Microsoft signs $9.7B AI cloud pact with IREN
Microsoft has secured a five‑year, $9.7B capacity deal with Australia’s IREN, including a 20% prepayment for access to Nvidia GB300‑based systems to be deployed at IREN’s Texas campus.
Microsoft has signed a five‑year, $9.7 billion agreement with Australian AI cloud provider IREN to secure access to Nvidia’s latest GB300 accelerator systems, as the software group has raced to add compute for fast‑growing AI workloads.
As per reports, the contract gives Microsoft reserved AI cloud capacity from IREN, including access to Nvidia GB300‑based infrastructure, with Microsoft having made a 20% prepayment under the terms.
To fulfil the agreement, IREN said it has separately agreed about $5.8 billion of purchases with Dell Technologies for GPUs and related equipment that are set to be deployed in phases through 2026 at IREN’s 750MW campus in Childress, Texas, alongside new liquid‑cooled data centres expected to provide around 200MW of critical IT load.
Reuters reported that Microsoft’s prepayment has helped fund IREN’s equipment buys and that the contract is contingent on IREN meeting delivery milestones.
Why has Microsoft turned to IREN?
Microsoft said the deal has added near‑term AI headroom without the long lead times and capital needs of building new sites, at a moment when the company cautioned that AI infrastructure tightness could persist well into 2026.
The arrangement complemented Microsoft’s parallel rollout of its own GB300 NVL72 supercomputer‑scale clusters on Azure.
IREN has positioned itself among “neocloud” providers focused on GPU‑accelerated compute. The company—co‑founded and co‑led by Daniel Roberts—said the Microsoft contract is its largest to date and underlined the value of its secured power capacity and vertically integrated data‑centre buildout.
Microsoft’s AI buildout continues
In parallel with third‑party capacity arrangements, Microsoft unveiled its first production‑scale Azure cluster based on Nvidia’s GB300 NVL72 systems, designed for advanced reasoning and multimodal AI.
The latest quarter has shown Microsoft’s heavy infrastructure outlay, with the company reporting $77.7 billion in sales and nearly $35 billion in capital expenditure, much of it for chips and data‑centre projects tied to AI demand, according to the Associated Press.
Satya Nadella, Microsoft’s chief executive, has repeatedly referred to such investments as foundational to Copilot and Azure AI growth.