Microsoft just locked down another massive piece of the AI infrastructure puzzle with a $9.7 billion deal that sends a clear signal about where the compute wars are heading. The five-year contract with Australia's IREN gives Microsoft access to cutting-edge Nvidia GB300 GPU infrastructure at a Texas facility, marking the tech giant's latest aggressive move to meet exploding AI demand from enterprise customers.
Microsoft is writing checks faster than a startup burning through Series A funding, and today's $9.7 billion contract with Australia's IREN proves the company isn't slowing down its AI infrastructure spending spree. The five-year deal announced Monday gives Microsoft access to compute infrastructure built with Nvidia's latest GB300 GPUs, deployed in phases through 2026 at IREN's facility in Childress, Texas.
The numbers tell the story of an industry in hyperdrive. IREN's Texas facility will support 750 megawatts of capacity - enough power to run a small city - and the Australian company is separately committing $5.8 billion to buy GPUs and equipment from Dell. That's nearly $15.5 billion in combined infrastructure investment for a single partnership.
This comes just weeks after Microsoft launched its first production cluster with Nvidia's GB300 NVL72 systems for Azure, which the company says are optimized for reasoning models, agentic AI systems, and multi-modal generative AI. The timing isn't coincidental - Microsoft is racing to deploy the most advanced chips available as competition for AI compute capacity reaches fever pitch.
The IREN deal follows Microsoft's October partnership with Nscale for approximately 200,000 Nvidia GB300 GPUs across four data centers in Europe and the U.S. These back-to-back agreements show Microsoft is diversifying its compute partnerships while betting big on Nvidia's latest architecture.
IREN's transformation tells a broader story about the AI gold rush reshaping entire industries. Like competitors including CoreWeave, IREN started as a Bitcoin mining operation but pivoted when executives realized their GPU collections were more valuable for AI workloads than cryptocurrency. The pivot paid off spectacularly - CEO Daniel Roberts told Bloomberg the Microsoft deal will generate about $1.94 billion in annualized revenue while using just 10% of the company's total capacity.
That math reveals the scale of demand Microsoft is preparing for. If this represents only a fraction of IREN's capability, the company is building infrastructure for AI workloads we can barely imagine today. The phased deployment through 2026 suggests Microsoft expects sustained growth in enterprise AI adoption, not a temporary spike.
