Brazil's energy security, Itaipu and the US–Paraguay AI infrastructure contest

1. Electricity security as a physical and institutional system

A country may produce abundant electricity over twelve months and still experience local blackouts because transmission equipment has failed. It can maintain large installed generating capacity yet fail to meet an evening peak if its available generation cannot be dispatched at the necessary moment. Electricity security involves resource adequacy, network reliability, flexibility, affordable costs, spare equipment and the capability to restore service after disruption.

Brazil's portfolio includes hydropower, wind, solar, thermoelectric plants and other resources with different operating characteristics. Reservoirs can provide flexibility within water and environmental constraints. Wind and solar depend on weather and time. Thermal resources can supply dispatchable generation but require fuel and financing. Batteries can support short-duration balancing without necessarily replacing large water reservoirs during prolonged drought.

Institutional coordination is essential. The National Electric System Operator coordinates Brazil's interconnected system; EPE undertakes long-term planning; ANEEL regulates critical parts of the market; public and private utilities own physical assets. Itaipu adds an international layer in which national authorities must respect a shared legal entity and negotiated entitlements.

2. The difference between Brazil's electricity mix and total energy mix

An electricity mix describes sources used to supply electricity. The broader energy balance includes petroleum, gas, biofuels and fuels consumed directly by factories and vehicles. Brazil may generate most of its electricity from renewable sources while remaining dependent on fossil fuels in transportation and industrial processes. Confusing the categories creates inflated claims of national energy independence.

EPE's 2026 National Energy Balance, covering 2025, reports 86.6% renewable electricity in internal supply, with 52.2% hydraulic participation. Internal electricity supply reached 783.3 TWh, including net imports of 7.4 TWh, while final consumption was 667.8 TWh. These are distinct accounting concepts, affected by system losses and other uses. The difference is not a reserve of electricity that can simply be allocated to new data centers. [7]

Brazil enjoys substantial low-carbon electricity advantages, but resource abundance does not guarantee cheap deliverable firm capacity at a specific site. The time profile of demand, transmission congestion and contingency requirements influence the costs of connection and reliable supply.

3. Formation of Brazil's electricity industry

Early Brazilian electrification developed around local cities, industrial customers, private concession operators and nearby hydroelectric sites. Growing urban populations and manufacturing forced the expansion of generation and the construction of lines connecting previously separated systems. Energy increasingly became an economic infrastructure question rather than a narrowly municipal utility service.

Twentieth-century public institutions and regional electricity companies sought to accelerate investment. Eletrobras and its subsidiaries played major roles in planning and engineering high-voltage grids and major hydroelectric works. Electricity became essential to steelmaking, transport, urbanization and the spread of consumer appliances.

Financing and technical equipment were not always available domestically. Foreign technology and capital could facilitate necessary investment while creating exposures to exchange rates and lender conditions. These dependencies were not proof that domestic actors lacked decision-making power; they were constraints that influenced the cost and timing of their choices.

4. The Paraná basin and geographical interdependence

The Paraná River belongs to a basin shared by Brazil, Paraguay and Argentina. Hydropower therefore intersects with navigation, downstream water use, flood management and cross-border law. A project affecting river levels and flows can impose costs or restrictions on another country even when the latter is not a shareholder in the operating company.

Itaipu's location provides access to large Brazilian power markets and is especially significant to the much smaller Paraguayan electricity system. Its creation followed earlier boundary and river-use disagreements involving the former Guaíra Falls. The engineering project required settling difficult questions of territory, legal rights and interstate cooperation.

A dam of this size is a political agreement expressed in infrastructure. Water, equipment and transmission lines cannot easily be separated from the treaties governing them. Unilateral attempts to alter important aspects of the system can carry legal, diplomatic and operational consequences.

5. The War of the Triple Alliance and Paraguayan national memory

The War of the Triple Alliance from 1864 to 1870 profoundly affected Paraguay's population, economy and political relations with Brazil and Argentina. Its legacy is part of the country's sense of vulnerability in relations with larger neighbors. This history does not dictate every modern negotiating position, but helps explain why control over Itaipu revenues and electricity commercialization remains politically sensitive.

Paraguay contributed to the legal and territorial basis of the shared enterprise and has a legitimate interest in using its allocation to support domestic industry. Brazil supported much of the financing and infrastructure needed to build the dam and absorb power that Paraguay initially could not use. Both facts matter to a complete historical accounting.

The structural asymmetry is considerable. Brazil possesses a larger domestic market, broader industrial capacity and more financing alternatives. Paraguay values the opportunity to convert a renewable resource into local productive activity. Sustainable cooperation requires recognizing that their optimal short-run choices will not always coincide.

6. The 1960s: military regimes and state-led development

Itaipu was conceived as Brazil was governed by a military dictatorship and Paraguay by Alfredo Stroessner's authoritarian regime. Large public works served industrial growth, territorial integration and the prestige of governing elites. Restrictions on democratic competition and public participation limited how environmental costs, displacement and distributive questions were debated at the outset.

Infrastructure built under dictatorship can produce genuine long-term economic benefits, but its institutional origins still matter. Engineering accomplishments do not erase deficiencies in consultation, civil liberties or the transparency of original decisions. Political history needs to describe both the productive outcomes and the conditions under which affected citizens could participate.

The broader Cold War is relevant because the United States had considerable influence in the region. Yet general American influence is not proof that Washington personally wrote, imposed or controlled the 1973 treaty. A claim of specific covert interference would require diplomatic records or other evidence particular to Itaipu.

7. The 1973 treaty and shared sovereignty over generation

The treaty was signed on 26 April 1973 and entered into force after ratification instruments were exchanged on 13 August 1973. It created a binational entity with equal rights and obligations for the contracting states, supplemented by annexes governing administration, technical facilities and financial services. It is a specialized international regime, not a Brazilian license issued to Paraguay. [8][9]

The arrangement recognizes an equal entitlement to power and sets conditions for the allocation of energy not used by one party. Electricity commercialization is undertaken through national entities and contracts. The physical flow of energy can differ from the juridical fifty-fifty entitlement because actual demand, frequency conversion, network design and deliveries vary.

Consequently, a statement that Paraguay receives fifty percent of Itaipu is a statement about treaty rights rather than a claim that exactly half of annual output physically serves Paraguayan consumers. The distinction is central to Paraguay's growing interest in local demand that could reduce the surplus historically made available to Brazil.

8. Annexes A, B and C: how the binational arrangement works

Annex A concerns the company's institutional organization. Annex B describes its engineering facilities. Annex C establishes financial arrangements and the provision of electricity services. Its cost-based framework includes contracted capacity, expenses, royalties, compensation for ceded electricity and other components. A rate charged for capacity is not the same as a wholesale energy price. [8][10]

The traditional Unit Cost of Electricity Service, expressed in US dollars per kilowatt-month, cannot be directly compared with a data center power purchase agreement denominated in dollars per megawatt-hour. A proper conversion requires assumptions about capacity, energy output, utilization, contract duration and additional charges.

Annex C provided for review after fifty years from the treaty's entry into force, considering debt amortization and the relative volumes of contracted capacity. That milestone was reached in 2023. It did not establish a predetermined new tariff or permit one country to dictate revised conditions without an agreement.

9. The engineering system: twenty turbines and different frequencies

Itaipu's twenty 700 MW generating units provide 14,000 MW of nominal installed capacity. Portions of the generation system were designed for Paraguay's 50 Hz network and others for Brazil's 60 Hz network. Conversion and specialized high-voltage transmission allow the systems to exchange the power required by their different operating standards.

Installed capacity is not annual energy production. The theoretical maximum of 14 GW sustained for all 8,760 hours in a conventional year is 122.64 TWh, while actual 2025 generation was 72.88 TWh. The ratio is approximately 59.4%, a calculated annual capacity factor that should not be mistaken for mechanical availability or reliability. [2]

Even when adequate water and generating units exist, a converter or transmission failure can disrupt deliveries to one side of the system. Reliability therefore requires maintenance, redundant routes, replacement equipment, sound protective relays and coordinated operational planning across borders.

10. Guaíra Falls and the costs of large-scale development

Itaipu's reservoir transformed the landscape and submerged the former Guaíra or Sete Quedas falls, a site of major scenic and ecological importance. People were displaced and ecosystems altered as the governments prioritized the construction of a major electricity source. Some losses cannot be fully represented in financial accounts.

The project also created productive capacity that supported industrial activity and urban life far beyond the river valley. Royalties and local investments offered mechanisms for compensating affected territories, without making every historical impact reversible.

New data centers and transmission lines raise smaller but related distributional questions. Their economic benefits may be realized in global corporate revenue while water, land and equipment construction affect local residents. Cost-benefit assessment must identify where burdens fall rather than treating national development claims as substitutes for consultation.

11. Argentina and the 1979 Tripartite Agreement

Argentina became an essential diplomatic actor because it is downstream on the Paraná River. During the 1970s, Buenos Aires was concerned that the Brazil–Paraguay project would constrain Argentina's own water use and hydroelectric plans. Academic historical research describes how technical and environmental arguments became intertwined with geopolitical competition. [11]

The Tripartite Agreement of 19 October 1979 among Brazil, Paraguay and Argentina established a framework for coordinating hydraulic uses and relevant levels and operational conditions in the common river system. Itaipu's institutional history records the accord as an important element in resolving the controversy surrounding the plant. [12]

This episode is more informative for energy security than a simplistic story of one dam dominating its neighbors. Shared basins create legitimate competing interests, and negotiated mechanisms can reduce disputes and support long-lived infrastructure.

12. The 1973 oil shock and a broader industrial strategy

The treaty's signing coincided with the first oil shock, which greatly increased the importance of alternative energy sources. Brazil expanded investment in hydropower, domestic petroleum development, ethanol and transport infrastructure while trying to manage import dependence and rapid industrial growth. [13]

Large infrastructure projects required long-term finance. Interest rates, currency changes and delays could materially increase obligations. Latin American countries accumulated significant external borrowing through the 1970s and encountered severe debt pressures in the following decade. [14]

This history demonstrates that energy independence cannot be pursued through construction volume alone. Secure electricity must remain financially sustainable, otherwise debt burdens and tariff distortions may constrain the maintenance and investment necessary for future security.

13. Itaipu's debt and the dispute over financial fairness

An influential Columbia Center on Sustainable Investment study described the project's debt financing, including a substantial Eletrobras loan in 1975 and smaller international equipment-related facilities. Currency adjustments, interest and the long amortization profile influenced the cost of electricity and debates over the division of benefits. [15]

Paraguayan perspectives emphasize the unequal economic advantages of the initial consumption and financing arrangement. Brazilian arguments highlight the construction risks borne by larger financial institutions and the provision of a market for electricity not originally demanded in Paraguay. The two narratives often refer to different kinds of cost and benefit.

Paying down construction debt before and around the 2023 review milestone changed the financial context. It did not eliminate operating expenses, maintenance, environmental obligations, reinvestment or prospective transmission upgrades. The phrase “the debt is paid” cannot rationally be interpreted as “the power is free.”

14. Democratization and the changing legitimacy of the treaty

Both Brazil and Paraguay eventually emerged from military rule, opening more space for parliaments, civil society and the press to debate the bilateral arrangement. Claims about unequal revenue distribution and industrial opportunities gained greater public visibility. The binational framework survived, but its political justification had to respond to changed democratic expectations.

Paraguay's increasing electricity demand gradually reduced the previous imbalance between its treaty entitlement and domestic consumption. The opportunity cost of ceding surplus became more important as electricity-intensive businesses explored local investment. Brazil diversified its generation through wind, solar and other resources, although Itaipu remained crucial to certain regions and the wider grid.

A treaty signed for a very different industrial era may require adjustment, but renegotiation also brings risks. Stability depends on keeping operational coordination intact while addressing legitimate claims from both governments.

15. Lula and Fernando Lugo: the renegotiation of 2009–2011

In 2009, presidents Luiz Inácio Lula da Silva and Fernando Lugo reached an understanding about compensation for Paraguayan electricity not consumed domestically. Brazil's Congress approved the necessary measures in 2011, tripling a widely cited annual amount from about US$120 million to US$360 million under the arrangements discussed at that time. Both legislative records and Itaipu's official history document the decision. [16][17]

For Paraguay, the increase was an assertion of economic rights and a means of securing a greater share of benefits. For Brazil, it reflected an assessment that regional partnership and diplomatic stability were worth financial concessions. The distribution of resulting costs to final consumers was a significant part of the political debate.

Historical dollar amounts must be interpreted by year, legal basis and compensation category. They cannot be compared with a contemporary AI data center's electricity costs as though all represented the same type of payment. The broader strategic question concerns whether public and private energy revenues help build lasting production and welfare.

16. The 2023 fifty-year review of Annex C

The treaty reached its fiftieth anniversary of entry into force in August 2023, triggering the scheduled review of Annex C. Itaipu's official statements make clear that the governments, rather than the plant's technical staff alone, bear responsibility for political negotiation. The company provides operational and financial advice. [9]

The economic environment has changed materially since 1973. Brazil's system is more diverse, Paraguay's domestic market has expanded, the original debt profile has substantially run its course and digital technologies are creating new categories of demand. Negotiations cover compensation, royalties, service costs and future commercialization rules.

The treaty review should not be presented as an automatic unrestricted international auction of all Paraguayan energy. It remains subject to bilateral legal instruments, electricity-system requirements and the agreed distribution of rights.

17. The 2024–2026 tariff and competing national priorities

In 2024, Brazil and Paraguay agreed on a service-cost reference of US$19.28 per kilowatt-month for 2024, 2025 and 2026. Paraguayan authorities emphasized royalties, compensation and social investment resources. Brazilian authorities emphasized maintaining affordable electricity and developing a longer-term framework that could change commercialization from 2027 onward. [18][19]

Paraguay has a legitimate interest in extracting greater economic value from its electricity, especially when domestic firms can now use more of it. Brazil has a legitimate interest in cost-effective power for households and industry and in preserving dependable supplies. These priorities are not automatically compatible.

The treaty's unit cost is not a quoted rate for Crusoe's proposed computing campus. Calculating its actual cost would require the separate Paraguayan retail or contracted tariffs, transmission charges, loss factors, taxes, firm-capacity commitments and any incentives negotiated with the ANDE.

18. The March 2026 negotiations and unresolved contractual terms

On 19 March 2026, foreign ministers Mauro Vieira and Rubén Ramírez Lezcano met in Asunción and reported progress on Annex C. Paraguay's foreign ministry explicitly stated that it was premature to announce the future tariff before the negotiators completed their agreement. [20][21]

Political statements about lower prices or market opening cannot substitute for the eventual binding text. Assertions that a particular energy price will apply in 2027 must therefore distinguish a proposal, an interim rate and an approved final arrangement.

Prospective market reforms also interact with transmission rights, existing contracts and the technical roles of ANDE, ENBPar and grid operators. Additional flexibility will not erase engineering constraints or the need for predictable binational coordination.

19. The 2025 Brazilian espionage controversy

In April 2025, Paraguay recalled its ambassador and suspended Itaipu-related talks following Brazil's acknowledgment that Brazilian intelligence had conducted surveillance against Paraguayan officials between June 2022 and March 2023. Lula's administration said the operation had been stopped after the government learned of it. The reported objectives and methods became subjects of political and investigative scrutiny. [22]

This is a specific episode of documented interference by Brazilian actors against Paraguay, rather than an established US operation. It demonstrates that the value of energy-negotiation information can motivate actions that severely damage confidence between governments.

The episode does not prove that the 2026 American–Paraguayan computing agreement involved cyber espionage or treaty violations. Those are separate claims requiring their own evidence. The historical precedent is relevant to institutional safeguards and diplomacy, not an invitation to presume wrongdoing in unrelated transactions.

20. American influence: from historical context to documented 2026 involvement

The United States exercised substantial political and economic influence in Latin America during the twentieth century, including during the Cold War. Finance, corporate technology and regional diplomacy affected national development strategies. Its general role is relevant background to the history of Brazil and Paraguay.

Still, structural influence is not equivalent to proof that Washington drafted Itaipu's tariff provisions or ordered a specific energy decision in 1973. Such an allegation requires records about the particular transaction. The treaty itself is a bilateral instrument between Brazil and Paraguay.

The 2026 evidence is more concrete: a senior American diplomatic visit, a publicly announced computing partnership and Paraguay's entry into a US-backed strategic-technology declaration. Influence can be examined through investment, contracts and institutional commitments without inventing covert activities.

21. Christopher Landau's visit and investment diplomacy

US Deputy Secretary of State Christopher Landau visited Paraguay during 19–21 October 2026, accompanied by announcements concerning digital infrastructure, minerals and other economic initiatives. Paraguay's public news service reported the Crusoe–Araico project, along with discussions of uranium investment and the initiation of talks about civilian nuclear cooperation. [23]

The official description portrayed the American relationship as an alternative to financing attributed to geopolitical competitors. This is evidence of strategic framing by the parties, not proof that the offered financing is automatically advantageous. Actual economic benefits depend on interest rates, procurement, local suppliers, taxes and enforceable obligations.

Brazil was not a named partner in the private announcement. That fact warrants scrutiny of treaty compliance and the allocation of electricity, but does not by itself establish infringement of Brazil's share. Paraguay's national authority to develop domestic businesses and customers remains relevant.

22. The industrial meaning of the Crusoe–Araico project

Crusoe announced on 22 October 2026 its strategic partnership with Araico for AI computing infrastructure powered by Itaipu. It described a location near an electrical substation, services to US and Latin American customers and an expected initial operating phase in mid-2027. [4]

The company distinguished an expansion path toward 106 MW of IT power from a longer-term option of approximately 280 MW. These figures should never be presented as an installed or metered load as of the announcement date. IT power also differs from the site's total power requirement, which includes cooling, conversion and supporting equipment.

The location may provide employment, construction and engineering contracts in Paraguay, while the cloud operator may hold high-value software, customer relationships and processing technology abroad. The distribution of value is determined by corporate agreements, tax structures and the extent of local participation, not the nominal size of the electrical connection alone.

23. Pax Silica as a new form of energy geopolitics

On 23 October 2026, Paraguay became the twenty-sixth signatory to the Pax Silica Declaration at the US Department of State. Paraguay's foreign ministry described a shared commitment to leverage critical minerals and hydroelectric electricity for advanced AI infrastructure in the Americas. [5]

The agreement's policy language links electricity, semiconductors, digital hardware and advanced computing to national economic security. It suggests a strategic shift: renewable electricity from smaller states can support international AI infrastructure, making energy contracts part of technology diplomacy.

The signature does not change ownership of Itaipu or create a military alliance against Brazil. To assess its consequences, observers must examine subsequent financing, investment screening, procurement and digital-security commitments. A declaration establishes political direction; execution remains a separate evidentiary question.

24. Quantifying the hypothetical 280 MW electricity load

Power is measured in megawatts; annual energy is power multiplied by time. 280 MW operated continuously for 8,760 hours would consume 2,452,800 MWh, or 2.4528 TWh. A 106 MW block operated under the same assumption would consume 0.92856 TWh. These figures represent conditional arithmetic, not observed demand or a confirmed energy purchase agreement. [4]

A data center's Power Usage Effectiveness (PUE) compares whole-facility electricity with IT electricity. At a hypothetical PUE of 1.20, a 280 MW IT load would correspond to 336 MW of whole-facility demand and approximately 2.94336 TWh per year if sustained continuously. Actual load depends on installed hardware, utilization, efficiency and cooling conditions.

Relative to Itaipu's actual 2025 generation of approximately 72.88 TWh, the servers-only case equals 3.37%, and the hypothetical whole-campus case 4.04%. This comparison establishes an order of magnitude, not a claim that the investment will reduce Brazilian entitlement by that fraction.

25. When hydroelectric exports become imported digital services

Low-cost electricity can attract manufacturers or data centers. The benefits retained by the host country depend on employment quality, tax receipts, procurement, knowledge transfer and the location where resulting goods or services are sold. AI facilities may involve tremendous capital expenditures while requiring relatively few permanent employees per megawatt.

Paraguay could benefit through local construction, electrical engineering, logistics, technical training and domestic software providers. Alternatively, it could become primarily a location where imported computing equipment uses subsidized power to sell services billed elsewhere. Similar choices confront Brazil under its domestic data-center policies.

The relevant opportunity cost is the value of electricity under realistic alternative uses: what other industries or consumers could purchase it, what network investments are required and who receives the resulting gains? “Digital sovereignty” cannot be demonstrated merely by locating foreign-owned processors near a hydropower station.

26. ANDE, Paraguayan tariffs and the missing contract terms

Paraguay's National Electricity Administration (ANDE) is responsible for delivering electricity to the country's consumers under its statutory and regulatory framework. Its published tariff documents include special provisions for energy-intensive customers and advanced industrial activities. This shows that large computing loads need to be considered within established pricing and regulatory categories rather than assumed to obtain electricity directly and without conditions from the binational plant. [24]

The public announcement of the Crusoe–Araico partnership had not, by 25 October 2026, disclosed a complete electricity purchase contract, the final rate applicable to the site, the cost allocation of any necessary grid reinforcement, firm-capacity rights or operational penalties. Paraguayan journalists reported trying unsuccessfully to obtain clarification from ANDE about the contract. Such missing public information does not establish illegality, but it creates a material limitation for economic evaluation. [25]

A proper assessment needs approved interconnection studies, power and energy tariff categories, network investment responsibilities and demand forecasts. If the project receives electricity below its incremental system cost, others may absorb the difference. If it pays a cost-reflective tariff, finances the required infrastructure and stimulates new local capabilities, it may create net economic value. Neither conclusion can be reached from the corporate press release alone.

27. Subsidies, electricity pricing and transfers of economic value

The price of electricity includes generation as well as substations, protective equipment, high-voltage lines, reactive power management, operational reserves and losses. A major customer with an attractive energy-only price may nevertheless impose additional system costs. When those costs are socialized through household or other industrial tariffs, the project receives an implicit subsidy even if its electricity purchase agreement looks commercially attractive.

The opposite may also occur. A stable long-term customer can justify grid investment that would otherwise not be financed, improving overall infrastructure use. The value depends on location, seasonal dispatch and the power-system requirements for contingencies. Proximity to a large dam alone does not determine whether a connection is economically efficient.

Authorities should make enough information public to evaluate the relation between incremental costs and expected revenues while protecting legitimate commercial confidentiality. The allocation of profits and taxes also matters: electricity can be bought domestically to support digital services whose main contracts and intellectual-property rights remain in another jurisdiction.

28. Brazil's REDATA law and industrial response in 2026

Brazil enacted Law No. 15,504 on 15 September 2026, creating the Special Tax Regime for Data Center Services (REDATA). The law provides tax incentives for qualifying equipment and establishes conditions linked to sustainability, national markets and research investment. It represents an attempt to attract AI infrastructure while reducing dependence on foreign cloud services. [26][27]

According to Brazil's Finance Ministry, firms benefiting from the programme generally must allocate at least 10% of their computing and data-services capacity to the domestic market, invest 2% of the eligible purchase value in local research-related commitments, secure sufficient clean electricity and meet a cooling-water efficiency standard of 0.05 liter/kWh or less. Regional adaptations and implementing rules need to be considered before applying the thresholds to particular projects. [27]

The programme does not establish that Brazil will automatically attract investment over Paraguay. Net electricity costs, taxes, grid availability, land, submarine cable connections, latency, borrowing conditions and customer access all influence site selection. Comparative evaluation requires actual project contracts and rules rather than promotional statements by either government.

29. Forecast grid connection requests versus actual electricity use

EPE's research on the impact of large data centers reported a growing number of connection requests to Brazil's transmission system. One update covering February 2025 identified 36 requested connections totaling approximately 15.7 GW by 2037. The number reflects a pipeline of proposed demand, not completed and operating computing plants. Individual projects may be delayed, resized, cancelled or mutually exclusive in their access requirements. [28]

If 15.7 GW were hypothetically consumed continuously, it would equal roughly 137.5 TWh a year. That is an important scale relative to Brazil's 667.8 TWh final electricity consumption in 2025. Yet this is only a mathematical upper-bound illustration, not EPE's forecast of energy actually used. Differences between requested connection power, IT load, facility overhead and operating utilization are fundamental. [7][28]

Many requests concentrate near major urban and industrial centers, placing pressure on specific corridors and substations. Wind output that cannot be delivered from the Northeast does not automatically solve a transmission binding constraint around São Paulo. The economic effects of new data centers must be modeled geographically and hourly.

30. Why computing campuses are distinctive grid customers

AI training and cloud infrastructure require substantial electrical availability, but different workloads have different degrees of flexibility. Some batch computing may be shifted to periods when renewable output is abundant. Real-time analysis, customer-facing applications and latency-sensitive services often need continuous capacity. Treating all computing loads as freely interruptible would overstate their usefulness for balancing electricity systems.

A constant demand profile may help absorb otherwise underused generation in some hours, but can also increase pressure during drought, outages and peak periods. Energy-storage agreements, credible demand-response commitments and independently verified operational flexibility influence whether a new project increases or decreases the broader system's cost.

The facility itself needs redundancy, cooling, backup supply, optical networks, fire protection and cyber defenses. Power consumption is only one component of its environmental and economic footprint. Water use, thermal discharge, backup generator emissions and equipment replacement may impose locally important costs.

31. The worldwide competition for low-carbon electricity and AI

Advanced AI models, cloud servers and scientific computing systems require chips whose electrical demand depends on architecture, cooling design and workload utilization. Firms therefore seek locations with comparatively low-cost power, stable grids, good communications infrastructure and predictable contracts.

Power marketed as hydropower can provide an attractive emissions profile, but claims about marginal emissions should account for electricity that might otherwise serve another customer. In interconnected systems, a change in one customer's demand may cause different plants to dispatch at another hour or in another region. The label renewable does not by itself establish the complete system-wide effects of a new large load.

Locating servers in a country is also not equivalent to transferring the ability to design AI chips, operate foundation models or control customer data. True technological autonomy depends on corporate governance, technical training, encryption, service contracts and national legal protections.

32. Renewable curtailment and the limits of apparent electricity surpluses

Brazil has been experiencing curtailment of wind and solar generation when supply exceeds system requirements or network and reliability constraints prevent generation from being used. In October 2026, ANEEL approved updated rules for the sequence of mandatory reductions, while the national system operator continued work on measuring and communicating generation restrictions. [29][30]

Electricity not generated during a sunny midday cannot automatically be used during an evening peak. Additional daytime load can be helpful in some locations, but a computing campus requiring round-the-clock service will still need night-time and contingency supplies. Storage and well-designed load flexibility are therefore important if the project is intended to absorb renewable surpluses.

Curtailment illustrates why a country can have abundant renewable potential while maintaining material system costs. Transmission, voltage control, operational reserve, forecasting and demand management are necessary complements to new wind and solar capacity.

33. Paraná hydrology and the competing risks of drought and flooding

Itaipu's annual output depends on river inflows, upstream reservoir operations and physical limits on the use of water. Climate variability and land-use change influence the availability and timing of water. Hydropower stations cannot create a water resource; they transform the energy of the flow they receive within technical and environmental constraints.

On 23 October 2026, Itaipu announced a strategy to lower reservoir levels in anticipation of very heavy rainfall, including forecasts exceeding 300–350 mm in parts of the region. The purpose was to manage downstream releases and reduce the coincidence of flood peaks in the Paraná and Iguaçu rivers. This is an observed flood-management event, not an indication of sabotage or of a scarcity emergency. [31]

Electricity production and public safety cannot be separated. At times, a hydropower operator must release water to preserve structural and downstream safety even if some potential generation is forgone. Evaluating that decision only in terms of lost megawatt-hours would ignore human life, infrastructure damage and legal responsibilities.

34. Drought resilience and Brazil's diversified generating fleet

Extended dry periods can reduce hydraulic generation and increase the use of thermal electricity, generally at higher marginal costs. Wind and solar provide complementary resources and can conserve reservoir water during favorable periods, but their availability is variable. The optimal portfolio combines resource diversity, transmission links and appropriately chosen flexible capacity.

Batteries, demand response and improved forecasting can mitigate hourly imbalances. They do not replace seasonal water storage under every condition. Planning needs to examine synchronized risks: low river inflows, high temperatures, high demand and simultaneous outages.

The expansion of data centers increases the importance of such combined scenarios. A project that depends on the hydrological conditions of a good year without contingency planning may expose other consumers to unexpected pressure in a dry year. Long-term contracts should address volume uncertainty and firm-power commitments explicitly.

35. Transmission as an asset equal in strategic importance to generation

Itaipu's power reaches consumers through an interlinked transmission network and equipment that accommodates different frequency systems. Transformers, converter stations, substations, lines and protective devices are indispensable parts of the supply chain. A generating plant can remain operational while a failed transmission corridor prevents a significant share of its output from reaching customers.

Maintaining such networks requires skilled labor, replacement equipment, financing and often internationally sourced components. Transformer lead times and grid-automation dependencies can make recovery expensive. These vulnerabilities are distinct from whether Paraguay and Brazil agree on the latest Annex C tariff.

Resilience should be designed around credible contingencies and economic costs. Redundancy can be valuable, but indiscriminate duplication of all equipment may be prohibitively expensive. The relevant measures are restoration times, critical spares, maintenance backlogs and the ability to reroute electricity after a component fails.

36. Extreme weather and the transmission failures of 2026

In October 2026, Brazil's grid operator described major weather-related damage to transmission towers following severe storms. Media reporting cited more than 290 affected structures over preceding months, especially in southern Brazil, while the operator characterized actual disruption to served demand as relatively limited owing to system resilience. [32]

The episode distinguishes physical exposure from system failure. Losing a line may be survivable when alternate paths and grid protections operate correctly. It still creates repair costs and dangerous working conditions and can expose smaller communities to localized interruptions.

Energy-security evaluation should focus on documented storm risks, fires, maintenance problems, vandalism and technical failures while retaining appropriate procedures for investigating deliberate interference if evidence appears. Attributing every outage to external geopolitics would misallocate resources and obscure practical preventive measures.

37. Industrial cybersecurity at dams, substations and data centers

Modern electricity networks rely on supervisory control, telemetry, protective relays, digital communications and industrial software. These systems improve operational efficiency but require disciplined cyber risk management. A digital intrusion can disrupt availability or compromise records without necessarily damaging generation equipment physically.

For a binational installation, good practice includes tightly managed access, vendor assurance, segmentation between business and control networks, recovery capability and coordinated incident response across institutions. Public discussion should remain at the level of governance and resilience rather than expose operational vulnerabilities.

A data center near a substation does not automatically connect its servers to the hydroelectric plant's industrial-control network. Engineering and security rules should preserve separation between corporate computing traffic, electricity delivery and the plant's own control systems.

38. Brazilian defense institutions and critical-infrastructure security

Brazil's Armed Forces have constitutional responsibilities for territorial defense and may participate in contingency planning for strategic facilities. That does not transfer electricity tariff decisions or industrial licensing away from the competent civilian agencies. Routine safety also relies on utilities, police, fire response and regulators.

A major facility on a border requires Brazil and Paraguay to coordinate physical protection and responses to serious incidents. Military actions crossing the border without authorization would present major legal and diplomatic risks and are not a rational interpretation of existing cooperation.

The American corporate interest in an AI facility is not evidence of foreign troops mobilizing to seize Itaipu. Planning for exceptional military conflict can exist as a contingency, but the documented priorities include secure grid operations, physical access management and natural-hazard readiness.

39. Economic leverage and changing bargaining positions

A state or corporation can influence energy decisions through capital, technological knowledge, long-term contracts and access to major industrial customers. These channels may change choices without military force. In Paraguay, attracting high-value computing projects can raise the opportunity cost of transferring electricity to Brazil, potentially strengthening its position in future negotiations.

American firms may seek trusted, low-carbon computing locations for clients and supply chains. Paraguay can use that demand to promote development. Brazil may face changing electricity volumes and incentives while possessing alternative generation capacity and opportunities for its own data-center market.

The term interference should be used precisely. Ordinary commercial diplomacy is an exercise of influence; coercion or corruption requires additional evidence and legal analysis. A diplomatic announcement made without Brazilian participation is not itself proof of unlawful control over binational electricity.

40. The 2027 market-opening question

The Brazilian government's 2024 agreement with Paraguay contemplated a framework for more flexible marketing of unused Paraguayan electricity beginning in 2027. The practical conditions require negotiated Annex C provisions and subsequent implementing arrangements. New domestic Paraguayan customers could reduce the amount available for sale abroad if they use electricity within that legal framework. [19]

That possibility should not be mechanically translated into an equivalent loss of electricity for Brazil from the 280 MW press-release figure. The Brazilian grid has other suppliers, and the project has not reached the stage of measured full load. Bilateral allocation, contractual rights and physical delivery conditions must all be examined.

Available Brazilian policy responses include network reinforcement, diversified generating capacity, long-term energy contracts and constructive bilateral negotiation. Each choice involves costs, and its value depends on a realistic comparison rather than an assumption that the cheapest alternative will always be to demand another country's surplus.

41. Crypto mining as a separate energy-intensive comparator

Paraguay has also attracted interest from energy-intensive digital businesses such as cryptocurrency mining. These activities share high electrical consumption with some AI infrastructure, but their revenue sources, price sensitivity, technological needs and labor impacts differ. The existence of an earlier energy-intensive industry can inform regulatory debates without establishing that all future computing projects behave identically.

A demand for special rates can create risks of cross-subsidy where other customers pay for infrastructure. Conversely, reliable industrial demand may justify useful investment. The relevant questions concern incremental costs, contract stability and the capability to enforce payment and operating conditions.

The AI project announced in 2026 should be examined on its own merits: cloud customers, national service provision, research capacity, fiscal treatment and the quality of the technology transferred to the local economy.

42. Argentina, Uruguay and regional electricity cooperation

Brazil has other electricity exchange experiences with Argentina and Uruguay, alongside its Itaipu and broader South American connections. Regional exchanges can lower costs when hydrological, seasonal or time-of-day conditions differ. Their success depends on grid links, stable commercial rules and governments' willingness to honor agreements.

Argentina's role as a downstream state in the historical Itaipu dispute demonstrates the potential for conflict over shared resources, while the 1979 settlement shows that institutions can reconcile important differences. Uruguay offers a smaller-market example of energy integration with different generation and policy structures.

A resilient Brazilian strategy should avoid exclusive dependence on a single bilateral source or an assumption of permanently frictionless regional relations. Diversified exchanges and predictable settlement mechanisms reduce risk while preserving mutually beneficial trade.

43. The comparisons with Quebec and Norway

Jurisdictions with abundant hydropower and favorable climates have become attractive to large computing installations because electricity and cooling affect costs and emissions. Quebec and Norway illustrate broader patterns of competition for low-carbon electricity, industrial development and transmission capacity.

Neither case demonstrates that hydropower alone creates a leading AI industry. Semiconductor access, fiber connectivity, software engineering, capital and customer markets also determine where advanced services grow. A region can supply energy while capturing only a small fraction of the economic value.

Brazil and Paraguay have different scales of domestic demand and different energy pricing and tax institutions. The best project for one country is not necessarily the largest campus in terms of announced megawatts. Value added, fiscal return and technical capabilities should carry weight alongside power consumption.

44. Four possible development paths through 2040

Under a productive integration path, Paraguay attracts computing investments with cost-reflective electricity, local infrastructure and enforceable training commitments. Brazil negotiates Annex C provisions constructively, expands its own computing industry and secures alternative electricity capacity. The binational plant remains an asset supporting two productive economies.

Under a digital dependency path, Paraguayan electricity is allocated cheaply to operators whose software, cloud clients and large margins remain outside the country. Brazil responds with poorly evaluated incentives that transfer resources without creating meaningful local research or manufacturing capability.

Under a tariff confrontation path, disagreements over compensation, contracted power and network costs damage political confidence and delay projects. Under a physical shock path, drought, floods or transmission failures make reliability more important than planned computing expansion. These are conditional mechanisms, not assigned numerical forecasts.

45. A ranked set of exposures for Brazil

Hydrological exposure remains significant because water availability affects power output and dispatch flexibility. Transmission exposure concerns storms, long replacement times and network expansion. Financial exposure arises from tariffs, service charges and potential subsidies to large consumers. External contractual exposure involves ongoing Paraguay negotiations and the possibility of future domestic uses for energy formerly ceded.

Digital exposure includes dependence on foreign clouds, processors, industrial software and cybersecurity providers. Diplomatic exposure was illustrated by the dispute following the 2025 surveillance revelations. Environmental exposure includes floods, reservoir management and the effects of large infrastructure development.

An American military operation to capture Itaipu has not been demonstrated by the publicly available evidence used here. It should not be placed in the same risk category as the observable vulnerabilities simply because American firms and diplomats are active in the region.

46. Industrial alternatives for scarce firm electricity

Electricity can support aluminium, fertilizers, hydrogen, steel, chemical processing or digital services. The economic value of an additional unit of electricity depends on realistic alternatives, not a predetermined ideological preference for either traditional manufacturing or data centers.

Computing campuses often involve large capital investment with relatively few permanent jobs per megawatt. They can nevertheless support valuable engineering and software ecosystems where skills, local procurement and public research are developed. The assessment should consider all these effects without confusing construction employment with long-term operations.

Projects benefiting from public policy should be evaluated for use of national services, cost-reflective electricity tariffs, water performance and durable domestic capability. Brazil can set rules for its own incentives; Paraguay will make decisions within its institutions and treaty obligations. Cooperation can protect the interests of both better than a zero-sum contest.

47. Metrics that reveal rather than conceal performance

Itaipu monitoring should separately report total output, deliveries to each country, electricity ceded from Paraguay's legal share, contracted capacity, effective tariffs, royalty payments and operating costs. Every indicator needs a date range, a unit and a clear source. A fifty-fifty legal entitlement cannot be replaced with a single year's physical delivery ratio.

Data-center metrics should distinguish planned IT power, approved grid connection, installed facility capacity, electricity actually measured each month, observed PUE, water usage, tariffs, grid cost allocation and local-service provision. Crusoe–Araico's future 280 MW should remain a project option until construction, energization and commercial operations justify reporting realized capacity.

National electricity-security monitoring needs reservoirs, weather exposure, transmission outages, renewable curtailment, reserve margins, imported fuels and restoration times. Stable historical series are more useful for accountability than rapidly changing promises of megawatts without clear definitions.

48. What Brazil can legitimately negotiate with Paraguay

Brazil can demand that binational rights and operating procedures be respected, request transparency about effects on treaty commitments and negotiate reliable future deliveries. It can propose grid investments, long-term electricity exchanges and mutually beneficial changes to pricing. It cannot simply claim the unrestricted right to control Paraguay's electricity because a foreign-owned company might become a customer.

Joint research, integrated digital infrastructure and transparent benefit sharing may create opportunities for both governments. The aim should be to identify shared gains instead of treating every Paraguayan industrial project as a Brazilian loss.

The absence of Brazil from a business ceremony is not itself proof that an international treaty has been violated. Stronger evidence would involve incompatible contract terms, actual unauthorized reallocation or decisions beyond an institution's competence. The public documentation available in October 2026 does not by itself establish those findings.

49. Itaipu's historical legacy for the century of computing

Itaipu emerged from a combination of major engineering, state authority, industrial requirements and asymmetric regional power. It changed electricity supply in two countries, required accommodation with a third and left enduring environmental and political legacies. The 1973 treaty still determines rights in an environment transformed by cloud services and AI.

Paraguay has legitimate rights to develop industry using its electricity allocation within the agreed legal framework. Brazil has legitimate interests in affordability, dependable supply and maintaining a functioning binational entity. American technology firms seek energy for computing and strengthen their position through investment and diplomacy. These motives can complement or conflict with one another depending on actual contract terms.

Sovereignty is measured through functioning institutions and legal control, not through the loudness of geopolitical accusations. Reliable grids, fair tariffs, scientific capability and transparent negotiated rights give countries more durable autonomy than unsupported claims of foreign invasion.

50. Energy and digital power through 2040

The pace of electrification and AI expansion will depend on water conditions, network capacity, storage technologies, financing and the efficiency of computing systems. States that combine dependable electricity with engineering talent, manufacturing and data governance are more likely to capture a greater share of digital value.

Brazil's challenge is to exploit its clean-power advantage without undermining affordability or reliability for households and firms. Itaipu should remain an instrument of beneficial regional integration, adapted to growing demand and new economic uses. Foreign capital is neither inherently a threat nor a guarantee of prosperity; its effects depend on competitive alternatives, contracts and local learning.

Electricity policy increasingly determines where computing infrastructure and associated economic power will reside. Itaipu's future will matter not only for megawatt-hours, but for whether its two owners can convert a historic shared resource into long-term productive capacity and institutional security.

Quantitative and documentary comparison

Analytical visualizations

References and documents

[1] Brazil, Decree No. 72,707 of 28 August 1973, Treaty of Itaipu: https://www2.camara.leg.br/legin/fed/decret/1970-1979/decreto-72707-28-agosto-1973-421325-publicacaooriginal-1-pe.html

[2] Itaipu Binacional, 2025 generation and productivity, January 2026: https://www.itaipu.gov.py/noticias/energia/central-hidroelectrica-itaipu-registro-una-productividad-record-en-el-ano-2025

[3] Itaipu Binacional, 2025 electricity supplied to ANDE, January 2026: https://www.itaipu.gov.py/noticias/energia/itaipu-suministro-25-768-gwh-de-energia-electrica-a-paraguay-en-el-2025

[4] Crusoe, corporate statement on the Araico partnership, 22 October 2026: https://www.crusoe.ai/resources/newsroom/crusoe-araico-hydroelectric-ai-infrastructure-partnership

[5] Paraguayan Ministry of Foreign Relations, Pax Silica accession, 23 October 2026: https://www.mre.gov.py/paraguay-se-suma-a-la-declaracion-de-la-pax-silica/

[6] Paraguay state news service, Yguazú Digital and Crusoe–Araico, 23 October 2026: https://mediosdelestado.gov.py/ip/2026/10/23/pena-presenta-en-ee-uu-proyecto-yguazu-digital-para-atraer-inversiones-en-infraestructura-digital-y-centros-de-datos/

[7] EPE, National Energy Balance 2026, 2025 data: https://dashboard.epe.gov.br/apps/livro-ben/livro/pt/capitulo_1.html

[8] Brazilian National Congress, 1973 approval of the Itaipu treaty: https://www2.camara.leg.br/legin/fed/decleg/1970-1979/decretolegislativo-23-30-maio-1973-346403-publicacaooriginal-1-pl.html

[9] Itaipu Binacional, fifty-year anniversary and Annex C revision: https://www.itaipu.gov.py/noticias/institucional/quincuagesimo-aniversario-de-vigencia-del-tratado-de-itaipu-abre-paso-a-la-revision-del-anexo-c

[10] Itaipu Binacional, original Annex C text: https://www.itaipu.gov.py/sites/default/files/u123/AnexoC.pdf

[11] Luiz Eduardo Pinto Barros, historical study of Argentina and Itaipu, 1966–1979: https://periodicostestes.bce.unb.br/index.php/emtempos/article/view/37048

[12] Itaipu Binacional, institutional history and the 1979 Tripartite Agreement: https://www.itaipu.gov.py/institucional/sobre-itaipu/nuestra-historia

[13] Inter-American Development Bank, Monetary and Fiscal History of Brazil, 1960–2016: https://publications.iadb.org/publications/english/document/The_Monetary_and_Fiscal_History_of_Brazil_1960-2016_en_en.pdf

[14] Federal Reserve History, Latin American Debt Crisis: https://www.federalreservehistory.org/essays/latin-american-debt-crisis

[15] Columbia Center on Sustainable Investment, Leveraging Paraguay's Hydropower for Sustainable Economic Development: https://ccsi.columbia.edu/sites/ccsi.columbia.edu/files/content/docs/our%20focus/extractive%20industries/Leveraging-Paraguays-Hydropower-for-Economic-Development-Final-CCSI.pdf

[16] Itaipu Binacional, increased compensation for ceded energy, May 2011: https://www.itaipu.gov.py/noticias/sin-categorizar/brasil-aprobo-triplicar-compensacion-al-paraguay-por-energia-de-itaipu

[17] Brazilian Chamber of Deputies, April 2011 approval of the payment increase: https://www.camara.leg.br/noticias/212411-CAMARA-APROVA-AUMENTO-DO-VALOR-PAGO-AO-PARAGUAI-POR-ENERGIA-DE-ITAIPU

[18] Paraguayan Ministry of Industry and Commerce, 2024–2026 tariff agreement: https://www.mic.gov.py/paraguay-y-brasil-acuerdan-historica-tarifa-de-itaipu-y-anuncian-el-inicio-de-las-negociaciones-del-anexo-c/

[19] Brazilian Mines and Energy Ministry, bilateral tariff agreement and 2027 plans: https://www.gov.br/mme/pt-br/assuntos/noticias/brasil-e-paraguai-fecham-acordo-estrutural-para-tarifas-de-itaipu

[20] Paraguayan Foreign Ministry, March 2026 Annex C meeting: https://www.mre.gov.py/paraguay-y-brasil-avanzan-en-la-revision-del-anexo-c-de-itaipu/

[21] Itaipu Binacional, 19 March 2026 foreign ministers' statement: https://www.itaipu.gov.py/noticias/institucional/cancilleres-de-paraguay-y-brasil-destacan-avances-en-la-revision-del-anexo-c-del-tratado-de-itaipu

[22] Associated Press, Brazil–Paraguay espionage dispute, April 2025: https://apnews.com/article/24bd56c84442acfc08c5b6985df344f2

[23] Paraguayan government news service, Christopher Landau visit, 22 October 2026: https://mediosdelestado.gov.py/ip/2026/10/22/inversiones-en-tecnologia-y-nuevas-fuentes-de-energia-se-formalizaron-en-visita-del-subsecretario-de-ee-uu/

[24] ANDE, applicable electricity tariffs and intensive-use categories: https://www.ande.gov.py/tarifas_vigentes.php

[25] ABC Color, requests for clarification of the Araico–Crusoe electricity contract, 23 October 2026: https://www.abc.com.py/economia/2026/10/23/araico-y-crusoe-usaran-energia-de-itaipu-para-un-centro-de-ia-la-ande-no-informa-sobre-el-contrato/

[26] Brazil, Law No. 15,504 of 15 September 2026, REDATA: https://www.presidencia.gov.br/ccivil_03/_ato2023-2026/2026/lei/l15504.htm

[27] Brazil's Finance Ministry, REDATA obligations, 16 September 2026: https://www.gov.br/fazenda/pt-br/assuntos/noticias/2026/setembro/regime-especial-de-tributacao-para-servicos-de-datacenter-e-sancionado

[28] EPE, Impacts of large data-center expansion on Brazilian electricity demand, 2025: https://www.epe.gov.br/sites-pt/publicacoes-dados-abertos/publicacoes/PublicacoesArquivos/publicacao-458/topico-801/GPL-018.pdf

[29] Brazil's National Electric System Operator, renewable curtailment data: https://www.ons.org.br/Paginas/faq_curtailment.aspx

[30] ANEEL, updated mandatory generation-restriction rules, 6 October 2026: https://www.gov.br/aneel/pt-br/assuntos/noticias/2026-defeso-eleitoral/aneel-aprova-regra-para-organizar-cortes-de-geracao

[31] Itaipu Binacional, flood management during the October 2026 rain event: https://www.itaipu.gov.br/noticias/utilidade-publica/itaipu-adota-estrategia-para-conter-elevacao-do-rio-parana-diante-da-previsao-de-chuvas-intensas

[32] Folha de S.Paulo, extreme-weather damage to transmission equipment in October 2026: https://c-level.folha.uol.com.br/infraestrutura/2026/10/efeitos-do-el-nino-derrubam-mais-de-290-torres-de-energia-no-brasil-desde-julho.shtml

[33] Brazilian Chamber of Deputies, Mines and Energy Ministry information on Itaipu 2025 deliveries: https://www.camara.leg.br/proposicoesWeb/prop_mostrarintegra?codteor=3134139&filename=Tramitacao-18-RIC-352-2026