The Queue Killer: AI Takes on Grid Delays

National Grid and GridCARE are using AI to slash interconnection wait times from years to months across upstate New York

14 Apr 2026

National Grid and GridCARE have announced a partnership to deploy artificial intelligence tools across upstate New York's power grid, aiming to accelerate connections for large electricity users facing waits that can stretch several years.

GridCARE's Energize platform uses physics-based modelling to identify spare capacity within the existing network that conventional planning methods tend to miss. Rather than defaulting to costly new infrastructure, the system looks for conditions under which current equipment can carry greater load, and flags distributed assets such as batteries that could provide additional headroom. The companies say connection timelines could compress to as little as six to twelve months.

The initiative arrives as upstate New York draws growing interest from data centre developers and semiconductor manufacturers attracted by its mix of nuclear, hydroelectric, gas, and wind generation. For many of these projects, speed of grid access is the deciding factor in site selection.

GridCARE can point to prior results. A collaboration with Portland General Electric in October 2025 unlocked 80 megawatts of capacity for data centre connections, demonstrating that software-led approaches can move quickly where conventional grid planning has historically stalled.

National Grid has positioned the programme as both an economic development and a bill-reduction measure. Serving new customers through existing infrastructure, rather than building new lines, lowers costs for utilities and, the company argues, for existing ratepayers. GridCARE has separately published research suggesting that well-managed large flexible loads can reduce grid costs across all customer classes.

Whether the approach scales is the open question. Upstate New York's concentrated demand corridors represent a favourable test case, but replicating the model across larger or more complex networks would require broader data-sharing arrangements and regulatory support that remain untested at scale.