Brazil as a Sustainable AI Hub in the Global South: A Comparative International Law Analysis
By: Isabela Vasconcelos Chelou
Brazil is rich not only in culture and population but also in natural resources. It is known for its diverse biomes and is the Latin American country that contains the largest portion of the Amazon rainforest. Brazil has an abundance of powerful rivers and forest reserves that sustain the water cycle, as well as strong winds and high solar incidence across its continental territory. The country’s extensive climate agenda is part of the reason why Brazil will host COP30 this year. The event will take place in the city of Belém do Pará, currently one of the largest cities in the Amazon and commonly referred to as a gateway to the forest.
Precisely because it leverages many advantages in terms of natural resources, Brazil is one of the leading countries in producing clean energy. According to Brazil’s EPE, 88.2% of the country’s electricity in 2024 was renewable, with solar and wind making up almost a quarter of the total. Looking at low-carbon electricity, which includes hydropower, the country’s power grid is one of the world’s most sustainable, with 90% of its 2024 electricity coming from low-carbon sources, including 56% hydropower and 24% wind and solar, according to data released by Ember.
Due to this great potential to generate renewable energy, Brazil has been attracting an increasing number of Big Tech companies, like Amazon and Microsoft, seeking to invest in the construction of large artificial intelligence data centers. Moreover, such interest demonstrated by Big Tech companies aligns closely with Brazil’s goals at both governmental and private levels, particularly because investments from such companies stimulate technological innovation and position Brazil as a pioneer in Green AI in Latin America.
Nonetheless, to become a superpower in sustainable artificial intelligence, Brazil will also have to deal with challenges, including but not limited to the high energy and water consumption required by data centers, a significant risk of “data extractivism”, and unequal access to energy. Big Tech companies may enjoy privileged access to a continuous energy supply, while vulnerable populations face limited access to electricity.
This can be particularly complicated, considering that Brazil does not yet have specific legislation regulating AI data centers. Although the country has strong environmental licensing laws—most notably Law No. 6.938/1981 (National Environmental Policy)—which requires an environmental impact study as part of licensing for major projects, these laws were primarily designed to regulate factories, hydroelectric dams, mines, and large infrastructures, as these typically have significant and visible ecological consequences that trigger the need for an impact study.
However, the environmental damage left behind by digital infrastructures is often less visible. The high water and energy consumption demanded by AI data centers can have long-term negative effects that may not be immediately apparent and thus may not trigger the need for an environmental impact study. This regulatory blind spot can mean that a large data center could pass through with a less rigorous process. Therefore, because the AI data centers are a recent technological development, the existing Brazilian laws have proven incapable of properly targeting digital infrastructures.
This legislative gap has not gone unnoticed by the Brazilian legislature, which is already working on Bill No. 3018/2024 (Projeto de Lei nº 3018/2024) (the “Bill’), filed in August of 2024. Still in the early stages of proceedings in the Federal Senate, the Bill, in its proposed Article 5º, enumerates the measures that AI data centers need to follow to guarantee energy efficiency and environmental sustainability. This list includes the use of renewable energy sources, efficient cooling systems, and optimized hardware. It also requires that energy audits must be conducted periodically and that annual energy reports must be published. Another interesting measure proposed by the Bill is the development of management plans that include greenhouse gas emission reduction targets, as well as the promotion of recycling and proper disposal of electronic equipment.
If enacted into law, the proposed Bill would resemble the model adopted by the European Union, which already has a regulatory directive for AI data centers in effect. This is binding legislation, and Member States are required to transpose it into national law and ensure its enforcement. According to Article 12 of the Energy Efficiency Directive (Directive (EU) 2023/1791), large data centers that operate above 500 kW are required to report annually on their energy performance.
This annual report must fulfill the minimum requirements set out in Annex VII of the Energy Efficiency Directive. Aside from basic identification information regarding ownership, location, installed power, and data traffic, the directive also specifies, in Subsection (c) of Annex VII, that Key Performance Indicators (“KPI”) shall be monitored and published in observance of Article 12. It defines a non-exhaustive list of such indicators, including energy consumption, power utilization, temperature set points, waste heat utilization, water usage, and use of renewable energy. The first evaluation following this directive was published on July 8, 2025, showing that this is innovative legal ground in the European Union, as it is in Brazil.
Finally, Article 12 makes it clear that the reporting obligations are not aimed solely at monitoring data centers. The collected information may result, when appropriate, in proposals for new legislation to further strengthen energy efficiency, with the goal of transitioning toward a net-zero emission data center sector. In other words, the European Union will use the data reported annually as sustainability metrics to examine and study ways to continue improving. This feedback mechanism will, in turn, incentivize policymakers to prioritize clean energy sources and reduce reliance on fossil fuels.
Because it is still in early stages, the Bill has yet to develop its reporting obligations, enforcement authority, and penalties. Additionally, Brazil needs to define its operational threshold and establish which AI data centers will fall under its scope. The European Union could serve as inspiration, and considering that Brazil is hosting COP30, it is the perfect opportunity to learn from other countries’ existing AI regulation models and use their strengths and shortcomings as guides when refining Bill No. 3018/2024—potentially becoming a reference for the next generation of sustainable technology law.
However, the Brazilian legislature is not the only one that should be drawing on outside references; the legislative gap is even deeper in the United States. With no binding federal regulation on energy efficiency standards or reporting obligations for AI data centers, the U.S. mainly relies on a self-regulatory approach. This means that large tech companies establish and enforce their own energy consumption standards, which end up being cost-driven rather than sustainability-focused. The Environmental Protection Agency (“EPA”) launched the ENERGY STAR program—a voluntary certification for U.S. data centers—in an attempt to promote better energy-use practices among Big Tech companies.
Yet companies are not required to participate in this program, despite the EPA’s recognition that “data center energy use was predicted to grow exponentially as digital computing expanded. However, gains in energy efficiency—both at the network hardware and storage level, as well as at the building level—have mostly offset this growth.”
Despite having two proposed bills to address the environmental impacts of AI data centers, notably—the Artificial Intelligence Environmental Impacts Act of 2024 (S.3732/H.R.7197) and the Department of Energy AI Act (S.4664)—the U.S. Congress has not yet introduced any bills mandating energy efficiency standards or reporting and monitoring duties.
This lack of incentive appears to stem from the nation’s continued reliance on fossil fuels. According to the U.S. Energy Information Administration, in 2024, only 26% of the country’s electricity came from renewable sources, while 54% came from fossil fuel sources and 18.6% from nuclear power.
The comparison between Brazil, the EU, and the U.S. highlights the necessity for a more harmonized global approach. Artificial intelligence is transnational, as it is embedded in more areas of society every day. Big Tech’s investments, emissions, and energy use go beyond jurisdictions; therefore, COP30 is an opportunity for recognition that global alignment is essential.
COP30 presents itself as the ideal moment to influence States’ behaviors through the Paris Agreement. Under Articles 6, 9, and 10, the framework could be used to encourage States to adopt incentives such as tax breaks, carbon market benefits, and subsidies to make investment in renewable energy AI data centers more attractive for tech companies.
Linking AI data centers to renewable energy incentives within the financial mechanisms provided by the UNFCCC framework could shape corporate investment decisions, steering Big Tech toward a more sustainable path and discouraging dependence on fossil fuels.
In this context, as COP30 approaches, Brazil should take the lead and establish itself as a trendsetter among the global south, bringing together AI growth and climate responsibility as pillars for an advanced and sustainable future for all, demonstrating how its renewable energy advantage can inspire a new global standard.
This article was written by PECC's Energy and Climate Law Scholar Isabela Vasconcelos Chelou, a Brazilian lawyer pursuing dual LL.M. degrees in comparative legal studies and environmental law at Elisabeth Haub School of Law at Pace University. It was originally published on November 10, 2025, in Volume 1, Issue 2 of the R.E.A.C.T. by PECC Newsletter.
Editors: Mercè Martí I Exposito, Frances Gothard, Carington Lowe