In this theCUBE + NYSE Wired: AI Factories – Data Centers of the Future segment, theCUBE’s John Furrier sits down with Dan Roberts, co-founder and co-CEO of IREN, to unpack how AI factories are reshaping the data center model into a new asset class. Roberts details IREN’s shift from Bitcoin mining to GPU cloud at scale, highlighting a fully vertically integrated approach that owns land, substations, grid connections, buildings and servers to deliver resilient, low-cost service. He shares timely metrics on surging demand (“next to insatiable”), the rapid ramp from ~1,900 servers at the end of June to ~23,000 operating or being installed, and headroom to 3x capacity. Roberts also outlines IREN’s footprint with almost three gigawatts of secured land and power, ~800 megawatts of operating facilities and ~160 megawatts ready for AI factories – while emphasizing a 100% renewable energy consumption record over the past seven years. The conversation dives into why GPU-dense architectures and redesigned facilities matter for scale, performance and efficiency – touching on NVIDIA and Dell roles across servers, networking, cooling and system reliability. Roberts discusses practical swap-outs of mining ASICs for NVIDIA GPUs, success with Dell’s air-cooled XE9680, and plans to roll out NVL72 systems with the Grace Blackwell 300s in partnership with Dell, alongside liquid-cooled facilities targeted for year-end capable of supporting GB300s. He explains how IREN’s model aligns economics with demand cycles (mid-market labs to hyperscalers), why building for the long term and listening to customers underwrites reliability, and how accelerating compute unlocks broader AI use cases – turning AI factories into the cornerstone of enterprise infrastructure for the next era of compute, efficiency and innovation.