HIVE launches Paraguay cloud cluster for Columbia University research

HIVE Digital Technologies has activated its BUZZ Artificial Intelligence Cloud platform in Asunción, Paraguay, with Columbia University researchers using the system for large language model training. The deployment is positioned as a proof of concept for scaling high-performance computing capacity in Paraguay.

HIVE Digital Technologies has brought its BUZZ Artificial Intelligence Cloud platform online in Asunción, Paraguay, with live GPU compute nodes now serving workloads from an academic research team at Columbia University in New York. The deployment is the first GPU cluster to go live under the company’s phased plan to add Artificial Intelligence and high-performance computing infrastructure to its renewable energy operations in Paraguay. HIVE is using the project to test whether Paraguay can support low-latency, high-availability GPU cloud services for customers outside the country.

The Columbia University team is using the BUZZ Cloud infrastructure for large language model pre-training and end-to-end foundation model training. The work focuses on optimization algorithms designed to improve model quality while reducing computational and memory costs, with performance evaluated through training metrics such as loss and perplexity as well as downstream benchmarks. Their work begins with small- to medium-scale models (0.2B to 2B parameters, including GPT-2-class and LLaMA-style architectures) and is scaling to larger models (8B+ parameters) using multi-GPU distributed training frameworks. In early experiments, Muon has shown roughly 1.3x greater efficiency1 than standard baselines by exploiting the structure of model weights.

HIVE said the live research workloads will provide operational data on latency, throughput, and workload management between New York and Asunción. The company plans to use that information to shape a broader expansion strategy, with initial deployment targets through 2027. It expects the current cluster to serve as a proof of concept before further Tier III data center development in Yguazú, where additional infrastructure upgrades would be needed to support larger-scale GPU cloud services.

The company argues Paraguay is well suited for this buildout because of abundant hydroelectric power and fiber connectivity. HIVE’s existing 300-megawatt (“MW”) renewable power base, sourced from hydroelectric generation, combined with the telecom partner’s enterprise-grade network infrastructure, creates a platform that can serve demanding workloads originating outside Paraguay’s borders, including from North American institutional clients. Frank Holmes said HIVE has 300 MW of renewable hydroelectric power operational in Paraguay, with another 100 MW in development. Management also pointed to Paraguay’s recent economic growth, stable governance, and openness to foreign infrastructure investment as supportive conditions for future expansion.

50

Impact Score

SK Group warns DRAM shortages could curb memory use

SK Group chairman Chey Tae-won warned that customers may reduce memory consumption through infrastructure and software optimization if DRAM suppliers fail to raise output. Demand from Artificial Intelligence data centers is keeping the market tight as memory makers weigh expansion against the long timelines for new fabs.

BitUnlocker bypasses TPM-only Windows 11 BitLocker

Intrinsec disclosed BitUnlocker, a downgrade attack that can bypass TPM-only Windows 11 BitLocker protections with physical access to a machine. The technique abuses a flaw in Windows recovery and deployment components and relies on older trusted boot code.

Micron samples 256 GB DDR5 9200 MT/s RDIMM server modules

Micron has begun sampling 256 GB DDR5 RDIMM server modules built on its 1-gamma technology to key ecosystem partners. The company positions the new modules as a higher-speed, more power-efficient option for scaling next-generation Artificial Intelligence and HPC infrastructure.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.