HIVE Digital Technologies lands a $220M AI infrastructure GPU cloud deal with Bell and Cohere

HIVE Digital Technologies said it has signed a major artificial intelligence infrastructure agreement valued at $220 million, centered on GPU cloud capacity delivered through its high performance computing unit, BUZZ High Performance Computing (BUZZ HPC). The company announced the deal on June 18, naming Bell and Cohere as key counterparties, and framing the arrangement as part of an expanded AI buildout.

According to the announcement, the agreement is intended to support the deployment of GPU-accelerated compute for AI workloads. GPU cloud refers to cloud infrastructure provisioned with graphics processing units, which are designed to speed up the training and running of machine learning models. For companies building AI products or running large-scale inference, access to GPU capacity can be a gating factor, since AI systems typically require substantial parallel computing.

HIVE’s role in the arrangement runs through BUZZ HPC, which the company uses to operate or supply high performance computing resources. By routing the deal through BUZZ HPC, HIVE is effectively positioning itself not just as an investor in AI, but as an infrastructure supplier for customers and ecosystem partners looking for scalable GPU capacity.

The announcement also ties the agreement to Bell, which is widely associated with communications services in Canada, and Cohere, a company known for commercializing natural language processing and other language-model capabilities. Together, the named partners suggest the deal could be aimed at enterprise use cases where large language models need sustained compute resources, whether for customer-facing applications, internal knowledge systems, or other AI-driven workloads.

While the headline value is $220 million, the company’s public framing did not provide additional deal structure details in the June 18 market post. HIVE did not specify in the announcement text, at least as presented by the reporting outlet, the length of the contract, the exact delivery schedule for compute, or whether the figure reflects revenue, contracted infrastructure spend, or total potential commitment.

HIVE is not the only player seeking to expand GPU cloud capacity as AI adoption accelerates, but the size of the reported commitment underscores how quickly infrastructure sourcing has moved into the mainstream of corporate technology budgets. In practice, deals like this often involve multi-year capacity planning, performance requirements, and service-level expectations, especially when customers need predictable throughput for AI training runs or production inference.

Cohere’s inclusion indicates a focus on language and AI services rather than a general-purpose compute play alone. Companies that provide or deploy language-model capabilities typically need GPU-heavy environments for model usage at scale, and infrastructure partners can become central to meeting throughput, latency, and reliability targets.

Still, the announcement leaves open what is not disclosed publicly. HIVE did not detail the specific technical configuration of the compute, the percentage of capacity allocated to each named counterparty, any customization or software stack agreements, or the unit economics behind the $220 million figure. Those omissions matter for assessing how quickly revenue could be recognized and how sensitive the deal may be to changes in GPU pricing or AI demand.

Investors and customers will likely watch for follow-through disclosures, such as contract duration, operational start dates, and any additional terms around capacity provisioning and performance. HIVE’s next reporting cycle will also be a key place to see whether the company quantifies expected revenue contribution, cost structure, and any related capital requirements for delivering the capacity implied by the deal.