Infrastructure, Not Chips, Is AI's Biggest Bottleneck, IREN Co-Founder Says

The race for AI computing capacity depends on more than GPU supplies. The electricity, land, cooling systems and grid connections required by data centers can also constrain deployment. IREN co-founder Daniel Roberts said physical infrastructure had become scarcer than chips as AI demand grew exponentially, prompting the company to integrate energy, data-center and NVIDIA GPU resources.

Roberts recently said the industry's focus was shifting from chip shortages to the availability of power and sites. IREN is therefore building an AI platform through a vertically integrated model that controls key components from energy to computing equipment. As of July 20, 2026, reports had not disclosed any additional investment amount, GPU count or specific completion date.