On October 27, 2026, UFMG officially inaugurated Soberano, a new supercomputer acquired as part of the GeoIDE project: Geoinformation Extraction Using Artificial Intelligence for Dual-Use Applications. The project is being developed in partnership with the Brazilian Army, through its Directorate of Geographic Service (DSG), and the Military Institute of Engineering (IME). The supercomputer will be used to train advanced AI models on Brazilian soil, process satellite imagery on a continental scale, and perform high-resolution climate and territorial simulations. These capabilities are expected to support the monitoring of compliance with environmental legislation and the development of national defense strategies. Read the article published by UFMG to learn more.

How can ocean warming change marine life in the Americas? That was the question a study led by a CSR researcher sought to answer using machine learning. The results show that, in a very high-emissions scenario, up to 69% of the studied area could experience a reduction of more than 5% in species richness by the year 2100. Among the most affected areas are Florida, the Gulf of Mexico, the Caribbean, northern South America, and southern Brazil. Read the full paper here.

Researchers at CSR have developed a machine-learning method to isolate the actual effect of the environment on biodiversity from the effect of insufficient data collected in certain areas. The computational model is capable of estimating species richness as if all locations had been surveyed with the same intensity. The Amazon emerges as the most biodiverse region on the planet in nearly all the groups analyzed, while Brazil’s Cerrado stands out as the richest tropical savanna among all those studied. The study’s findings were published in the journal Frontiers in Ecology and Evolution.