Microsoft invested $9.7 billion in the Australian startup IREN: What should you know about it?

Microsoft Corporation has signed a large five-year agreement worth about $9.7 billion with the Australian cloud provider IREN. The deal involves leasing computing power for artificial intelligence tasks using Nvidia GB300 systems at IREN’s data center in Texas. The contract includes a 20% upfront payment and is the largest in the partner company’s history, according to Bloomberg.

Details of the Deal and Its Impact on IREN

The signing took place on November 3, 2025. Microsoft will gain access to high-performance Nvidia GB300 GPUs optimized for AI computing, including model training and generative workloads. Deployment of the systems is planned in stages throughout 2026 at IREN’s Chil­dress (Texas) site, which has a capacity of 750 MW. Once fully implemented, the deal will bring IREN about $1.94 billion in annual revenue.

In response to the partnership, IREN announced the purchase of graphics processors and equipment from Dell Technologies worth $5.8 billion. This will expand its infrastructure, including liquid cooling systems to support 200 MW of critical IT load in four phases. The company plans to finance the project through cash reserves, client prepayments, operational cash flow, and additional borrowing.

IREN, formerly known as Iris Energy, belongs to the growing “neocloud” segment — specialized data center operators for AI. The company started as a Bitcoin miner but quickly pivoted to cloud computing. The AI investments have already paid off: IREN’s NASDAQ stock has risen more than 500% since the start of the year, and after the announcement of the deal — another 35% in pre-market trading. Analysts note that the partnership with Microsoft confirms IREN’s reliability as a provider for hyperscalers such as Meta and OpenAI.

Microsoft’s Strategy in the AI Era

For Microsoft, this is part of a global expansion of Azure’s cloud infrastructure. Demand for AI services, including Azure AI Services and OpenAI support, is growing exponentially. In October, the company launched its first production cluster with Nvidia GB300 NVL72 for tasks such as agent-based AI systems and multimodal models. A similar deal with Nscale for 200,000 GPUs covered Europe and the United States.

The partnership with IREN strengthens Microsoft’s position in the U.S., where GPU and data center shortages are a key issue. IREN highlights “significant interest” in its facilities in Sweetwater (with a 2 GW reserve), opening opportunities for additional clients. IREN CEO Daniel Roberts called the deal a “milestone,” emphasizing the platform’s vertical integration — from data centers to the GPU stack.

Prospects and Risks of “Neoclouds”

IREN, along with competitors such as CoreWeave, Crusoe, and Nebius, uses mining experience for AI: large GPU farms can easily be repurposed for new tasks. The “neocloud” market is booming — in 2025, investments in AI infrastructure are expected to exceed $200 billion. For IREN, this is a chance to become a key player, though risks remain: dependence on a single client (Microsoft — 20% prepayment), scaling execution risks, and volatility of small-cap stocks.

Microsoft, in turn, is diversifying its supply chain, avoiding overreliance on Nvidia. Analysts predict that such deals will accelerate the transition to proprietary modems and chips, reducing costs.

In Brief…

Microsoft signed a $9.7 billion, five-year contract with Australia’s IREN to lease Nvidia GB300 systems at a Texas data center — the largest deal in IREN’s history, sending its stock up 35%. IREN is purchasing $5.8 billion in equipment from Dell for 2026 deployment. The agreement strengthens Azure’s AI infrastructure, while IREN exemplifies the “neocloud” model — former miners shifting to cloud services. The market is expanding rapidly but comes with dependency and scaling risks.