HIVE Digital Technologies (TSX-V:HIVE, NASDAQ:HIVE) has announced the expansion of its NVIDIA GPU infrastructure through a $30 million investment, which will see the deployment of two advanced GPU clusters in Quebec.
The company said the new supercomputing clusters will expand revenue opportunities and enhance its ability to deliver AI solutions, including AI model training, real-time analytics, and generative AI services.
The first of the two clusters, the NVIDIA H100, is comprised of 248 GPUs arranged in 32 nodes interconnected with Infiniband.
This cluster has already arrived in Quebec, has been fully configured, and is expected to be operational before the end of 2024.
HIVE projects that this deployment will generate approximately $15 million in annualized run-rate revenue for its HPC business in the following quarter.
The second cluster, the NVIDIA H200, consists of 508 GPUs organized in 64 nodes, also connected with Infiniband.
It has been shipped and is expected to arrive in early January 2025. Deployment of the H200 cluster is scheduled for the first quarter of 2025, with anticipated revenues exceeding $20 million annually by the second quarter of 2025.
"The deployment of our NVIDIA H100 and H200 GPU clusters represents a key progressive step in our HPC strategy and a notable evolution in our business model, as the revenue potential from just 10 MW of HPC NVIDIA chips serving global AI demand is comparable to that of 100 MW of Bitcoin mining," HIVE executive chairman Frank Holmes said.
Holmes emphasized that this expansion underscores the strategic value of diversifying into high-performance computing to seize opportunities in the rapidly growing AI market, while staying true to HIVE's core principles of sustainability, innovation, and Bitcoin mining.
"As the global demand for AI-driven solutions grows, we are riding the wave with sustainable, high-performance infrastructure that supports industries and society while staying true to our environmental values," he said.
Aydin Kilic, HIVE CEO, highlighted that the company has deepened its expertise in operating GPUs for AI applications by collaborating with NVIDIA and their OEM vendors over the past year.
"Looking forward, we are very excited to build on this momentum and expand our server capacity with next-generation AI compute powered by NVIDIA's latest GPUs, based on NVIDIA reference architecture," Kilic said.
"Further, we are preparing for the provision of liquid-cooled data center technology in our existing HIVE data centers for the next generation of high-density compute."