IREN co-founder and co-CEO Daniel Roberts said every additional AI data center adds to, rather than eases, overall AI demand. He argued the current AI infrastructure cycle is being misread when compared with earlier technology booms, because more capable AI requires more compute power and lower-cost AI tends to drive broader adoption. Roberts said each step-change in AI capability creates new use cases that did not exist a year earlier, allowing demand to grow at software speed while supply remains constrained by physical infrastructure such as power connections, transmission lines, concrete, steel and construction labor. In his view, earlier tech cycles often allowed innovation to help supply catch up with demand, but this time the technology itself is accelerating demand exponentially while real-world buildout cannot expand fast enough.