AI Demonstrates Its Value in Grid Operations

Building a bridge from yesterday to tomorrow

The grid of yesterday was predictable — stabilized by inertia and centrally controlled by operators. Today, utilities are navigating a journey toward a grid that is far more dynamic and less predictable, stabilized not by inertia but by increasingly autonomous control systems.

Some AI applications now in use help manage the growing number of connected devices on the grid without increasing operational costs. Others reduce outages and improve event response. 

Moving forward, AI will play a critical role in grid operations, as organizations turn to grid-specific systems that operate within a more autonomous control model. Operators will shift from direct control to orchestrating intelligent systems, intervening only when behavior falls outside defined parameters.

To navigate this transformation, utilities must put systems in place that support current operations while laying the foundation for future capabilities.

Enter adaptive intelligence

In utilities, adaptive intelligence refers to AI that continuously learns from grid behavior and conditions, improving coordination, efficiency and resilience over time. 

Systems that use adaptive intelligence continually expand their use of machine learning and AI to process data locally, collaborate across devices and make real-time decisions within operator-defined limits.

While administrative or operator-assisted systems may perform well with general AI models such as large language models (LLMs), many grid tasks require specialized AI — often referred to as domain-specific models (DSMs).

Embracing adaptive intelligence will be key to supporting new data flows and AI-driven functions.

AI and the grid in action

This year's theme at GridFWD 2025 was, "AI and the Grid: The rising tide of applications and demands."

To bring that theme to life, we were proud to co-host a demo room with NVIDIA that showcased adaptive intelligence systems powered by domain-specific models and trained for specific grid tasks. 

These intelligent systems operate across control rooms, transmission and distribution networks, and the grid edge.

Our demonstrating partners included:

• Argonne National Laboratory, which develops grid-specific foundation models, collaboration and training tools, and established a consortium to advance AI and grid research.
• ThinkLabs, which deploys physics-informed neural networks and other DSMs to address a range of grid planning and operations challenges — including a use case that could drastically reduce the time required for interconnection request assessments.
• Buzz Solutions, which uses specialized computer vision models to automate and enhance condition-based maintenance for overhead distribution assets, allowing utilities to monitor and maintain the grid more efficiently.
• Utilidata, which offers a digital platform that collects data from grid-edge devices such as meters and transformers, performing AI-enabled processing directly at the device and sending back actionable insights.

These partners demonstrate the range of approaches that make AI actionable across the grid — from advanced modeling and predictive planning to real-time edge processing and automated maintenance.

By integrating adaptive intelligence capabilities, WWT and NVIDIA help utilities reimagine grid operations, helping them build a bridge from yesterday's grid to tomorrow's.