IREN CEO Warns 1-Gigawatt AI Factories Started Now May Wait Until 2030 for Power
Updated
Updated · 24/7 Wall St. · May 25
IREN CEO Warns 1-Gigawatt AI Factories Started Now May Wait Until 2030 for Power
3 articles · Updated · 24/7 Wall St. · May 25
Daniel Roberts said a 1-gigawatt AI factory launched today may not get its first compute online until 2030 because power connections have become the key constraint.
Utilities may need 18 to 24 months just to determine whether a site has available capacity, underscoring how grid interconnection is now slowing AI buildouts before construction even begins.
That shifts the bottleneck beyond Nvidia GPUs and memory chips to electricity, cooling, fiber and land near transmission lines, even as Microsoft, Amazon and Meta pursue multi-gigawatt expansion.
The International Energy Agency expects data-center electricity demand to more than double by 2030, suggesting companies that already control powered sites and interconnects could gain an edge over cash-rich rivals starting from scratch.
As AI's energy demand soars, will small nuclear reactors become the tech industry's most critical new partner?
The AI revolution requires a trillion-dollar energy overhaul. Can our planet's power grids actually support it?
With power, not chips, as the new AI bottleneck, who will win the race to build the digital future?
The Real AI Bottleneck: Grid Capacity, Energy Infrastructure, and the Battle for Compute Power
Overview
The report reveals a major shift in the AI industry's growth challenges, as highlighted by IREN CEO Daniel Roberts. The main bottleneck has moved from acquiring advanced AI chips to securing reliable power and grid infrastructure. This means that having money to buy chips is no longer enough; companies now face delays from utility approvals and building transmission lines. As a result, project timelines and capital planning for new data centers are directly affected. The ability to scale AI depends on existing infrastructure and access to power, making these factors critical for future AI expansion.