100 Million Meters Walk Into an AI Lab

Itron's grid edge AI platform has reached 100M endpoints, promising US utilities 10% fewer outage minutes and 20% more capacity

1 Apr 2026

America's electricity meters have long been passive observers of consumption. Itron wants to make them active participants in keeping the lights on.

The utility technology company has built a network of more than 100 million connected grid endpoints across the United States, and is now deploying artificial intelligence at the edge of that network to help utilities manage a grid growing more complex by the day. More than 16 million distributed-intelligence meters have already been shipped. In 2025 alone, the system dispatched 70 GWh of flexible load and generation from customer devices. These are live deployments, not proofs of concept.

At the DTECH 2026 conference in San Diego in February, Itron unveiled an expanded Grid Edge Intelligence portfolio and a newly formed Resiliency Solutions segment, consolidating capabilities from recent acquisitions. The segment targets a 10% reduction in outage minutes and up to 20% more grid capacity through better use of existing assets. For utilities facing rising demand and stretched capital budgets, the appeal is obvious: more performance from the infrastructure already in the ground.

The engineering logic deepened in March 2026, when Itron and NVIDIA demonstrated AI inference running directly on grid endpoints using NVIDIA's Jetson edge computing hardware. By processing high-frequency waveform data at the meter itself, the system can identify faults, wildfire risk indicators, and other emerging threats in real time, without routing everything through a central server first. Speed and resilience improve; dependence on costly centralised computing falls.

Integration with Schneider Electric's EcoStruxure distribution management system means utilities can act on this edge intelligence within existing control workflows, sparing them disruptive and expensive platform overhauls. The result is a continuous data link from the distribution transformer to the control room, a view that becomes more valuable as solar panels, batteries, and electric vehicles multiply on the network.

The underlying proposition is straightforward: the grid is getting harder to manage, and building entirely new infrastructure to cope is neither fast nor cheap. Itron's bet is that the smarter use of what already exists, 100 million endpoints now capable of sensing, computing, and responding, offers utilities a more practical path. Whether that bet pays off will depend less on the technology than on whether regulators and utilities move quickly enough to let it.