IREN recently reported that demand for its cloud services from hyperscalers, enterprises, and frontier labs is surpassing both current and planned capacity. This comes as cloud contract economics improve, with new deals priced above previous contracts, including $15 million per megawatt for the latest cohort compared to NVIDIA’s and Microsoft’s contracts. The most recent agreements emphasize long-term relationships, featuring a weighted-average duration of over three years and substantial customer prepayments that account for about 45% of GPU capital expenditures. This aligns with broader trends where AI infrastructure agreements increasingly incorporate multi-year terms and upfront payments to facilitate hardware acquisition.

IREN: IREN, operating as Iris Energy, develops and operates renewable energy-powered data centers that support both Bitcoin mining and AI/high-performance computing workloads. The company is actively expanding its GPU cloud offerings to meet needs from large-scale AI users. In this news, IREN highlights persistent excess demand for its capacity and improving economics on new GPU contracts.
NVIDIA: NVIDIA designs and supplies graphics processing units and related platforms optimized for AI training, inference, and data center applications. Its latest Blackwell-generation chips power advanced AI systems. The news notes that IREN’s newest customer agreements center on NVIDIA’s B200 and B300 GPUs.
Microsoft: Microsoft operates one of the world’s largest cloud platforms and procures substantial GPU capacity for its AI and enterprise services. Its existing GPU agreements provide market reference points for pricing and terms. The report compares IREN’s recent deals favorably against Microsoft’s contract rates.
John Todaro: John Todaro is a technology infrastructure analyst at Needham & Company. He authored the note summarizing a management call with IREN on demand trends, contract details, and capacity outlook.

Hardware Trends: NVIDIA’s Blackwell-series GPUs are being incorporated into new AI cloud deployments across the industry.
Contract Structures: Recent AI infrastructure agreements increasingly include multi-year terms and customer prepayments to support hardware acquisition.
AI Infrastructure Demand: Hyperscalers, enterprises, and frontier AI labs continue to seek additional GPU capacity beyond currently available supplies.