Alupar Investimento S.A. (ALUP4.SA / ALUP11.SA) - Deep Dive Research Report

Prepared 11 September 2026 | B3 Nível 2 | Tickers: ALUP3 (ON), ALUP4 (PN), ALUP11 (Unit = 1 ON + 2 PN)

A note on the reporting period used in this report

Alupar has a 31 December fiscal year-end and reports quarterly (ITR filings with the CVM). Working forward from that cadence, the most recent release due before today was Q2 2026 (2T26), expected around mid-August 2026. That release exists and has been verified: the earnings release went out after market close on 6 August 2026, and the conference call was held 7 August 2026 at 11:00 Brasília time, confirmed both by Alupar's own Results Center and by the teleconference notice filed with the results package. The six calls used throughout this report are:

The most recent call is 35 days old. All six are used in Sections 7 and 9.

1. What the company does

Alupar builds and owns the physical wires that carry electricity across Brazil and, increasingly, across the Andes. It is not a utility in the sense most people picture. It does not sell you electricity, it does not read your meter, and it does not own the power station at one end or the substation at the other. It owns the 500,000-volt steel-and-aluminium corridor in between, and it gets paid a fixed annual fee for keeping that corridor available.

That distinction is the whole business. Under Brazil's regulatory framework, a transmission concessionaire earns a Receita Anual Permitida (RAP), an Annual Permitted Revenue set at the moment the concession is auctioned and then indexed to inflation each July. The RAP does not vary with how many megawatt-hours flow down the line. It varies only with whether the line is available. If a hydro drought halves the power moving from Pará to São Paulo, Alupar's revenue on that corridor does not move. If a new data centre triples the load, it does not move either. The concessionaire is paid to hold an asset ready, and is penalised through an availability deduction when the asset is not ready. As one industry description of the mechanism puts it, the owner "gets paid a fixed amount of money every year regardless of how much electricity actually flows through the line," which eliminates volume risk and produces a bond-like asset.

The revenue itself is collected by the system operator. Grid users - generators plugging in at one end, distributors and large industrial consumers drawing at the other - sign System Use Agreements with the ONS (Operador Nacional do Sistema Elétrico) and pay a transmission tariff, the TUST, calculated monthly on a nodal basis. ANEEL, the regulator, pools those payments and allocates them to concessionaires according to their RAP. Alupar therefore has no customer credit exposure to any individual counterparty in the ordinary sense. Its counterparty is a regulated clearing mechanism into which the entire Brazilian power system pays.

Alongside this, Alupar owns a smaller portfolio of renewable generation - four large hydro plants, four small hydro plants, wind complexes in Ceará and Rio Grande do Norte, and one solar plant - plus a recently launched energy trading arm selling into Brazil's free market. That side of the house behaves nothing like the transmission side. It carries price risk, volume risk and, since 2024, a severe and growing grid-curtailment problem. The two businesses sit inside one holding company for reasons that are historical rather than industrial, and understanding why is most of what you need to know about Alupar.

How it got here

The corporate entity dates from 2007, but the operating business is older. Per Alupar's own history page, the group began operating in transmission in 2000 and entered hydroelectric generation in 2005. In the second half of 2007 a corporate reorganisation folded twelve transmission companies and four generation special-purpose vehicles into a single holding, Alupar Investimento S.A. The twelve included names still on the books today - EATE, ECTE, ENTE, ERTE, ETEP, ETES, Lumitrans, STC, STN, Transirapé, Transleste and Transudeste - and the four generation SPEs were Foz do Rio Claro, Ijuí, Queluz and Lavrinhas.

The company listed on B3 in 2013, raising approximately BRL 740 million, and has traded since as three lines: ordinary shares (ALUP3), preferred shares (ALUP4) and a Unit (ALUP11) bundling one ordinary with two preferred. Control has never changed hands. Guarupart Participações Ltda., the Godoy Pereira family holding company, holds roughly 52% of total capital, with the balance in the public float.

The founding-family character of the company is unusually visible in the current structure, and it is worth stating precisely because it is both a governance fact and an explanation of the company's behaviour. Per Alupar's management disclosure page: Paulo Roberto de Godoy Pereira, a business administration graduate of Mackenzie University, has been Chief Executive Officer and a board member since 2006, and was president of ABDIB, the Brazilian infrastructure industry association, from 2004 to 2014. José Luiz de Godoy Pereira, a civil engineering graduate of Mackenzie University, has served as Executive Vice President and Chief Financial and Administrative Officer of Alupar since 11 July 2007, and chairs the Board of Directors. Both were re-elected to the board on 16 April 2025 and to executive office on 8 May 2025. Two further family members sit as alternate directors: Ana Helena Godoy Pereira de Almeida Pires, a Pace University/Lubin graduate who worked in Citibank Private Banking and has been an NGP director since 2004, and Fernando Carvalho de Godoy Pereira, a 2017 FGV EAESP business administration graduate with five years in investor relations and finance.

Two non-family executives run the operating and market-facing functions. Ênio Luigi Nucci, Commercial and Technical Officer, has been with Alupar since 2006 and the filing credits him with more than twenty-five years in infrastructure and energy. Luiz Eduardo Muniz Coimbra, Investor Relations Officer, is a PUC-SP economist with a corporate finance postgraduate qualification from FIA who joined Alupar as a trainee in 2009 and now fronts every quarterly call. Two independent directors complete the board: Humberto Falcão Martins, who holds a Ph.D. in Administration and a Master's in Public Administration from EBAPE/FGV, and Rogério Paulo Calderón Peres, a PwC partner from 1981 to 2003 who subsequently held director roles at Itaú Unibanco and HSBC.

The core value proposition, and what is actually hard about it

Because RAP is fixed at auction, a transmission concession is economically a construction problem wearing a utility's clothes. ANEEL publishes a reference capex for each lot and a ceiling RAP. Bidders compete by cutting the RAP. Whoever accepts the lowest annual revenue wins, and then has four to six years to build the thing.

The entire return therefore comes from three places, none of which is "the tariff":

1. Building for less than the regulator's reference capex. ANEEL's number is a planning estimate. A company with its own engineering organisation, standing supplier relationships for towers, conductor and transformers, and a land-acquisition team that has negotiated thousands of rights-of-way can beat it. Alupar's management said explicitly on the Q2 2026 call that on the Lot 7 project it expects roughly 30% savings against the R$1,089 million reference capex.
2. Financing cheaply and long. Brazilian infrastructure debentures are tax-advantaged for domestic investors and can be issued at very long tenors against a contracted, inflation-linked revenue stream. In early 2026 the TECP subsidiary issued R$2.45 billion of 12-year debentures at IPCA plus 6.9928%, a spread management described on the Q1 2026 call as NTN-B 2040 minus 0.40%.
3. Finishing early. RAP begins when the asset is energised and accepted, not when the contract says it should be. Every month pulled forward is a free month of a 30-year annuity. Alupar energised Phase I of the TECP Centro substation in November 2025, eleven months ahead of its June 2026 deadline, stepping its share of that project's full RAP from 11.84% to 33.08%.

The hard part is not the electrical engineering. High-voltage transmission is mature technology. The hard part is doing the following, repeatedly, across seven-year timelines and thousands of kilometres of rural Brazil: acquire the land corridor from hundreds of individual owners; obtain a federal environmental licence from IBAMA covering that corridor; procure transformers and conductor in a market where lead times move with global demand; and mobilise line-construction crews into the Amazon, the cerrado and the Andes. A company that is good at this earns an equity return on a contracted annuity. A company that is bad at it discovers that the RAP it bid is fixed while its costs are not.

"We are a company of greenfield, our capital cost is a little bit higher than other companies... greenfield has given better results."

• CFO José Luiz de Godoy Pereira, Q3 2025 conference call, 7 November 2025

That sentence is the strategy in one line. Alupar does not buy operating transmission assets from other people at a market-clearing price. It bids for the right to build them, accepts the construction risk, and keeps the construction margin.

A concrete walk-through: Lot 7, São Paulo

On 3 July 2026, ANEEL held the second stage of Transmission Auction 01/2026. Alupar bid through the Consórcio Olympus XX, a partnership with the Brazilian asset manager Perfin, for Lot 7.

Lot 7 is a piece of urban surgery. It comprises a new São Miguel 345/88 kV substation in a (9+1R) x 133.33 MVA configuration, plus four underground 345 kV circuits: Norte-São Miguel circuits 1 and 2 at 8.2 km each, and São Miguel-Ramon Reberte Filho circuits 1 and 2 at 9.2 km each. Roughly 17.4 km of route, all of it buried, all of it inside the São Paulo metropolitan area.

ANEEL's reference capex was approximately R$1.08 billion and the ceiling RAP was set accordingly. The Olympus XX consortium bid a RAP of R$96.7 million, a 52% discount to the ceiling. The concession runs 30 years; the energisation deadline is June 2031. As reported, the winning bid produced a RAP-to-capex ratio of 12.4% against reference capex, which rises further if the 30% construction saving management flagged is actually realised.

Now follow the money. From June 2031 to 2061, an inflation-indexed R$96.7 million per year lands in the consortium's accounts as long as the four circuits and the substation are available. Between now and 2031, the consortium spends roughly a billion reais it does not have, funded with long-dated debentures, while earning nothing on the asset. That five-year hole is the entire risk, and the ability to fund it, build through it, and come out the other side with real capex well below the regulator's number is the entire business.

2. Business segments

Alupar reports two operating segments, Transmission and Generation, plus a small and recently added energy trading activity. They are genuinely different businesses with different risk profiles, different regulators, and different economics, and the difference has widened sharply over the last two years.

Revenue mix on a regulatory basis in Q1 2026 was approximately 74% Transmission, 26% Generation. On a segment-EBITDA basis the skew is more extreme, roughly 81% Transmission, 19% Generation, because transmission runs at a structurally higher margin than any merchant generation business can.

2.1 Transmission

What it does. Alupar holds interests in concessions for 44 electricity transmission systems totalling roughly 9,999 km of lines in operation and under construction, and around 39 substations, spread across Brazil, Colombia, Peru and Chile. As of the Q4 2025 call, 31 of those systems were operational and 13 were under construction with completion dates spread across 2026 to 2029. Brazilian and Peruvian concessions run 30 years; the Colombian and Chilean concessions are indefinite, which materially changes their present value character.

The Brazilian footprint is national rather than regional. Assets include EATE's 924 km 500 kV Tucuruí corridor across Pará and Maranhão, STN's 541 km Teresina-Fortaleza 500 kV line, TPE's 541 km 500 kV Poções-Governador Valadares link across Bahia and Minas Gerais, EBTE's 775 km of 230 kV in Mato Grosso, and TNE's 715 km 500 kV Manaus-Boa Vista line. Smaller assets sit in Santa Catarina (ECTE, ETSE, STC), Espírito Santo (ETES, ETC, ESTE), Minas Gerais (Transleste, Transirapé, Transudeste, ESDE), Rio Grande do Sul (Lumitrans), and São Paulo and Rio de Janeiro (TSM, ELTE).

The core capability. What Alupar knows how to do, and what took twenty-five years to build, is the unglamorous middle of the project: land, licence, logistics. Its quarterly disclosures are dominated not by construction percentages but by land-acquisition percentages - the Q1 2026 pack reported TPC land negotiation at 91%, up from 65% of contracts, and noted that TCN, TEO and TET had reached 100% on substation areas. A transmission project in Brazil does not fail because nobody can string a conductor. It fails because one landowner on a 551 km corridor will not sign, or because IBAMA's unified installation licence sits in a queue for three years.

The TNE project is the case study. Alupar and Eletronorte won ANEEL Auction 004/2011 for the Manaus-Boa Vista line. Commercial operation began on 16 September 2025 - fourteen years later. The line connects the state of Roraima, previously an electrical island, to the National Interconnected System, and carries an estimated RAP of R$561.7 million for the 2025/26 cycle. A fourteen-year gap between auction and energisation is what licensing risk looks like in practice.

Note on TNE ownership: Alupar's own corporate site states the concession is held 51% Alupar / 49% Eletronorte, consistent with the original 2011 consortium. Subsequent reporting has referenced a higher Eletrobras/Axia economic stake following capital injections into the SPE. The two figures cannot both be current, and Alupar has not published a reconciliation in accessible materials. Readers should treat the economic split as unconfirmed.

Why it is separate. It is separate because it is regulated by ANEEL under a concession contract, remunerated through the TUST pool, and carries no merchant risk. Nothing about it resembles selling electricity.

Competitive position. Alupar describes itself in its corporate profile as the largest 100% privately-controlled transmission company in Brazil by RAP. In auctions it competes against Axia Energia (the former Eletrobras), ISA Energia Brasil, Engie Brasil, Taesa, State Grid Brazil Holding and a rotating cast of financial consortia. It wins where its construction cost base and financing cost let it accept a lower RAP than a bidder who must buy engineering in the open market. It loses where a state-backed or infrastructure-fund bidder is prepared to accept a lower cost of capital than Alupar's, which increasingly happens.

Strategic role. This is the group. Management's capital allocation, funding and messaging are all built around it.

2.2 Generation

What it does. Alupar's corporate profile states a portfolio of 16 generation assets totalling 798.44 MW - four large hydro plants, four small hydro plants, seven wind farms and one solar park - across Brazil, Peru and Colombia. The named assets and their nameplate capacities, as published on Alupar's own companies page, are:

The per-asset nameplate figures published on that page sum to more than the 798.44 MW group headline in the corporate profile, which is consistent with the group figure being stated on an ownership-adjusted or otherwise different basis. Alupar does not publish the reconciliation.

The core capability. Honestly, less than in transmission. Running a run-of-river hydro plant is an operations job, not a moat. What the segment does provide is a physical hedge and a commercial platform: Brazilian hydro assets participate in the Mecanismo de Realocação de Energia (MRE), which socialises hydrological risk across participating plants, and the portfolio gives Alupar something to sell in the free market. Management described the commercial approach on the Q3 2025 call as active positioning rather than passive selling:

"We position our portfolio year-over-year... constantly managing based on sales and exposure."

• Q4 2025 conference call, 6 March 2026

Why it is separate. Different regulator relationship, different revenue mechanism, different risk. Transmission revenue is availability-based and inflation-linked. Generation revenue is a function of how much the plant produces, what price it was contracted at, and increasingly whether the ONS lets it produce at all.

Competitive position. Weak, in the sense that there is no moat. Alupar's wind farms compete with everyone else's wind farms into the same oversupplied Brazilian market. This is a price-taker business with a cost position no better than the industry's.

Strategic role. It is no longer the growth bet. Management's Investor Day messaging and all six recent calls point capital at transmission, particularly Latin American transmission. Generation is a cash contributor that has become a source of negative surprises.

2.3 Energy trading (Alup)

Alupar launched Alup, an energy trading arm, into Brazil's free market, focusing initially on medium-voltage clients - the customer tier that became eligible when the free market opened to all consumers with demand above 500 kW from January 2024. It sells the group's own generation and third-party energy. It is small, it is new, and it is best understood as a distribution channel for the generation segment rather than a third business.

Segment comparison

3. Products and business detail

3.1 The transmission product, in detail

A transmission concession is not one product; it is a bundle of physical assets sold as an availability service. Alupar's concessions fall into recognisable families.

Long-haul 500 kV bulk corridors. These move large blocks of power hundreds of kilometres from generation basins to load centres. EATE (924 km, Pará/Maranhão), ENTE (464 km, Pará/Maranhão), ETEP (323 km, Pará), STN (541 km, Piauí/Ceará), TPE (541 km, Bahia/Minas Gerais), TME (348 km, Mato Grosso), TCC (288 km, Minas Gerais/Espírito Santo), ESTE (236 km, Espírito Santo), TSM (330 km, São Paulo/Rio de Janeiro), ETB (446 km, Bahia), TNE (715 km, Amazonas/Roraima). This is the backbone of the portfolio and the highest-RAP-per-kilometre category.

Regional 230 kV and 345 kV systems. Shorter, often serving a specific industrial or urban load. Transleste (150 km, 345 kV, Minas Gerais), Transudeste (140 km, 345 kV), Transirapé (65 km, 230 kV), ETEM (235 km, 230 kV, Mato Grosso), EBTE (775 km, 230 kV, Mato Grosso), ERTE (179 km, 230 kV, Pará), STC (195 km, 230 kV, Santa Catarina), ETES (107 km, 230 kV, Espírito Santo), ELTE (40 km, 230 kV, Baixada Santista, São Paulo), AETE (193 km, 230 kV, Mato Grosso, 50% held).

Substation and reinforcement projects. Increasingly the growth category, and different in character: no line at all, just transformation capacity, bay expansions, and voltage control equipment inside an existing substation. TECP's four-phase modernisation of the Centro substation in São Paulo is the marquee example. ETVG (Várzea Grande, Mato Grosso), ETSE (Santa Catarina), ETC (Espírito Santo), ESDE (Minas Gerais) and ETAP (Rio Grande do Norte) fall in this family. These projects are attractive because they are short, urban, low-licensing-risk and carry high RAP-to-capex ratios - Alupar approved an ETVG reinforcement in Q1 2025 at a 13.7% RAP/capex ratio.

Underground urban transmission. New for Alupar with Lot 7. Burying 345 kV circuits under São Paulo is materially more expensive per kilometre than overhead line but eliminates the right-of-way negotiation that is the dominant delay risk on rural corridors. It also requires cable-jointing and thermal-management competence that overhead construction does not.

International concessions. TCE in Colombia (237 km, 500 kV, Nueva Esperanza near Bogotá to La Virginia near Pereira, with a RAP of approximately US$28.5 million), a Chilean transmission position, and a growing Peruvian portfolio.

3.2 What is under construction right now

As of the Q2 2026 call on 7 August 2026, the construction book looked like this:

The Peruvian Group 3 award deserves separate mention. In September 2025 ProInversión, Peru's investment promotion agency, awarded Alupar Group 3 of the 2023-2032 Transmission Plan: four transmission and substation projects totalling 247.4 km of lines across Palca, La Pascana, Arequipa, Lima, Apurímac and Puno, serving roughly 2.3 million people. Reported investment is US$214-220 million, with an aggregate RAP of US$31.8 million, implying a RAP-to-capex ratio of roughly 14.5% - materially above what Brazilian auctions now clear at. Commercial operation is scheduled for November 2029 under a 30-year concession.

3.3 What entered service in 2025

2025 was the year three long-dated projects landed, and it is the single best evidence in the file about Alupar's execution:

• ELTE (Empresa de Transmissão Baixada Santista), 40 km of 230 kV plus two substations serving the Santos region of São Paulo: commercial operation July 2025. Separately, the Manoel da Nóbrega substation reinforcement was brought forward in May 2025, adding R$14.2 million of annual permitted revenue at 2024/25 cycle values.
• TNE (Transnorte Energia), the 715 km 500 kV Manaus-Boa Vista double-circuit line: commercial operation 16 September 2025, connecting Roraima to the national grid for the first time. Estimated project RAP R$561.7 million for the 2025/26 cycle.
• TCE (Transmisora Colombiana de Energía), 237 km at 500 kV between Nueva Esperanza and La Virginia: commercial operation registered 22 October 2025, Alupar's first operating transmission asset in Colombia. RAP approximately US$28.5 million.

3.4 The revenue indexation structure

This is unusual and worth understanding, because it explains a lot of the reported quarterly volatility. Alupar's contracted RAP for the 2026/27 cycle is indexed on three different bases:

• IPCA (Brazil's consumer price index): 58.9%
• IGP-M (Brazil's general price index, more commodity- and FX-sensitive): 24.1%
• PPI / USD-equivalent (the Latin American concessions): 17.0%

Three indices, three different behaviours. IGP-M can diverge violently from IPCA when the real moves or commodity prices swing. The 17% dollar-linked slice is a natural hedge against a weakening real but introduces translation noise, and Peruvian sol movements produced a negative FX impact that management flagged specifically on the Q1 2026 call.

On the other side of the balance sheet, the debt is indexed differently again: as of Q2 2026, 63.8% IPCA, 20.5% CDI, 13.7% TJLP, 2.0% foreign currency, with debentures accounting for 67.1% of funding sources and development banks a further 12.5%. The IPCA-linked debt is the immediate cause of the profit compression reported through 2026: inflation accretes onto the debt balance and into financial expense continuously, while the RAP that offsets it resets once a year, in July. When inflation rises, the cost hits first and the revenue catches up later.

3.5 Geographies

Brazil is the core, with assets in at least fourteen states from Amapá and Roraima in the far north to Rio Grande do Sul in the south.

Colombia since 2011 (PCH Morro Azul authorisation) and materially since 2016, when Alupar won the UPME 07-2016 tender that became TCE. Colombian transmission concessions are indefinite, not 30-year.

Peru since 2013 (La Virgen hydro authorisation), and now the principal international growth market with TCN, Maravilla and Puno Sur in construction and the Group 3 award running to 2029.

Chile since 2023-2024, with an Atacama-region position adjudicated alongside Transelec and Engie for tension-control services in the Antofagasta and Atacama regions, remunerated over a 25-year useful life beginning in 2027.

Management told its Investor Day that Latin American assets are expected to represent at least 18% of consolidated revenue once the eight projects under development enter operation, and framed the rationale as higher returns (RAP/capex above 11%), dollar-denominated revenue, longer concessions, and lower country risk premia than the marginal Brazilian lot now offers.

4. Customers

Who actually pays

For the transmission segment, the honest answer is: everyone who touches the Brazilian grid, through a pooling mechanism. Alupar does not negotiate with customers. Grid users sign System Use Agreements (CUST) with the ONS. The ONS calculates the TUST monthly on a nodal basis, differentiating between generators and consumers. ANEEL sets the total RAP pool annually - for the 2026/27 cycle it approved R$54.95 billion of permitted revenue across all concessionaires in commercial operation, 9.41% above the prior cycle - and allocates it.

This has three consequences that are easy to miss.

There is no customer concentration in the normal sense. No single offtaker can leave. The "customer" is the aggregate load of Brazil.

There is no sales cycle and no commercial relationship. The buying decision was made once, at the auction, by ANEEL, on the single criterion of lowest RAP. Everything that looks like a commercial function in a normal company - pricing, account management, contract renewal - does not exist here. What exists instead is a bid-preparation function: engineering cost estimation, land and licensing due diligence, and financing structuring, all compressed into the weeks before an auction.

Switching costs are infinite and irrelevant. A 500 kV line between two substations cannot be replaced by a competitor's line. The concession is exclusive for 30 years. But that exclusivity buys nothing beyond the contracted RAP, because the RAP is fixed. Alupar cannot raise prices on a captive customer. The moat protects the revenue; it cannot grow it.

The real credit exposure is systemic, not counterparty-specific. If the Brazilian tariff pool comes under political pressure - and Brazilian electricity tariffs are a recurring political issue - the risk arrives through regulation, not through a customer defaulting.

Generation and trading customers

Here Alupar does have real customers and a real sales motion. The generation portfolio sells into two markets:

The regulated market (ACR), where energy is sold to distribution companies through government-run auctions at long-dated prices. Predictable, low-margin, and the buyer is a regulated utility.

The free market (ACL), where large commercial and industrial consumers contract directly. This is where Alup, the trading arm, operates, targeting medium-voltage customers who became eligible when the threshold dropped to 500 kW of demand in January 2024. The buying decision here sits with a plant manager or a corporate energy buyer, the criteria are price and contract flexibility, and the sales cycle is weeks to months. Switching costs are close to zero; contracts are typically one to five years and customers shop them.

Management has been explicit that free-market pricing has been working against them. On the Q2 2026 call, the generation segment's revenue decline was attributed to lower average prices in the free market and lower contracted volumes in both the free and regulated markets.

Contract structure and revenue predictability

The mix is unusually clean:

• Transmission: ~74% of revenue, 100% contracted, 30 years (indefinite in Colombia and Chile), inflation-indexed, availability-based. Predictability is close to absolute at the revenue line. The variability is entirely below the EBITDA line, in financial expense and tax.
• Generation: ~26% of revenue, split between medium-term PPAs and spot exposure. Predictability is poor and has deteriorated, because curtailment now removes output that was contracted and in some cases not compensated.

5. Competitive landscape

The structure of the industry

Brazilian transmission is a repeated-auction oligopoly. There is no market share to defend, only a pipeline of new lots to bid on. Existing concessions are untouchable for their term; future concessions go to whoever accepts the lowest RAP. Market share is therefore a running tally of auctions won, and it resets every few months.

The competitive intensity is measurable and it is rising. In Transmission Auction 01/2026 (first stage, March 2026), five lots representing R$3.3 billion of investment cleared at an average discount of 50.68%, described in Brazilian trade press as a record. Individual lots: Lot 1 (MG/RJ/SP) at 46.85%, Lot 2 (PR/SC) at 46.89%, Lot 5 (MT/PA) at 50.89%, sublot D at 54.77%. Winners were Cimy (lots 1 and 6), Engie (lots 2 and 3) and Consórcio BR2ET (lot 4). Alupar won nothing in that round.

In the second stage (3 July 2026), four lots worth R$1.75 billion were auctioned. Alupar's Olympus XX consortium took Lot 7, the largest, concentrating 65.6% of the round's contracted RAP and 62.0% of its regulatory investment. Axia Energia Sul took Lots 8, 9 and 10, bidding as aggressively as 59% discount on Lot 8.

Discounts above 50% mean the RAP alone no longer carries an equity return. As one Brazilian sell-side note on that auction put it, return theses now depend almost entirely on a company's ability to deliver real capex below the regulatory capex, secure low-cost debt, run tax-efficiently, and ideally energise early. That is a description of a business where operational execution has replaced bidding as the source of return, and it is exactly the ground Alupar has chosen to fight on.

Named competitors

Where Alupar wins

Construction cost. The company has built continuously since 2000 and has the engineering, procurement and land teams standing. When it says it expects 30% savings against ANEEL's reference capex on Lot 7, that is the accumulated benefit of twenty-five years of doing this.

Early energisation. TECP Phase 1 landed eleven months ahead of deadline. The Manoel da Nóbrega reinforcement came forward. In a world of 52% discounts, a year of RAP pulled forward on a 30-year asset is a meaningful piece of the return.

Lot selection discipline. Alupar sat out the March 2026 round entirely and then took the single largest lot in the July round. Management's framing on the Q4 2025 call was blunt:

"Each auction is an auction, depends on the number of aggressive players."

• Q4 2025 conference call, 6 March 2026

International arbitrage. With Brazilian lots clearing above 50% discount, Alupar has redirected capital to Peru at roughly 14.5% RAP/capex and to Colombia and Chile where concessions are indefinite. Very few Brazilian mid-caps have the operating footprint to do this.

Where Alupar is exposed

Cost of capital. Against Axia (state legacy, now privatised at enormous scale), State Grid (Chinese sovereign), Celeo (Dutch pension money) and Engie (European utility parent), Alupar's cost of capital is structurally higher. Management effectively concedes this: the greenfield quote from Q3 2025 begins "our capital cost is a little bit higher than other companies." In an auction where the winner is whoever accepts the lowest return, a higher cost of capital is a permanent handicap that must be overcome by construction efficiency.

Scale. Alupar is competing for the same lots as companies several times its size, with construction books to fund, and it must maintain leverage discipline while they need not.

Generation. In generation there is no moat at all. Alupar's wind farms are exposed to exactly the same curtailment as everyone else's, and in June 2026 Alupar was reported as among the worst-affected portfolios in Brazil.

Barriers to entry

Moderate, and lower than a casual reading of "30-year concession" suggests. There is no licence scarcity: ANEEL runs auctions open to any qualified bidder, and financial sponsors with no operating history win lots regularly (Cimy, BR2ET, Perfin). The genuine barriers are:

1. Capital at scale and long tenor. Funding a billion-real project with five years of negative cash flow requires access to the Brazilian infrastructure debenture market at institutional size.
2. Construction organisation. A financial sponsor must buy EPC in the open market at market price, which structurally caps how deep a discount it can bid. This is Alupar's real edge.
3. Land and licensing competence. Fourteen years from auction to energisation on TNE is the barrier made visible.
4. Operating track record as a qualification criterion. ANEEL and ProInversión pre-qualify bidders.

None of these stops a well-capitalised new entrant. All of them make it expensive.

Structural shifts under way

• Discount escalation. Fifty-plus percent discounts are the new baseline, compressing returns for everyone and shifting the winner from "who will bid lowest" to "who can build cheapest."
• From lines to substations and storage. The upcoming auctions are increasingly weighted to transformation capacity and voltage control rather than greenfield corridors. ANEEL's October 2026 auction carries 13,564 MVA of transformation against 1,866 km of line. Brazil has also opened a battery-storage regulatory framework, a category Alupar has said it is watching.
• Curtailment reshaping the generation market. Investment in Brazilian wind and solar has stalled on curtailment, which paradoxically strengthens the case for transmission investment - the cuts exist because the grid cannot move the power.

6. Industry

What drives demand

Demand for new transmission in Brazil is driven by a geographical mismatch that is getting worse. Brazil's cheapest new generation is wind and solar in the Northeast and hydro in the North. Its load is in the Southeast. Every gigawatt of renewable capacity added in Bahia, Rio Grande do Norte, Piauí or Ceará requires wires to move it 1,500 to 2,500 km south, and the wires have lagged the turbines badly.

The evidence for the lag is the curtailment data. In 2025, 20.6% of total Brazilian solar and wind output was curtailed, more than double the 9.3% cut in 2024. Through 10 May 2026, curtailment averaged 3 GW, or 17.8% of potential generation from those sources. The ONS expects this to persist: 3.11 GW average in 2026, rising to 3.5 GW in 2027 and 3.65 GW by 2029. Curtailment-related losses have been estimated at roughly US$300 million in 2024 and US$370 million in 2025.

That is a demand signal for transmission written in the plainest possible terms. Every curtailed megawatt-hour is evidence that the grid needs building.

Secondary drivers: electrification of transport and industry, data centre load growth, the integration of previously isolated systems (Roraima joined the national grid only in September 2025), and reliability requirements after system disturbances.

Industry size and growth

The regulated revenue pool is the cleanest measure. For the 2026/27 tariff cycle (1 July 2026 to 30 June 2027), ANEEL set total permitted revenue for transmission facilities in commercial operation at R$54.95 billion, up 9.41% on the prior cycle.

The forward auction pipeline as disclosed:

Brazilian sell-side has framed the transmission auction pipeline as capable of attracting on the order of R$66.5 billion of investment through 2027.

Where Alupar sits in the chain

Squarely in the middle. Upstream sit the equipment suppliers (transformers, conductors, towers, switchgear, cable) and the EPC contractors. Downstream sit the ONS, which dispatches and bills, and ANEEL, which sets and allocates the revenue. Alupar is the asset owner and, unusually for a mid-cap, is substantially its own project developer and construction manager rather than a pure asset holder outsourcing to an EPC.

Import dynamics

There is no import-substitution story here in the conventional sense; power cannot be meaningfully imported into Brazil at scale. The import dynamic that matters is capital and equipment. Large power transformers, HVDC and specialised switchgear are globally sourced, and lead times and prices move with global grid investment. A transmission developer that locked a RAP in 2026 and buys transformers in 2029 carries that exposure unhedged. The competitive dynamic is also import-shaped in the bidder pool: State Grid of China, ISA of Colombia and European utilities all compete directly for Brazilian lots.

Regulation

This is a regulation-defined industry and three items dominate the current agenda.

The annual tariff revision. ANEEL recalculates and publishes RAP each June for the cycle beginning 1 July. This is when inflation indexation actually lands in revenue.

Tax reform. Brazil's consumption tax overhaul began its transition in 2026 and completes in 2033. PIS/Cofins is extinguished in 2027, replaced by CBS; ICMS and ISS phase out between 2029 and 2033. For transmission specifically, ANEEL approved removing PIS/Pasep and Cofins amounts from the RAP calculation base from 1 July 2026 for concessions contracted through 31 December 2005. Estimates suggest concessionaire revenue moves from a 3.65% PIS/Cofins rate to roughly 11-12% under CBS between 2027 and 2028, roughly a tripling of the headline rate on tariff revenue. The governing principle is that reform should be neutral for concessionaires, producing neither gain nor loss, with the burden grossed up in the tariff. Alupar's IR officer addressed this directly on the Q2 2026 call:

"Most of our contracts, the revenue of RAP is net, grossing up the taxes."

• Luiz Coimbra, Q2 2026 conference call, 7 August 2026

Neutrality is the intent. Execution of neutrality across dozens of contracts of different vintages is where the risk lives.

Federal tax incentive reduction. LC 224/2025 imposes a linear 10% cut to federal tax incentives and benefits, phased as 5% in 2026 and a further 5% in 2027. Separately, SUDAM and SUDENE regional incentives (a 75% IRPJ reduction for qualifying projects in the North and Northeast, for a ten-year enjoyment period) expire asset by asset. Alupar's Q2 2026 profit decline was attributed partly to increased income tax expense following the expiration of tax benefits, which is this mechanism arriving.

Curtailment compensation. In July 2026 Brazil set compensation rules for curtailed wind and solar covering eligible cuts from 2023 to 2025. Critically, only external unavailability (transmission delays or limitations) and reliability-driven cuts ordered by the ONS qualify. Curtailment caused by energy oversupply is excluded. Since oversupply is the dominant and growing cause, the framework leaves most of the loss with generators.

Cyclicality

Transmission is close to acyclical at the revenue line. RAP does not respond to GDP, to power demand, or to commodity prices. What is cyclical is the auction pipeline - it responds to the national expansion plan and to government investment appetite - and the cost of capital, which responds to the Selic rate and to the country risk premium. Brazilian transmission companies have historically been sensitive to real interest rates in exactly the way long-duration bond proxies are.

Generation is fully cyclical and additionally weather-dependent: hydrology, wind resource, spot prices (PLD), and free-market contracting all swing.

Tailwinds and headwinds

Tailwinds: a R$54.95 billion and growing regulated revenue pool; a visible multi-year auction pipeline of R$8.9 billion (Oct 2026) and R$12.9 billion (2027); curtailment creating an undeniable case for more grid; electrification and data centre load; Latin American markets offering better returns than Brazil; battery storage emerging as a new adjacent auction category.

Headwinds: record auction discounts compressing sector returns; the consumption tax transition creating multi-year uncertainty; federal tax incentives being cut; environmental licensing timelines that can stretch a project by a decade; and a high domestic real interest rate that raises the funding cost of long-duration assets while making their contracted cash flows less attractive relative to government paper.

7. Growth triggers

All items below are drawn from the six conference calls listed at the top of this report, or from company announcements discussed on those calls.

• Lot 7 São Paulo underground project won, targeting June 2031 energisation. New 345/88 kV São Miguel substation plus four underground 345 kV circuits totalling ~17.4 km, won through the Olympus XX consortium with Perfin at a 52% discount. Reference capex R$1,089 million, RAP R$96.7 million, RAP-to-capex 12.4% against reference figures. (Q2 2026 concall, 7 August 2026)

  Management stated it expects approximately 30% savings against the reference capex, which would lift the realised RAP-to-capex ratio materially above the 12.4% headline.

• TAP received its full IBAMA installation licence in July 2026, releasing construction on the largest project in the book. Two 500 kV lines (Silvânia-Nova Ponte 3 and Nova Ponte 3-Ribeirão Preto) totalling 551 km across Goiás, Minas Gerais and São Paulo, plus expansions at three substations. Estimated capex R$2,051.8 million; RAP of R$276.8 million for the 2026/27 cycle; RAP-to-capex 11.67%. (Q2 2026 concall, 7 August 2026. Repeated trigger - TAP progress was reported at 28% physical advance on the Q4 2025 concall, 6 March 2026.)

  "It's the biggest project we have..."

  • Luiz Coimbra, Q2 2026 concall, 7 August 2026

• TCN in Peru reached 79% physical completion and is expected operational during 2026. Progress jumped from 28% at Q1 2026 to 79% at Q2 2026, one of the fastest advances in the book. (Q2 2026 concall, 7 August 2026; prior status Q1 2026 concall, 8 May 2026.)

• TECP Centro substation Phase 2 at 90% complete, with additional RAP on completion. Phase 1 was energised in November 2025, eleven months ahead of its June 2026 contractual deadline, stepping Alupar's share of the project's integral RAP from 11.84% to 33.08%. (Q2 2026 concall, 7 August 2026; Q3 2025 concall, 7 November 2025, where management first flagged the 11-month advance.)

• Peruvian Maravilla and Puno Sur projects received environmental licences in May and June 2026 with 100% of substation land secured, positioning them for accelerated execution. Part of the Group 3 award from ProInversión: four projects, 247.4 km, US$214-220 million investment, aggregate RAP of US$31.8 million (roughly 14.5% RAP-to-capex), commercial operation scheduled November 2029. (Q2 2026 concall, 7 August 2026; project introduced on the Q3 2025 concall, 7 November 2025, as "Peru 250 km line, five substations, US$220 million, physical advance 23.8%, commercial operation targeted November 2029.")

• Chilean Atacama tension-control service scheduled to begin during 2027, remunerated as an annuity over a 25-year defined useful life. (Referenced across the Q2 2025 concall, 8 August 2025, where management described a block of eight international projects - three in Chile, four in Peru, one in Colombia.)

• Brazilian implementation-cycle funding is complete. The TECP subsidiary raised R$2.45 billion of 12-year infrastructure debentures at IPCA plus 6.9928% in early 2026. (Q1 2026 concall, 8 May 2026; announced at Q4 2025 concall, 6 March 2026.)

  "Essa emissão resolve praticamente todo o funding do nosso ciclo de implantação no Brasil." ("This issuance essentially resolves all the funding of our implementation cycle in Brazil.")

  • Luiz Coimbra, Q4 2025 concall, 6 March 2026

• Thirteen transmission systems under construction with completion staged from 2026 to 2029, out of 44 concessions of which 31 are operational. Management sized the aggregate programme at "more or less $2 billion, roughly R$10 billion total." (Q4 2025 concall, 6 March 2026.)

• Deleveraging begins once the construction book converts. Management guided net debt to EBITDA peaking at 3.9x-4.0x in 2028 before declining as projects energise. (Q4 2025 concall, 6 March 2026.)

  "When you have a very high backlog, your leverage level will increase during implementation process. Then it goes down, and it goes down very fast."

  • CFO José Luiz de Godoy Pereira, Q3 2025 concall, 7 November 2025

• Battery storage auctions are under active evaluation as an adjacent category. Management confirmed interest on the Q1 2026 call because of the "synergy with our strategy," but has held back pending regulatory clarity, and turned more cautious by Q2 2026, citing a "high level of supply and registered amounts" creating uncertainty. (Q1 2026 concall, 8 May 2026; Q3 2025 concall, 7 November 2025; Q2 2026 concall, 7 August 2026. Repeated across three calls with the tone moving from interested to disciplined.)

• Latin American assets targeted at 18% or more of consolidated revenue once the development pipeline operates, with management citing higher RAP-to-capex ratios, dollar-denominated revenue, longer or indefinite concessions, and lower country risk premia as the rationale. (Alupar Investor Day, as reported by Genial Investimentos.)

• ANEEL Transmission Auction 04/2026 scheduled for 30 October 2026, offering nine lots (up to ten concession contracts via sublots 4A and 4B), R$8.9 billion of estimated investment, 1,866 km of new lines and sectioning, and 13,564 MVA of transformation capacity across Bahia, Goiás, Mato Grosso do Sul, Paraíba, Paraná, Rondônia and São Paulo. Management has consistently declined to pre-commit to participation, evaluating lot by lot. (Auction confirmed by ANEEL; participation stance stated on the Q4 2025 concall, 6 March 2026.)

Trigger summary

8. Key risks

1. Auction discounts have removed the margin of safety on new growth

Mechanism. RAP is fixed irrevocably at the auction. When average discounts hit 50.68% (Auction 01/2026 first stage) and individual bids reach 59% (Axia on Lot 8, July 2026), the contracted revenue no longer covers a normal equity return on the regulator's reference capex. The winner is betting it can build for 25-30% less than the reference. If it cannot - because a transformer price moves, because a corridor negotiation drags, because an EPC contractor fails - the shortfall is absorbed entirely by equity, for thirty years, with no mechanism to repair it.

Calibration. High probability of margin compression across the industry; moderate probability that any given Alupar project misses badly. This is the central risk in the business model as currently practised.

Management's own framing is that they only bid where the maths works, which is credible given they sat out the March 2026 round entirely. But the counterpart to discipline is that if discounts stay at this level, Alupar's growth stops. The company is choosing between a shrinking pipeline and a shrinking return.

2. IPCA-indexed debt hits profit before IPCA-indexed revenue arrives

Mechanism. As of Q2 2026, 63.8% of consolidated debt is indexed to IPCA, and 58.9% of RAP is indexed to IPCA. The indices match; the timing does not. Inflation accretes onto the debt principal and flows into financial expense continuously through the year. The RAP that offsets it resets once, in July. In a rising-inflation environment, the cost arrives first and the revenue lags by up to twelve months. Layer on a construction book where the debt is drawn years before the corresponding RAP begins, and the drag compounds.

This is not hypothetical. Alupar's Q2 2026 regulatory net profit fell year on year, and the company attributed it to increased financial expense pressured by the IPCA-indexed debt profile alongside investments in transmission projects under development.

Calibration. High probability, moderate and recurring drag. It is a structural feature, not an event.

3. Curtailment is destroying the economics of the generation portfolio

Mechanism. The ONS orders wind and solar plants to stop producing when the grid cannot absorb or move their output. The generator loses the revenue. Brazil's curtailment rate more than doubled from 9.3% of wind and solar output in 2024 to 20.6% in 2025, ran at 17.8% through May 2026, and is forecast by the ONS to keep rising to 3.65 GW average by 2029.

Alupar is not a bystander here. In Q2 2026, SIN-wide wind curtailment was 15.7% (2.2 GW average), up from 11.7% in Q2 2025, and reporting on second-quarter results placed Alupar among the most affected portfolios in June, with cuts exceeding 28% of potential generation. Alupar's own quarter reflected a generation shortfall of 13.1 percentage points versus expected output, of which 10.5 points were non-reimbursable and only 2.6 points fell into the reimbursable category.

The July 2026 compensation framework does not fix this. It covers grid-related and reliability-driven cuts from 2023 to 2025 only, and explicitly excludes curtailment caused by oversupply, which is the dominant and fastest-growing cause.

Calibration. High probability, already occurring, moderate impact at group level because generation is roughly a quarter of revenue and a fifth of segment EBITDA. But it is getting worse, not better, and it affects the segment with no contractual protection.

4. Execution risk concentrated in a four-year construction window

Mechanism. Thirteen transmission systems are under construction with completion staged from 2026 through 2029, against an aggregate programme management sized at roughly R$10 billion on the Q4 2025 call. Every one of those projects has a contractual energisation deadline with ANEEL or ProInversión, and missing it triggers revenue deferral and, in the Brazilian framework, penalties. The projects are simultaneously the entire growth story and the entire near-term drag: capex flows out, no RAP flows in until energisation.

Management has guided net debt to EBITDA peaking at 3.9x-4.0x in 2028, up from 3.2x at Q2 2026 and 3.3x at end-2025. That guidance assumes the programme lands roughly on schedule. A one-year slip on TAP - the single largest project at approximately R$2.05 billion of capex - would extend the leverage peak and push it higher, precisely when Alupar would want balance-sheet room to bid the October 2026 and 2027 auctions.

Calibration. Moderate probability of individual project slippage, low probability of systemic failure. Alupar's own track record in 2025 argues for the low end. But the base rate for Brazilian transmission is not reassuring: TNE took fourteen years from auction (2011) to commercial operation (September 2025).

5. Environmental licensing can add years, not months

Mechanism. No amount of engineering competence compensates for a licence that has not been issued. TAP is the live illustration: a project of 551 km and roughly R$2.05 billion of capex that could not begin construction until IBAMA issued the unified installation licence in July 2026. Management had been reporting TAP at 28% physical progress as of the Q4 2025 call in March 2026, which reflected preparatory work, land acquisition and procurement rather than line construction.

The risk compounds because the RAP clock does not stop for a licence. Alupar carries land, financing and mobilisation costs while waiting.

Calibration. High probability of some delay on any given project, and it is the single largest historical source of Brazilian transmission overruns.

6. Tax: two separate pressures arriving at once

Mechanism, part one - expiring regional incentives. Alupar's northern and northeastern assets have benefited from SUDAM and SUDENE incentives, which reduce IRPJ by 75% for a ten-year enjoyment period on qualifying projects. Those windows close asset by asset. Alupar's own Q2 2026 profit decline was attributed in part to higher income tax expense following the expiration of tax benefits. Separately, LC 224/2025 imposes a linear 10% reduction to federal tax incentives, phased 5% in 2026 and 5% in 2027.

Mechanism, part two - the consumption tax transition. PIS/Cofins is extinguished in 2027 and replaced by CBS. Estimates put concessionaire tariff revenue moving from a 3.65% rate to roughly 11-12%, approximately a tripling of the headline rate. ANEEL has already begun adjusting: from 1 July 2026 it removed PIS/Pasep and Cofins amounts from the RAP calculation base for concessions contracted through 31 December 2005. The governing principle is neutrality.

Management's position is that the contracts protect them:

"Most of our contracts, the revenue of RAP is net, grossing up the taxes."

• Luiz Coimbra, Q2 2026 concall, 7 August 2026

Note the word "most." A portfolio of 44 concessions signed across a quarter century will not have uniform tax gross-up language. Neutrality is the regulatory intent; contract-by-contract implementation across a 2033 transition is where leakage occurs.

Calibration. The regional incentive expiry is high probability and already visible in reported results. The CBS transition is moderate probability of some leakage, low probability of a material adverse outcome, but high uncertainty over a seven-year window.

7. Latin American country and currency exposure

Mechanism. Roughly 17% of contracted RAP for the 2026/27 cycle is PPI or USD-equivalent indexed, which is the international book. That is a genuine hedge against real depreciation, but it introduces two new exposures. First, translation: management flagged a negative FX impact of approximately R$20 million from Peruvian sol depreciation on the Q1 2026 call. Second, regulatory: Peru, Colombia and Chile each have their own tariff-setting regimes, and Alupar has far less institutional history with them than with ANEEL. Management acknowledged international regulatory risk explicitly as a caution on the Q2 2025 call.

The Peruvian exposure is growing fastest - TCN in construction, Maravilla and Puno Sur licensed, Group 3 running to 2029 - exactly as management targets international assets at 18%+ of consolidated revenue.

Calibration. Moderate probability, moderate impact. The offsetting argument is that Colombian and Chilean concessions are indefinite rather than 30-year, which is a real structural improvement over the Brazilian contract.

8. Cost of capital in an auction where the lowest bidder wins

Mechanism. Alupar's competitors in ANEEL auctions include a sovereign-backed Chinese utility, a newly privatised national champion with a R$158 billion market capitalisation, a European utility subsidiary, and Dutch-pension-backed infrastructure vehicles. Every one of them can accept a lower return than a Brazilian family-controlled mid-cap financing itself in the domestic debenture market at IPCA plus roughly 7%. Management has conceded the point in plain language:

"We are a company of greenfield, our capital cost is a little bit higher than other companies..."

• CFO José Luiz de Godoy Pereira, Q3 2025 concall, 7 November 2025

The company's answer is construction efficiency. That answer works only as long as the construction advantage exceeds the financing disadvantage. At 50%+ discounts, the margin between the two narrows.

Calibration. High probability of continued competitive pressure; the mechanism by which "no growth" rather than "bad growth" becomes the outcome.

9. Systemic risk to the tariff pool and to system risk parameters

Mechanism. Two related versions. First, political: Brazilian electricity tariffs are a recurring political flashpoint, and the RAP pool is funded by those tariffs. A government intervention in the TUST mechanism would reach every concessionaire at once.

Second, operational: Alupar's CFO has publicly warned that pressure from market participants to relax the conservative risk parameters used in the system's dispatch models (the CVaR settings that govern how defensively the ONS holds reservoir water) could push the system toward a critical position. Management reiterated caution on proposed CVaR changes on the Q1 2026 call, 8 May 2026. This is unusual and worth noting: a transmission company warning about generation adequacy is warning about the health of the system it is paid by, not about its own line availability.

Calibration. Low probability, high severity. Not a 2026 concern; a standing structural one.

10. Concentrated family control

Mechanism. Guarupart Participações holds roughly 52% of total capital, and the Godoy Pereira family occupies the chair, the chief executive role, the CFO role and two alternate board seats. Alupar lists on B3's Nível 2, not the Novo Mercado, so preferred shares carry limited voting rights. Minority holders of ALUP4 and ALUP11 have very little ability to influence capital allocation, related-party arrangements, or the auction discipline the entire thesis depends on.

The counterargument is that this same concentration has produced twenty-five years of consistent strategy and a management team whose own wealth sits in the stock. It cuts both ways.

Calibration. Low probability of acute governance failure; permanent structural feature that limits minority recourse.

9. Walk the talk

The six calls under review: Q1 2025 (9 May 2025), Q2 2025 (8 August 2025), Q3 2025 (7 November 2025), Q4 2025 (6 March 2026), Q1 2026 (8 May 2026), Q2 2026 (7 August 2026).

A word on how to judge this management before judging it. Alupar issues no revenue guidance, no earnings guidance and no capex guidance in numeric form. Every quarter in the period under review was reported as an "earnings miss" by sell-side aggregators, and every one of those misses is an analyst modelling error rather than a management shortfall, because management never put a number on the table to miss. What management does commit to is physical: project completion percentages, energisation dates, funding milestones, and a payout policy. That is what can be tracked, and it is what follows.

Starting point: Q1 2025

On the May 2025 call, management laid out a staging schedule with unusual specificity. TCE in Colombia was reported at 99% complete. ELTE at 98%. TNE at 84%. Alongside this, they announced a new ETVG reinforcement in Mato Grosso at R$9.9 million of RAP and a 13.7% RAP-to-capex ratio, and declared an interim dividend of R$0.21 per unit at a 52% payout of regulatory net income.

Three assets reported at 84-99% complete is a falsifiable claim with a short fuse. Within six months, all three energised:

• ELTE: commercial operation July 2025. Additionally, the Manoel da Nóbrega substation reinforcement was brought into commercial operation ahead of schedule in May 2025, adding R$14.2 million of annual permitted revenue at 2024/25 cycle values. That was announced in the same month as the call.
• TNE: commercial operation 16 September 2025, connecting Roraima to the national grid.
• TCE: commercial operation registered 22 October 2025, Alupar's first operating transmission asset in Colombia.

Three for three, on the stated sequence, within two quarters. That is the strongest single data point in this section.

Q2 2025: the staging holds

The August 2025 call reported that TC in Colombia had concluded works and was moving to commissioning with the regulatory bodies, that STEP had advanced from roughly 10% at the start of 2025 to 20%, and that a further block of eight projects was in development - three in Chile, four in Peru, one in Colombia. Management flagged caution on energy acquisition costs and on international regulatory risk.

The Colombian commissioning claim was confirmed by the October 2025 energisation. The eight-project international block is the pipeline that by Q2 2026 had become TCN at 79%, Maravilla and Puno Sur licensed, and the Chilean Atacama service scheduled for 2027. Nothing in that block has been quietly dropped across four subsequent calls, which is itself informative.

Q3 2025: an early delivery, and a promise about leverage

The November 2025 call carried the most quotable commitment of the period. Management reported TECP Phase 1 at 33% completion and stated it was running "11 months ahead of original deadline."

The verification is immediate and unambiguous. Phase I of the TECP Centro substation was energised in November 2025, against a contractual deadline of June 2026. Eleven months early, exactly as claimed, and the RAP consequence was concrete: Alupar's entitlement stepped from 11.84% of the project's integral RAP, which it had been receiving since April 2024, to 33.08%.

The same call introduced the leverage argument that management has repeated ever since:

"When you have a very high backlog, your leverage level will increase during implementation process. Then it goes down, and it goes down very fast."

• CFO José Luiz de Godoy Pereira, Q3 2025 concall, 7 November 2025

And the Peru project was described with a full specification: 250 km, five substations, US$220 million, 23.8% physical advance, commercial operation targeted November 2029. By Q2 2026, the associated Peruvian assets had their environmental licences and 100% of substation land.

Q4 2025: the funding promise, and a number attached to the leverage claim

The March 2026 call is where management converted the qualitative leverage argument into a falsifiable number. They guided net debt to EBITDA peaking at 3.9x-4.0x in 2028, having ended 2025 at 3.3x (down from 3.5x in 2024), and said the descent would follow as projects entered operation. They also put a figure on the programme: "we're talking more or less about $2 billion, roughly R$10 billion total," to be completed by 2029.

Most importantly, they made a specific funding claim:

"Essa emissão resolve praticamente todo o funding do nosso ciclo de implantação no Brasil." ("This issuance essentially resolves all the funding of our Brazilian implementation cycle.")

• Luiz Coimbra, on the R$2.45 billion TECP debenture, Q4 2025 concall, 6 March 2026

This was verified on the following call. The Q1 2026 call confirmed the R$2.45 billion was raised at IPCA plus 6.9928% over twelve years, which management characterised as NTN-B 2040 minus 0.40%. Fitch reaffirmed AAA on the national scale and BB+ internationally with a stable outlook. The funding claim was delivered, at a spread through the sovereign real curve, and no equity was issued to do it.

On auctions, management refused to pre-commit:

"Each auction is an auction, depends on the number of aggressive players."

• Q4 2025 concall, 6 March 2026

That is not a promise, but it became a testable statement of discipline three weeks later, when Auction 01/2026's first stage cleared at a record 50.68% average discount and Alupar won nothing. They then took the single largest lot in the July second stage. Stated discipline, observed behaviour, match.

Q1 2026: the first visible cost of the strategy

The May 2026 call is where the bill for the construction cycle started showing. Management reported TECP Phase 2 at 31.6%, TAP and TAPT funded, and the Peruvian sol depreciation costing approximately R$20 million. General and administrative expenses were up 17.2%, attributed to operating the newly commissioned assets - which is the correct and expected consequence of energising three projects in 2025.

Management also reaffirmed the dividend cadence with another R$0.21 per unit interim, and confirmed interest in battery storage auctions because of the "synergy with our strategy."

What was not repeated on this call, and is worth flagging: the specific 3.9x-4.0x 2028 leverage peak was not restated. Reported leverage at Q1 2026 was 3.2x on a regulatory basis and 2.9x on a corporate basis, i.e. moving in the opposite direction from the guided path. That is not a broken promise - the peak is guided for 2028, not 2026 - but it does mean the single most important quantitative commitment in the file is two years from being testable.

Q2 2026: delivery and a cooling note

The August 2026 call delivered on the two things investors had been waiting for. TAP received its full IBAMA installation licence in July 2026, releasing construction on the largest project in the book after a licensing wait that management had been navigating since at least the Q4 2025 call. And Alupar won Lot 7, its first Brazilian auction win of 2026, at a 52% discount with an expected 30% capex saving.

TCN in Peru jumped from 28% to 79% complete in a single quarter and was reaffirmed as operational during 2026. TECP Phase 2 reached 90%. The dividend was R$0.21 per unit again.

The cooling note was on batteries. Having expressed interest in Q3 2025 pending regulatory clarity, and confirmed interest in Q1 2026 citing strategic synergy, management by Q2 2026 was citing a "high level of supply and registered amounts" creating uncertainty and emphasising a disciplined approach. That is a three-call arc from interested to cautious, and it is the correct pattern to see from a management team that says it will not bid unless the maths works. It is also, if you are looking for growth, a soft withdrawal from an adjacency that had been talked up.

Promise versus outcome

Assessment

This is management that does what it says on the things it commits to, and is careful not to commit to much.

The physical record is genuinely good and it is the record that matters in this industry. Three projects staged and delivered in 2025 on the sequence disclosed. A substation phase energised eleven months early with a verifiable RAP consequence. A multi-billion-real funding round executed at a spread through the sovereign curve, with no equity dilution, exactly as promised one quarter earlier. An auction sat out when discounts hit a record, and a lot won when the maths worked. Across six calls there is no instance of a stated completion percentage or energisation date being walked back.

The one commitment that carries real weight and cannot yet be graded is the 2028 leverage peak. Everything about the investment case - the construction book, the auction discipline, the dividend trajectory - resolves into whether net debt to EBITDA tops out near 4.0x in 2028 and then falls "very fast." That is the promise to hold them to.

10. Shareholder friendliness index

Dividends. Per Alupar's own dividend disclosure, distributions were R$347.4 million (R$1.14 per unit) in 2023, R$275.7 million (R$0.87 per unit) in 2024, and R$356.0 million (R$1.08 per unit) in 2025, with R$138.4 million (R$0.42 per unit) paid year to date in 2026. The path is not a clean trend: 2024 fell roughly 21% per unit before 2025 recovered around 24%. The 2024 dip was a payout-ratio and earnings-base effect rather than a policy change - the board's policy since 2022 is to distribute at least 50% of regulatory net income, and the realised payout was 55% in 2023, 51.5% in 2024 and 50.8% in 2025. Two things are worth flagging. First, per-unit figures across years are not strictly comparable because Alupar capitalises reserves into a ~4% stock bonus most Aprils, so the unit count grows. Second, the payout ratio has declined in each of the last three years and the Q2 2026 interim was struck at 48.9%, below the stated 50% floor, which is consistent with management's Investor Day message that capital is being prioritised toward the project pipeline over distributions. Dividends have been paid in each of the last thirteen consecutive years.

Buybacks and dilution. Alupar announced its first-ever repurchase programme on 11 December 2024, authorising up to 3,720,536 units (1.17% of total capital) over eighteen months, i.e. to roughly June 2026, funded from profit reserves, with units to be held in treasury, cancelled or resold. That is the only programme identified in a three-year search of Brazilian exchange announcements, financial press and the company's own IR materials; nothing else was found for 2023, 2024, 2025 or 2026 to date. The quantity actually repurchased under that authorisation could not be verified - Alupar has not published an execution summary in accessible materials, and the CVM filing detail was not retrievable. On share count, the direction is the opposite of retirement: the company executed 4% stock bonuses in both April 2023 (35,164,450 new shares, a R$328.8 million capitalisation) and April 2025 (38,033,869 new shares, a R$349.5 million capitalisation), both funded by capitalising the Investment Reserve rather than by new shareholder contributions. These are economically neutral, like a stock split, but the share count is growing roughly 4% a year, not shrinking. Units outstanding stood at 329,626,867 as of 6 November 2025.

Verdict: Returns Capital, but decreasingly so. Thirteen unbroken years of dividends at a policy-anchored payout is a genuine commitment, yet the payout ratio has fallen in each of the last three years and slipped below its own 50% floor on the most recent interim, because management has explicitly redirected capital toward a R$10 billion construction programme running to 2029.

11. Insider activities

Source access, stated plainly

Brazil's regime for insider disclosure is CVM Resolution 44 (which consolidated and revoked Instructions 358, 369 and 449). Article 11 requires the company, its direct and indirect controlling shareholders, its directors, its executive officers and its fiscal council members to report their securities trading and holdings on a monthly basis. Article 12 requires disclosure when any holder crosses 5% of a share class.

I made a genuine attempt to retrieve the Article 11 filings and could reach the category but not the documents:

• Alupar's own IR site does carry the category - the CVM Filings page lists "Negotiated and Held Securities (Art. 11 of CVM Instr. nº 358)" as a document type - but the individual filings are behind a year-and-category selector that returns no document list to a non-interactive fetch, and the direct Portuguese-language paths return 404.
• The InvestSite aggregator page indexed specifically for ALUP4's "valores mobiliários negociados e detidos" category returns HTTP 403.
• The CVM's own RAD filing portal is an authenticated ASP.NET system that does not serve document lists to external retrieval.

Therefore: individual director and executive officer transaction data for Alupar over the last 12 months could not be located within the search budget. I am not going to infer or construct transactions that I did not see filed. What follows is what I could verify from primary disclosure.

Verified disclosures, last 12 months

Buys - reading the signal

There is no verified open-market purchase by any director or executive officer of Alupar in the last twelve months in the sources available. That is a statement about source access, not about insider conviction, and it should be read as such. Absence of retrievable evidence is not evidence of absence, and I am explicitly not treating it as a bearish signal.

The one purchase visible in primary disclosure is BlackRock's accumulation through late 2025 and January 2026 to 5.012% of the preferred class. This is not a conviction signal in the sense the term is normally used. BlackRock's own notification states the holding is for financial purposes with no intent to influence control, and a position of this shape at an index manager is overwhelmingly a function of benchmark weighting and fund flows rather than a discretionary view on Alupar's transmission pipeline. The subsequent drift back below 5% by May 2026 reinforces that reading: this is a position oscillating around an index weight, not a manager building a stake.

Sells - working out the why

There is no verified insider sale by a director or executive officer in the last twelve months in the sources available.

The one reduction visible is BlackRock moving from 5.012% to approximately 4.97% of the preferred class between January and May 2026. Reason not disclosed, and none needs to be inferred: a 4-basis-point move in a class position at a passive-weighted manager is mechanical rebalancing. It is not a signal about the business.

One structural point that changes how any Alupar insider data should be read. Control is held by Guarupart Participações, the Godoy Pereira family holding, at roughly 52% of total capital and a reported 74.71% of the ordinary class. Family exposure to the company sits in that holding vehicle, not in personally-held ALUP11 units, so the individual-director trading file would in any case be a thin proxy for insider alignment here. The meaningful alignment signal at Alupar would be a change in Guarupart's stake, and no such change was disclosed in the period reviewed.

Net assessment

On the evidence that could be verified: neutral, with an explicit data gap. There is no cluster buying to flag and no distribution to worry about. The controlling family has not reduced, which for a founder-controlled Brazilian infrastructure company is the most relevant single fact and is mildly reassuring. The only active hand visible in the register over twelve months is a passive institution crossing and re-crossing a 5% threshold, which carries no information about the business.

The honest read for a potential investor is that the insider signal here is unavailable rather than absent, and anyone who wants it should pull Alupar's monthly Article 11 filings directly from the CVM's RAD portal or from the company's IR "Negotiated and Held Securities" section using an authenticated session, which would resolve the question in a few minutes of interactive browsing that could not be replicated here.

12. Scenarios

Bull case

The construction book converts on schedule and Alupar emerges from 2029 as a materially larger company with the same risk profile it has today.

The sequence starts with TAP. The IBAMA licence arrived in July 2026 and the 551 km across Goiás, Minas Gerais and São Paulo gets built without a repeat of the corridor problems that have historically dogged long Brazilian lines. Real capex lands meaningfully below the R$2.05 billion reference, as it has on prior projects, and the line energises early enough that the RAP starts flowing before the leverage peak management guided for 2028. Lot 7 in São Paulo proves the thesis on urban underground work: the 30% capex saving management flagged materialises, buried circuits turn out to be exactly the kind of project where Alupar's advantage is largest because there is no right-of-way to negotiate with hundreds of landowners, and the company establishes itself as the natural bidder for the metropolitan reinforcement lots that ANEEL's pipeline is increasingly weighted toward.

In parallel, the Latin American bet pays. TCN energises in Peru during 2026 as guided, Maravilla and Puno Sur come through with licences already in hand and land secured, and the Group 3 portfolio hits its November 2029 date. Those assets earn at RAP-to-capex ratios well above what Brazilian lots now clear at, they are paid in dollar-linked revenue, and the Colombian and Chilean concessions run indefinitely rather than expiring at thirty years. Alupar hits the 18%-plus international revenue share it has targeted, and the market starts valuing that slice differently from the Brazilian book because it is longer-dated and less exposed to a single regulator and a single currency.

The financing side cooperates. The R$2.45 billion already raised covers the Brazilian cycle, so Alupar never has to come back for equity. Brazilian real rates come down over the period, which does two things at once: it lowers the cost of the remaining funding and it makes a long-duration inflation-linked annuity more attractive to the domestic institutions who are the natural buyers of this asset. Leverage tops out near 4.0x in 2028 exactly as guided and then falls quickly as thirteen projects turn on, one after another, each adding contracted revenue with almost no incremental cost.

And on the generation side, the thing that has been hurting stops getting worse. The grid catches up with the turbines, curtailment peaks rather than compounding, and the compensation framework gets widened beyond the narrow 2023-2025 grid-and-reliability window. The wind and solar portfolio goes from a source of negative surprises back to a quiet cash contributor. With the construction cycle behind it and a deleveraging balance sheet, Alupar resumes bidding from a position of strength into a pipeline that still has R$8.9 billion in October 2026 and R$12.9 billion in 2027 behind it, and the payout ratio climbs back toward the 55% it ran at in 2023.

Base case

Nothing breaks, nothing dazzles, and the company spends three years being a construction site.

The thirteen projects under way mostly land within a year of their contractual dates. One or two slip - the base rate for Brazilian transmission says at least one will - and the slippage is absorbed without penalty drama because Alupar has built enough schedule buffer into projects like TECP that the average holds. TAP, being the largest and the most recently licensed, is the one to watch, and it probably arrives closer to the back of its window than the front. TCN in Peru energises during 2026 as stated, because 79% complete in August with a year-end target is a claim that is difficult to miss by much.

Leverage rises roughly along the guided path toward 3.9x-4.0x in 2028. It does not fall as fast afterward as management's "very fast" suggests, because each auction Alupar wins in 2026 and 2027 adds a new capex tail just as the old one runs off. That is not a failure; it is what a growing greenfield developer looks like. The company continues to participate selectively in ANEEL auctions, wins perhaps one lot a year at discounts above 50%, and earns a return on those lots that depends entirely on construction execution rather than on the tariff.

Reported profit stays under pressure for several more quarters. The IPCA-indexed debt keeps charging financial expense ahead of the July RAP reset, the SUDAM and SUDENE incentives keep rolling off asset by asset, and the CBS transition adds noise from 2027. None of it touches the contracted revenue stream, which continues to grow as assets energise, but it does mean the gap between "the business is fine" and "the earnings line looks weak" persists, and the stock keeps getting reported as missing consensus estimates that management never set.

The generation segment stays a drag. Curtailment does not improve on the ONS's own forecasts, the compensation framework stays narrow, and Alupar's wind portfolio keeps giving back a share of its expected output. Because generation is roughly a quarter of revenue and a fifth of segment EBITDA, this is annoying rather than dangerous. Management does not build more of it.

The dividend keeps coming, at a payout ratio hovering around the 50% policy floor rather than the 55% of 2023, and the annual 4% stock bonus keeps the unit count creeping up. Guarupart does not sell. By 2029 Alupar is a bigger, more internationally diversified transmission company with roughly the same business model it has today, having spent the intervening years converting debt into contracted annuities.

Bear case

The construction cycle goes wrong at exactly the moment the balance sheet has no room, and the growth engine stalls at the same time.

The specific adverse path starts with TAP. It is the single largest project in the book, it only got its full installation licence in July 2026, and it crosses three states. A corridor dispute, an archaeological or indigenous land finding, or a supplier failure on transformers pushes energisation out by two years. The RAP does not start, the debt service does, and the 2028 leverage peak that management guided at 3.9x-4.0x arrives at something materially higher and arrives later. Covenant headroom, which management has said it is renegotiating, becomes the operative constraint rather than a background detail. Alupar stops bidding, not by choice but because it cannot fund a new lot while the last one is unfinished. The October 2026 and 2027 auctions pass and competitors take them.

Simultaneously, the auction market prices Alupar out permanently. Axia, post-privatisation and carrying a balance sheet many times larger, keeps bidding 55-59% discounts. State Grid and the pension-backed vehicles keep accepting returns Alupar cannot match on its domestic cost of capital. The company's own CFO has already conceded that "our capital cost is a little bit higher than other companies." At 50%-plus discounts, a construction advantage of 25-30% against reference capex is no longer enough to bridge that gap. Alupar becomes a company with a fine existing concession book and no way to add to it - a runoff asset with a thirty-year tail rather than a growth business, and the market revalues it accordingly.

The Latin American bet, which was supposed to be the escape route, turns into a second front. Peru's regulatory framework proves less predictable than ANEEL's, or the sol keeps depreciating, or political instability delays the Group 3 projects past their November 2029 deadlines. Management already booked a roughly R$20 million negative FX hit from the sol in a single quarter. Chile and Colombia are new enough that a first regulatory dispute would be genuinely unfamiliar ground.

On generation, curtailment keeps compounding past the ONS's 3.65 GW 2029 forecast because the grid build-out that Alupar is part of does not arrive fast enough. Having already lost more than 28% of portfolio potential generation in a single month, and with more than three-quarters of the shortfall falling outside the reimbursable category, the wind assets become a structurally impaired business rather than a cyclically weak one, and the compensation framework's exclusion of oversupply curtailment means no relief is coming.

And then the tax transition does not work as advertised. Management's phrasing was that "most" contracts gross up taxes. Across 44 concessions signed over twenty-five years, the ones that do not become visible as PIS/Cofins gives way to CBS at roughly triple the headline rate between 2027 and 2028, and the regional SUDAM and SUDENE incentives expire on the northern assets that carry some of the largest RAP in the portfolio. Contracted revenue is unchanged; what reaches shareholders is not. The payout ratio, already drifting down each year and already below its own 50% floor on the latest interim, gets cut outright to protect the construction programme, and the one thing that reliably brought income investors to this stock goes away.