Large Data Center Campus Planned in Pindamonhangaba, Brazil, with 150MW First Phase and R$5bn Investment
June 26, 2026
Large Data Center Campus Planned in Pindamonhangaba, Brazil, with 150MW First Phase and R$5bn Investment
A major hyperscale data center campus is being planned in Pindamonhangaba, São Paulo state, Brazil, marking one of the largest AI-focused infrastructure projects ever proposed in the country. The development, spearheaded by Riverhook Village 18, is expected to significantly bolster Brazil’s position as a growing hub for cloud computing and artificial intelligence workloads in Latin America.
The first phase of the campus will deliver 150 megawatts (MW) of critical IT capacity, with total investment reaching R$5 billion (approximately $1 billion USD). The project is designed to host hyperscale AI and machine learning operations, reflecting the surging demand for high-density computing power in the region. According to the developers, the site will leverage Brazil’s expanding renewable energy grid and favorable climate conditions to optimize energy efficiency and operational costs.
The announcement comes as global technology companies and data center operators increasingly turn to Latin America to diversify their infrastructure portfolios. Brazil, in particular, offers a combination of robust fiber connectivity, growing digital economy, and government incentives for large-scale tech investments. The Pindamonhangaba campus is expected to create thousands of construction and permanent jobs, while also attracting ancillary businesses such as network providers and equipment suppliers.
Industry analysts note that the scale of this project underscores a broader shift toward AI-optimized data centers that require significantly more power and cooling capacity than traditional facilities. The campus is being designed with advanced liquid cooling and modular architecture to support next-generation GPU clusters. As AI adoption accelerates across industries in Brazil and beyond, projects like this will be critical to meeting the computational demands of large language models and real-time analytics platforms.
Source: datacenterdynamics
