Artificial intelligence is no longer a laboratory topic. It has become central to corporate strategies, economic policies, and geopolitical disputes. For the Brazilian electricity sector, this change is not just a distant global trend—it is a window of opportunity and risk that demands immediate action.
The Special Report “Energy and AI”, published by the International Energy Agency (IEA) in 2025, presents the first comprehensive global analysis of the intersection between energy and AI. This article translates the main findings into the reality and vocabulary of the Brazilian electricity sector, focusing on data, projections, and practical implications.
“There is no AI without energy — specifically, without electricity. At the same time, AI has the potential to transform the future of the energy sector.” — Fatih Birol, Executive Director of the IEA.
The explosion in demand: data and projections.
Where are we today?
In 2024, data centers consumed approximately 415 TWh of electricity—about 1,5% of global electricity consumption. This number has grown 12% annually since 2017, a rate four times higher than the total growth in global electricity demand.
For comparison, a modern AI-focused data center consumes as much electricity as 100.000 homes. The largest complexes currently under construction are equivalent to 2 million homes.
Geographically, the concentration is striking: almost half of the US data center capacity is in just five regional clusters. In Ireland, data centers account for 20% of the measured electricity supply. In the US state of Virginia, that figure reaches 25%.
Where are we going: scenarios up to 2035
The IEA has constructed four scenarios for the evolution of electricity consumption by data centers. The Base Case projects more than double the current consumption by 2030 — exceeding the total consumption of Japan today.
The spectrum of uncertainty is significant: by 2035, the range will be from 700 to 1.720 TWh, depending on the rate of AI adoption and advances in efficiency.
Strategic insight: Even in the most conservative scenario (Headwinds), electricity consumption in data centers will grow by 60% by 2030. For electricity sector planning, this floor already represents a significant challenge for generation and transmission expansion.
Which generation mix will meet this demand?
The IEA indicates that no single source will be sufficient on its own. The winning combination involves renewables leading the way, natural gas as a dispatchable backup, and nuclear entering progressively — including the first Small Modular Reactors (SMRs) from 2030 onwards.
Renewables lead the way due to their availability, economic competitiveness, and short deployment times. Technology companies are already structuring long-term Power Purchase Agreements (PPAs) directly with renewable energy generators—a model that the Brazilian market is beginning to see through negotiations with large consumers in the ACL (Free Contracting Environment).
Natural gas plays a crucial role in dispatchability — AI requires uninterrupted power supply (99,99% uptime), which variable renewables alone cannot guarantee. The delivery time for gas turbines already faces a backlog of several years, which is a warning sign for system planners.
The bottleneck in the networks: the weakest link.
The IEA report identifies transmission and distribution networks as the main bottleneck in the cycle. It is estimated that, without corrective action, approximately 20% of planned data center projects globally will be at risk of delay due to lack of network connectivity.
Key point for operators and planners: The IEA estimates that applying AI to network management could unlock up to 175 GW of existing transmission capacity globally—without building a single new line. This surpasses all projected data center load growth through 2030 in the Base Case.
For Brazil, with the National Interconnected System (SIN) operating near its limits in critical corridors such as the North-Southeast, the application of AI in transmission management is not a distant future: it is a present necessity. The report indicates that AI-based fault detection systems reduce the duration of outages by 30 to 50%.
AI applied to the electricity sector: concrete opportunities
Power System Operation
The IEA maps several high-impact applications already in use by system operators around the world. Predicting variable source generation with AI reduces curtailment and optimizes dispatch. Machine learning algorithms applied to PMU (Phasor Measurement Unit) data improve real-time stability.
Generation and storage
For power plants, AI is already delivering significant gains: turbine performance optimization, predictive fault diagnosis in generators and transformers, and intelligent maintenance scheduling.
In the battery sector—critical for establishing renewable energy generation—AI is accelerating the discovery of new materials and chemistries, potentially reducing the development time of next-generation technologies by decades.
Technological Innovation Accelerated by AI
One of the most striking findings of the report is the potential of AI to compress innovation cycles in energy. In biomedicine, AI has led to a 45.000-fold acceleration in mapping protein structures. In energy, the potential is similar:
A revealing fact: Only 2% of the capital raised by energy sector startups went to companies with a value proposition based on AI. There is a huge asymmetry between the potential and the current investment in AI for energy.
Energy Security in the Age of AI
New Risks
The IEA raises concerns that go beyond electricity demand. The supply chain for data centers is highly concentrated and globalized. A key mineral is gallium — an essential input in next-generation, high-efficiency chips.
China holds approximately 99% of the world's refined gallium supply, and IEA estimates indicate that data center demand for gallium could exceed 10% of current supply by 2030.
In the cyber realm, attacks on energy utilities have tripled in the last four years and have become more sophisticated with the use of AI by attackers. The Brazilian electricity sector, with its increasing digitalization via smart grids and smart metering, is directly exposed to this trend.
AI as a Security Solution
The same report that identifies the risks presents solutions. Satellites equipped with AI and smart sensors detect incidents in critical energy infrastructure 500 times faster than traditional ground-based methods.
AI-based predictive analytics is being used by operators to identify vulnerabilities before they become incidents. For the Brazilian National System Operator (ONS) and utility companies, this is a strategic investment area with measurable ROI.
Brazil in a global context: opportunities and risks
The report positions emerging economies at a crossroads. Developing economies (excluding China) account for 50% of the world's internet users, but less than 10% of global data center capacity.
Brazil presents unique characteristics in this context:
The IEA is explicit: countries with a reliable and affordable energy supply will be better positioned to attract data center growth and capture the benefits of AI development. Brazil's renewable energy matrix is a real competitive advantage—but it needs to be combined with resilient grids and agile regulatory processes.
The recent expansion of data center projects in Brazil by companies like Google, Amazon, and Microsoft is no coincidence: the combination of abundant renewable energy, a favorable time zone for Latin America, and a significant domestic market creates a strong value proposition.
The challenge for regulators and planners is to ensure that electricity infrastructure does not become the limiting factor for this growth.
The emissions debate: neither catastrophe nor magic solution.
The IEA report is balanced on the climate impact of AI. Emissions associated with data center consumption are projected to grow from 180 million tons of CO2 equivalent today to approximately 300 Mt in 2035 in the Base Case—remaining below 1,5% of total energy sector emissions. This is significant, but not catastrophic.
On the other hand, the widespread adoption of applications of IA Energy optimization could generate emission reductions equivalent to about 5% of global energy emissions by 2035.
To put this in context: the electricity savings in buildings alone, if all currently available AI solutions were scaled up, would be equivalent to the annual electricity generation of Australia and New Zealand combined — approximately 300 TWh.
AI can be a tool for reducing emissions, but it's not a silver bullet. Rebound effects—such as a shift from public transport to autonomous vehicles—can offset some of the gains. Proactive policies remain essential.
Action agenda for the Brazilian electricity sector
Based on the IEA findings, it is possible to structure a priority agenda for operators, concessionaires, regulators, and investors in the national electricity sector:
The closing thought
The central message of the IEA report is unequivocal: technology and the energy sector are becoming inseparable. AI needs energy to exist; the energy sector needs AI to survive the increasing complexity of grids, the variability of renewables, and the pressure for efficiency.
For Brazil, the equation has favorable components rarely found in other markets: a predominantly renewable energy matrix, an already significant consumer market for AI, and a strategic geographic location.
Converting these advantages into real data center growth, smarter networks, and more efficient businesses depends on coordinated action between regulators, operators, investors, and the government.
There are many uncertainties — from the pace of AI adoption to the efficiency trajectory of the models. But one thing is certain: waiting to see is not a strategic option for a sector that operates with assets that have a lifespan of decades and long planning cycles.
Reference
International Energy Agency (IEA). “Energy and AI” – World Energy Outlook Special Report. Paris: IEA, 2025. Available at: www.iea.org
all the content of Canal Solar is protected by copyright law, and partial or total reproduction of this site in any medium is expressly prohibited. If you are interested in collaborating or reusing part of our material, please contact us by email: redacao@canalsolar.com.br.
