- within Accounting and Audit, Food, Drugs, Healthcare and Life Sciences topic(s)
In early September, the Minister of the Ministry of Management and Innovation in Public Services (MGI) reaffirmed the Federal Government's previous announcement regarding the allocation of BRL 23 billion over four years for initiatives aimed at the development of artificial intelligence. In her remarks, the Minister emphasized key points of the AI for the Good of All Plan, an initiative under the Brazilian Artificial Intelligence Plan (PBIA), structured into five pillars and 54 actions. The initiative reflects a context in which artificial intelligence is no longer merely a driver of technological innovation, but also a matter of digital sovereignty, compliance with the LGPD, and protection of the State's strategic information.
Among the PBIA's most relevant projects is the Sovereign Cloud, through which the government intends to migrate confidential public administration data to infrastructure controlled by Serpro and Dataprev. The proposal is not technological isolation, but strengthened governance: it aims to ensure that critical information circulates within a national environment with encryption keys managed by the State, guaranteeing privacy and availability in accordance with the LGPD. Investments exceeding BRL 1 billion are projected for this pillar through 2027.
Another highlight is the proposal to develop an AI supercomputer designed to position Brazil among the world's top five nations in processing capacity. The upgrade of Santos Dumont — housed at the National Laboratory for Scientific Computing (LNCC) — is expected to receive BRL 1.8 billion in funding by 2028 and could transform the national research ecosystem by enabling large-scale models and critical applications. This effort will be complemented by investments in regional National High-Performance Computing Centers (CENAPADs) (BRL 125 million) and high-speed connection networks (BRL 52.4 million). The expansion, however, will still face Brazil's structural dependence on imported hardware, a critical bottleneck for full sovereignty.
In the business domain, the PBIA includes the development of domestic datacenters powered by renewable energy (BRL 2.3 billion), with priority to the North and Northeast regions, as well as systemic support for the AI value chain (BRL 667 million) and the creation of an AI startup fund (BRL 400 million). These measures aim to establish a robust production chain capable of reducing bottlenecks and fostering global competitiveness for Brazilian companies.
On the software front, noteworthy mention goes to SoberanIA, a Portuguese-language large language model trained within Brazil using local infrastructure and curated datasets, such as the Jabuticaba Dataset. This model not only facilitates public-service applications — citizen assistance, school management, health, environmental monitoring — but also enables potential integration with private solutions.
The advantage of domestic LLMs rests on three factors: they allow deployment in sovereign clouds or on-premise environments, enable fine-tuning with national sector-specific data (legal, administrative, health, etc.), and reduce regulatory friction by natively adhering to the LGPD, eliminating the need for international data-transfer mechanisms. This creates space for a local ecosystem of providers capable of meeting specific demands from government and business without reliance on broad external platforms. The potential of such models, however, depends on the availability of large volumes of high-quality national data for training, a technical and governance challenge still to be overcome.
In practical terms, the pursuit of digital sovereignty in Brazil should not be confused with the notion of technological self-sufficiency, which remains unrealistic in the short and medium term. What emerges is an agenda of relative autonomy, in which the country seeks to reduce critical vulnerabilities without isolating itself from global innovation networks. Along this path, Brazil observes and absorbs elements from different international models: the economic dynamism of the United States, China's selective state control, and the European Union's regulatory orientation. The intention is to shape an original hybrid approach that balances economic development, control over critical information, data protection, and the appreciation of cultural diversity.
However, the goals announced by the government involve significant challenges that go beyond integration. The first is execution: governing such a complex plan requires inter-ministerial coordination and continuity, historically difficult in long-term Brazilian initiatives. The second is human capital, as the country must not only train but retain AI talent while competing with a heated global market that fuels brain drain. A third critical point is hardware dependency: data and software sovereignty rests on supercomputer and datacenter infrastructure whose core technologie s— such as semiconductors and GPUs — are fully imported, creating strategic vulnerability.
Therefore, the success of the PBIA will depend on overcoming substantial challenges. Beyond the need for integration across its pillars, success will hinge on resolving historical bottlenecks, such as the State's execution capacity, the development and retention of talent in a competitive global environment, and the ongoing dependence on imported hardware. Additionally, establishing a legal framework for AI that fosters innovation with legal certainty will be decisive. If Brazil manages to navigate these barriers, it may indeed transform artificial intelligence into a vector for economic development, social inclusion, and institutional strengthening.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
