AUTO-UPDATED

What's slowing down the AI buildout

The rapid expansion of AI data centers is currently constrained by inefficient electric grid interconnection processes rather than a fundamental shortage of electricity generation capacity across the United States.

Key Points

  • Major AI projects like the $40 billion Stargate campus require gigawatt-scale power, straining existing infrastructure and grid capacity.
  • The median wait time for power plant grid interconnection has surged from under 20 months in 2005 to 55 months by 2023.
  • Current "first-come, first-served" queue systems are backlogged with speculative projects, preventing high-value infrastructure from connecting efficiently.
  • Grid operators face significant congestion costs, which rose 45% to $11.5 billion in 2023 due to transmission bottlenecks.
  • Proposed solutions include auctioning grid capacity, adopting "connect and manage" policies, and incentivizing data centers to use on-site backup during peak demand.

Why it Matters

The inability to connect new power-hungry infrastructure to the grid threatens to stall the economic potential of the AI boom and broader industrial electrification. By reforming outdated interconnection policies, regulators can ensure that critical technology projects receive the energy they need without compromising grid reliability or driving up consumer costs.
Worksinprogress.news Published by Works in Progress, Chris Gillett
Read original