Event date: June 18–19, 2026  |  Category: Energy, Infrastructure Policy

The Problem: AI’s Insatiable Appetite for Power

Industry analyses circulating around the same time estimated that 30% to 50% of roughly 140 planned U.S. data centers, representing about 16 gigawatts of capacity, could miss their 2026 timelines or be cancelled outright. The bottlenecks were consistent across projects: multi-year waits for transformers and batteries, slow grid connection approvals, and, increasingly, local opposition citing water and energy usage. Only a fraction of announced projects were under active construction.

What FERC’s Order Changes

The reform is widely seen as a win for hyperscalers and AI labs, who have warned for months that the United States risks losing the AI infrastructure race to jurisdictions that can move faster on permitting and power access. By streamlining aspects of the interconnection process, the order aims to shorten the gap between a data center being announced and a data center actually being able to draw power.

The Cost Question: Who Pays?

The order is not without controversy. Consumer advocates have already signaled they may challenge it in court if FERC does not build stronger cost-allocation protections into the next phase of rulemaking, arguing that ordinary electricity ratepayers should not be left subsidizing the grid upgrades that AI companies need. For businesses in energy-adjacent sectors, or any company factoring data center capacity into its own AI roadmap, this ruling is worth watching closely: it will likely shape both the pace of AI infrastructure expansion and the political temperature around it for the rest of 2026.