Hive Live GPU Cluster Powers Columbia University LLM Research on 300 MW Renewable Platform
Hive’s BUZZ AI Cloud in Asunción operates live GPU nodes running Columbia University LLM training workloads from New York, marking its first AI cluster deployment on a 300 MW renewable energy footprint. The proof-of-concept will guide planned Tier-III data center expansions in Yguazú with high-availability, low-latency GPU compute capacity by 2027.
1. BUZZ Cloud Launch in Asunción
Hive’s BUZZ AI Cloud platform in Asunción, Paraguay is now operational with live GPU compute nodes hosted in a Tier-III data center run by the country’s largest telecommunications provider, delivering high-performance AI model training and inference workloads.
2. Columbia University LLM Research Engagement
A Columbia University research team in New York is using the GPU infrastructure to pre-train large language models from 0.2B to over 8B parameters, generating real-world performance data on latency, throughput and workload management; early tests show the Muon algorithm improving efficiency by roughly 1.3x.
3. HPC Expansion Roadmap to Yguazú Through 2027
The live deployment serves as a proof-of-concept to inform planned Tier-III data center expansions in Yguazú, leveraging Hive’s existing 300 MW hydroelectric power base and Paraguay’s fiber backbone to provide low-latency, high-availability GPU cloud compute for commercial AI demand.