Particle.news
Download on the App Store

NTT/ThoughtLab Warns AI-Driven Data Center Demand Could Outstrip Supply Chains and Grids

A July 2 report says fast growth in AI compute will strain power, equipment, land and labor unless industry and regulators coordinate planning.

Overview

  • The NTT Data and ThoughtLab study published Thursday models three growth paths and finds the most likely scenarios see annual data-center demand rising about 23 to 30 percent through 2030.
  • The report names decisive bottlenecks that could delay builds: limited grid capacity and interconnection, long lead times for processors, transformers, switchgear and backup generators, and shortages of suitable land and specialized construction labor.
  • Researchers say more than $60 billion of data-center projects have been delayed or canceled worldwide and that U.S. cancellations climbed from 2 in 2023 to 6 in 2024 and 25 in 2025, reflecting rising local opposition and permitting hurdles.
  • To avoid a capacity crunch the study recommends early co-planning with utilities, longer-term and diversified procurement of long-lead equipment, investment in advanced cooling and workload efficiency, and clearer community messaging about local benefits.
  • The findings signal trade-offs for communities and grids because operators are already leaning on on-site or contracted firm power and modular builds, which could raise local environmental impacts and shift costs to utilities and ratepayers if planning and regulation do not change.