclear Power Reactors in the World," International Atomic Energy Agency, IAEA-RDS-2/45, 2025.
[3] "Projeto Angra 3," Comissão de Minas e Energia, May 2025 [Angra 3 Project].
[4] "Balanço Energético Nacional 2025, Relatório Síntese, Ano Base 2024," Empresa de Pesquisa Energética, Ministério de Minas e Energia, Brazil, May 2025. [National Energy Balance 2025, Synthesis Report, Base Year 2024].
[5] "Sumário Fiscobras 2017, Usina Termonuclear de Angra III," Tribunal de Contas da União, August 2017. [Summary: FISCOBRAS 2017, Angra 3 Nuclear Power Plant, Federal Audit Court of Brazil].
[6] "Resolução Homologatória No. 3.432, de 10 de dezembro de 2024," Agência Nacional de Energia Elétrica (ANEEL), Estabelece a Receita Fixa e a Tarifa das Centrais Geradoras Angra 1 e 2 para 2025. Diário Oficial da União, 12 dez. 2024, Seção 1. Processo No. 48500.003427/2024-31. [Homologatory Resolution No. 3,432: Establishes Fixed Revenue and Tariff for Angra 1 and 2 Nuclear Power Plants for 2025].
[7] "InfoLeilão nº 035: 37º Leilão de Energia Nova (A-5),," Câmara de Comercialização de Energia Elétrica, Brazil, April 2023. [InfoAuction No. 035: 37th New Energy Auction (A-5)].
[8] "Sentença: Ação Penal No. 0510926-86.2015.4.02.5101, Justiça Federal, Seção Judiciária do Rio de Janiero, 7ª Vara Federal Criminal, 3 Aug 16 [Federal Criminal Court Sentence, Operation Radioactivity. Conviction upheld on appeal: TRF2, 1st Specialized Panel, Mar. 2022; sentence reduced to 4 years and 10 months].
[9] "Relatório de Auditoria, Fiscobras 2025, Usina Termonuclear de Angra 3," Tribunal de Contas da União, January 2026. [Audit Report: FISCOBRAS 2025, Angra 3 Nuclear Power Plant].
[10] "Continuidade da obra de Angra 3," Tribunal de Contas da União, May 2025. [Continuation of the Angra 3 Construction].