Overview
- Reporting this week confirmed forecasts that global data center electricity use could roughly double by 2030 and that U.S. demand may more than double in the near term to roughly 66 gigawatts.
- The primary constraint is deliverable power—interconnection queues, transmission buildouts and permitting—with the IEA estimating about 20% of planned projects face grid-related delay risk and roughly one-fifth of U.S. projects planning on-site gas to bypass waits.
- Supply-chain bottlenecks for transformers, cables and switchgear have lengthened lead times, and new transmission lines typically take four to eight years to build, which stretches deployment schedules even after permits and financing are secured.
- A new pressure point is inference traffic, which spreads demand across more locations and sharply raises needs for high‑capacity data‑center interconnects and smarter network controls such as 1.6 Tb/s optics and AI-driven routing.
- Grids and industry are pursuing near-term workarounds—batteries inside data centers, demand response and dynamic line ratings—and longer-term solutions like SMR offtakes and transmission builds while regulators debate how to allocate infrastructure costs so consumers are not left paying higher bills.