The expansion of data centers in Brazil requires planning, transparency, and regulations that ensure benefits for the country and reduce socio-environmental impacts. The conclusion was drawn by researchers during a debate held in Niterói, in Rio’s metropolitan region, on Monday (Jul. 27) as part of the 78th Annual Meeting of the Brazilian Society for the Advancement of Science.

Professors Mauro Oliveira, from the Federal Institute of Ceará, and Dan Levy, from the Federal University of São Paulo, argued that the installation of large data processing centers must be accompanied by public policies and greater involvement from the scientific community.

Notícias relacionadas:

• Justice says Brazil’s sovereignty must prevail over US threats.
• AI used in 6 out of 10 federal and state government agencies.

In Professor Oliveira’s view, Brazil runs the risk of repeating a historical pattern of exploiting natural resources without securing benefits for national development.

“The problem isn’t hosting data centers. The problem is accepting these investments without negotiation. Brazil needs to define what it wants to demand in return,” he said.

The discussion about data centers, he argues, is still limited to a few sectors and should involve universities, the government, and society. Professor Oliveira believes these projects pose challenges related to energy consumption, digital sovereignty, and control over technological infrastructure.

He cited the project to build a large AI data center in Ceará state. The facility, he noted, is expected to consume about 300 megawatts of power – a demand equivalent to the electricity supply capacity for millions of people.

“The world today cannot produce enough energy to meet the projected demand from AI data centers,” he stated.

In addition to energy, Professor Oliveira argued that Brazil should establish permanent mechanisms for monitoring, auditing, and ensuring transparency regarding these projects.

“The role of universities is to participate in this debate. We are not against data centers. We are against the absence of universities in the discussion about the conditions under which they are being implemented,” he said.

Impact

From an environmental law perspective, Professor Dan Levy argued that the impacts caused by data centers go beyond water and energy consumption. To his judgment, effects such as noise pollution, pressure on natural resources, territorial conflicts, social inequalities, and risks to vulnerable communities must also be considered.

“The digital transition and the ecological transition cannot be treated as separate processes. We must incorporate criteria of sustainability, transparency, and social participation,” he stated.

The communities directly affected need to participate in decisions related to the implementation of these projects and not just bear the brunt of their impacts, he went on to say.

He also linked the expansion of data centers to the concept of digital colonialism, whereby countries in the Global South provide natural resources – such as energy, water, and land – to sustain the technological infrastructure of large corporations in the Global North.

“We’re not against development. The issue is how this development is carried out and who bears its costs,” he concluded saying.

Artificial intelligence (AI) can already be found in 59 percent of Brazilian federal and state agencies, as per data from an official report on 2025 released Thursday (Jul. 23).

The Judiciary leads in the adoption of the technology, reaching 94 percent of agencies in 2025, followed by the Prosecution Service (92%), the Legislative (83%), and the Executive (55%).

Notícias relacionadas:

• Hate, the Big Tech, and the far right: How the gears of misogyny turn.

According to the study, among the different levels of government, the use of AI is more widespread in federal agencies (70%) than in state agencies (58%). Other advanced technologies are less widespread – the internet of things (IoT) is used in 20 percent of federal agencies and 29 percent of state agencies, while blockchain was cited by 25 percent of federal agencies and 17 percent of state agencies.

Among public agencies that use AI, the most common application is the automation of workflow processes, adopted by 39 percent of institutions (45 percent at the federal level and 38 percent at the state level).

Next are text mining and natural language processing, used by 35 percent of agencies (46 percent federally and 34 percent at the state level). Machine learning tools for prediction and data analysis follow, used by 31 percent (39 percent at the federal level and 30% at the state level).

Municipal governments

The study mapped the use of AI in Brazilian municipal governments and found that 27 percent of them had used the technology in the 12 months prior to the survey. The data reveal significant disparities based on the size of the municipality – adoption reaches 85 percent in cities with more than 500,000 residents and 81 percent in state capitals, while in municipalities with up to 10,000 residents, the percentage is 22 percent.

This pattern is also observed in the use of other digital technologies, which tends to be higher among municipalities of different sizes.

In municipal administrations, the most common AI applications are process automation (14%) and text mining (12%). City governments reported using these technologies primarily to improve services provided to the public (17%), streamline internal operations (17%), support public policies and core activities (14%), and monitor the use of public resources (10%).

According to survey manager Alexandre Barbosa, the results show that AI is already a reality for a significant portion of the Brazilian public sector.

“At the same time, [the data] highlight that expanding training for public servants is a fundamental step toward ensuring these technologies are incorporated in an increasingly effective manner and contribute to the actual improvement of public services provided to the people,” he stated.

Generative AI

The survey presents a broad overview of the use of generative AI in the Brazilian public sector. At the federal level, 65 percent of public agencies have initiatives to use this technology in internal processes, while 29 percent use it in services provided directly to citizens.

With regard to supporting and preparing teams to use generative AI, 59 percent offer courses or training, 46 percent have acquired specific tools for use in work processes, and 43 percent have established guidelines or user manuals for employees.

“The lack of training emerges as the main barrier to AI adoption in Brazilian public organizations. This challenge was cited by 19 percent of federal agencies, 21 percent of state agencies, and 40 percent of municipal governments,” the survey says.

In municipal administrations, managers also cited as obstacles the fact that technology is not considered a government priority (36%), a lack of need or interest (35%), difficulties related to data availability and quality (35%), and high costs (34%).

Instant messaging tools have established themselves as key channels for serving the public. In city governments, the use of WhatsApp or Telegram for requesting public services increased from 56 percent in 2023 to 64 percent in 2025, making it the second most widely used contact channel by municipalities, behind only the conventional telephone (83%).

At other levels of government, 39 percent of federal public organizations and 43 percent of state public organizations offered mobile messaging capabilities in 2025.