Microsoft has entered into a five-year agreement worth US$9.7 billion with data-centre operator IREN, granting the software giant access to high-performance AI cloud capacity built on Nvidia’s GB300 graphics processing units (GPUs).

The deal reflects Microsoft’s urgency to scale its AI offerings without the delays and capital intensity of building large new data-centres.

Under the contract, IREN will deploy compute infrastructure at its Childress, Texas facility, with phased roll-out through 2026 linked to its broader plan to deliver 750 megawatts of capacity at the site.

IREN has separately committed to purchasing approximately US$5.8 billion of GPUs and related equipment from Dell Technologies, enabling deployment of the Nvidia GB300 systems.

Microsoft will become IREN’s largest customer, with the deal expected to generate about US$1.94 billion in annualised revenue for IREN.

A portion of the payment — roughly 20 per cent — is being made in advance, helping fund IREN’s hardware purchases.

IREN, formerly focused on bitcoin mining, has pivoted its business toward large-scale AI cloud infrastructure, leveraging its renewable-power-backed data centres across North America.

The company says its facilities are fully grid-connected and designed for high-density compute.

For Microsoft, the arrangement offers a strategic avenue to expand its AI-cloud footprint while sidestepping the heavy capital outlay and time lag of constructing new proprietary data centres.

The partnership also addresses an ongoing shortage of advanced compute capacity needed for generative AI, agentic systems and multimodal reasoning models.

Market reaction was swift: IREN’s shares rose sharply on the announcement, while the move reinforces broader industry momentum around infrastructure for frontier AI workloads.

The contract includes performance milestones and is contingent upon IREN’s delivery schedule.

The deal underscores how hyperscaler cloud providers are increasingly outsourcing compute capacity to specialist “neo-cloud” operators that evolved from crypto-mining businesses, converting large GPU-rich installations into AI-optimised platforms.

IREN’s shift away from bitcoin mining toward enterprise AI infrastructure exemplifies this trend.

With AI-driven demand for computing power showing little sign of abating, Microsoft’s commitment signals confidence in the medium-term outlook for its AI business and reflects the company’s strategy to secure bottlenecked resources.

IREN, for its part, will rely on the partnership to scale its operations across its secured power portfolio and deliver step-changes in GPU deployment.

The contract offers both firms a pathway to capture a larger share of the rapidly evolving AI-cloud market.