Skip to content
Cloud & Data

Argentum AI and Kyndryl Sign $630 Million Deal to Expand AI Data Centre Capacity

Kyndryl data centre infrastructure supporting Argentum AI's Nvidia B300 GPU deployment in Boulder, Colorado.

GPU cloud provider Argentum AI and technology services company Kyndryl have signed a 20-year, $630 million data centre collaboration to expand infrastructure for enterprise artificial intelligence workloads. The initial agreement provides Argentum with 10 megawatts (MW) of capacity at Kyndryl's facility in Boulder, Colorado.

The companies plan to use the site to deploy Nvidia B300 GPU systems in a Tier III data centre environment. Kyndryl announced the agreement on November 17, saying deployment was already underway. The infrastructure is intended to support high-performance AI inference, allowing businesses to run trained models and process AI workloads at scale.

Argentum AI Kyndryl Deal Centres on Nvidia B300 Systems

The agreement gives Argentum access to data centre capacity designed to accommodate the power and cooling requirements of current-generation GPU systems. The initial 10 MW allocation will support Nvidia B300 deployments at the Boulder site.

B300 systems are designed for demanding accelerated computing workloads. Their deployment will provide Argentum with dedicated infrastructure for AI inference services, where processing speed, capacity and operational reliability are important for enterprise customers.

The Tier III environment is intended to provide infrastructure resilience and maintenance flexibility. Kyndryl will contribute its data centre and technology operations experience, while Argentum will deploy and operate its GPU cloud infrastructure.

A 20-Year Partnership for Enterprise AI Infrastructure

The $630 million agreement spans 20 years, giving Argentum a long-term infrastructure arrangement as demand for AI computing expands. The companies have also indicated that they intend to broaden their collaboration as Argentum's requirements grow.

Argentum operates GPU clusters and provides dedicated bare-metal computing capacity to AI developers, inference providers and enterprises. Rather than requiring every customer to build and manage its own data centre infrastructure, the provider offers access to specialised GPU resources through its cloud business.

For Kyndryl, the partnership puts existing facility capacity to work for a specialist AI infrastructure provider. Its role combines physical infrastructure with operational and managed technology services.

The companies have not announced a separate breakdown of the agreement's revenue or financial contribution to Kyndryl. The $630 million headline value covers the 20-year collaboration and should not be interpreted as an immediate payment or annual revenue figure.

Data Centre Capacity Becomes Critical for AI Expansion

The agreement reflects a broader shift in data centre demand as businesses deploy more AI applications. Running large models requires specialised accelerators, substantial electricity capacity, advanced cooling systems and reliable connectivity.

GPU cloud providers are seeking long-term arrangements with infrastructure operators to secure the space and power needed for these deployments. Partnerships can reduce the time required to establish new facilities while allowing specialist providers to concentrate on GPU systems and customer services.

For Argentum, the Boulder agreement establishes an initial infrastructure base for its enterprise AI inference business. For Kyndryl, it creates an opportunity to serve the expanding market for high-density computing.

The next measure of the partnership will be how quickly the B300 systems become operational and how effectively the facility serves customer demand. The companies' intention to expand their collaboration suggests that the initial 10 MW could form the starting point for a longer infrastructure relationship.