According to IREN co-CEO Daniel Roberts, the primary bottleneck in building new AI data centers is physical infrastructure like utilities and construction, not the supply of NVIDIA GPUs.

According to IREN's co-CEO Daniel Roberts, the primary bottleneck for building new AI data centers is not the supply of NVIDIA GPUs, but rather infrastructure challenges like utilities and physical co...

According to IREN's co-CEO Daniel Roberts, the primary bottleneck in building new AI data centers is not NVIDIA's GPU supply, but rather infrastructure challenges like utilities, concrete, and labor.

A new gigawatt-scale AI data center project started today would likely not have its first computer online until the 2030s.

It can take 18 to 24 months for a data center developer to even get a response from a utility about power availability for a potential site.

The primary bottleneck for building new AI data center capacity is no longer silicon, but rather power, steel, and concrete.