Microsoft enters –$9.7B contract with IREN for access to Nvidia chips (apnews.com)

Microsoft has signed a five‑year, $9.7 billion cloud services contract with Australian AI cloud provider IREN that gives it access to Nvidia GPUs, with a 20% prepayment as part of the deal. IREN, which bills itself as a fully integrated AI cloud operator (data centers through GPU stack) with secured power capacity, is also buying chips and related hardware from Dell for roughly $5.8 billion; those GPUs are planned to be deployed in phases at IREN’s Childress, Texas campus. The announcement sent IREN shares up ~22% premarket; Microsoft’s stock rose slightly. For the AI/ML community this is notable because it reflects Microsoft’s aggressive strategy to lock in scarce GPU capacity and scale AI infrastructure quickly without building every asset itself. Microsoft has been pouring capital into cloud and AI — nearly $35 billion in Q3 capex, much of it for chips and data‑center expansion — and this partnership helps secure Nvidia accelerators for large training and inference workloads. The deal also underscores broader industry dynamics: vertical partnerships (cloud provider + infrastructure operator + OEM) are becoming a primary route to expand GPU supply, shorten deployment timelines, and meet surging demand for LLMs and other compute‑heavy AI workloads.