Table of Contents
HIVE operates data centers across Canada, Sweden, and Paraguay, supporting both digital asset mining and AI-related workloads
In sum – what to know:
GPU cluster live – HIVE activates its first AI cloud deployment in Paraguay, supporting LLM training workloads from Columbia University as part of a cross-border compute setup.
Proof of concept – The project will generate performance data to guide future HPC and AI infrastructure expansion, with scaling plans extending toward 2027.
Energy-driven location – Paraguay’s hydroelectric power and fibre connectivity underpin HIVE’s strategy to develop cost-efficient AI infrastructure in South America.
HIVE Digital Technologies has brought online a GPU-based AI cloud deployment in Asunción, Paraguay, marking the first operational cluster under its plan to expand high-performance computing (HPC) infrastructure in the country.
The system is part of HIVE’s BUZZ AI Cloud platform and is hosted in a Tier III data center operated by a local telecommunications provider. It is designed to support AI model training, inference, and other compute-intensive workloads.
The deployment forms part of a broader strategy to build AI and HPC capacity on top of HIVE’s existing renewable energy operations in Paraguay. The company currently operates a 300 MW hydroelectric-powered base in the country and is evaluating further expansion based on demand and available capital.
As an initial use case, a research team from Columbia University is running large language model (LLM) training workloads on the infrastructure from New York. The project is non-commercial and intended to generate performance data on latency, throughput and workload management across long-distance compute environments.
HIVE said the results will inform its plans to scale AI infrastructure in Paraguay, including potential future Tier III data center developments in Yguazú. The company is targeting additional capacity deployments through 2027, depending on market uptake.
The Columbia research team is working on LLM pre-training and optimization techniques, testing methods to improve model efficiency and reduce computational and memory requirements.
The project also serves as a proof of concept for cross-border AI workloads, with data processed on infrastructure located in South America while controlled from North America.
HIVE indicated that Paraguay’s energy availability and network infrastructure could support further development of AI and HPC capacity aimed at both regional and international users. Future expansion plans will depend on customer demand for AI cloud services and broader adoption of high-performance computing in the region.
“BUZZ is the first AI native cloud to be launched in Latin America. We see Latin America as one of the most compelling growth stories in global AI infrastructure over the next three to five years. The region sits at an inflection point rising demand for sovereign compute, an abundance of renewable energy resources, and a growing recognition among governments and enterprises that dependence on foreign cloud providers creates strategic risk,” Craig Tavares, president and COO at BUZZ HPC, told RCR Wireless News.
“Paraguay offers some of the lowest-cost clean hydroelectric power in the world, which directly addresses the two biggest constraints on AI infrastructure scaling: energy cost and sustainability. As AI workloads grow more power-intensive particularly with the shift toward large-scale training and inference on next-generation GPU architectures access to affordable, reliable, clean power becomes a decisive competitive advantage,” Tavares said.
The executive said the firm expects demand for AI infrastructure in Latin America to be driven by a combination of regulatory, market, and capacity factors over the coming years. Increasing data sovereignty requirements are pushing enterprises and governments to keep compute workloads within national or regional borders, supporting the case for localized infrastructure, said Tavares.
At the same time, the region’s technology ecosystem is expanding, with growing activity across AI startups, fintech, and agritech sectors driving demand for GPU-based compute. In parallel, capacity constraints in established hyperscale markets are prompting a shift of workloads toward new geographies that can offer sufficient power, connectivity, and stable operating conditions, he added.
“Our approach is to build sovereign AI factories while leveraging each region’s natural advantages. In Paraguay, that’s clean hydro power and proximity to other growth markets in Latin America,” the executive added.
The company operates data centers across Canada, Sweden and Paraguay, supporting both digital asset mining and AI-related workloads.
