Artificial Intelligence Hits The Grid As Utilities Race To Keep Up

an abstract representation of solving problems using artificial intelligence to increase reliability and reduce losses and accidents during the transmission of electrical energy

getty

Duke Energy didn’t turn to artificial intelligence to chase the latest tech trend. It shifted to AI because every storm season now tests grid resilience. In 2024, the utility’s investment arm backed AiDash, a startup using high-resolution satellite imagery and machine-learning algorithms to spot tree growth and excessive vegetation along power-line corridors—long before crews see the threats on the ground. The goal: fewer outages, lower wildfire risk, and a grid ready for surging demand from data centers and electrified industries.

That’s the new watershed in the grid business: utilities are no longer just delivery systems. They are data-driven networks that manage load growth, operational risk, and the machines that drive the machines—automated systems that communicate directly with other equipment to manage power flows in real-time. The time has come for a broader transformation: utilities must invest in AI and emerging technologies to future-proof the grid, bolster reliability, and thrive in the era of electrification.

“AI is our problem, but it’s also potentially our salvation,” Steve Smith told me in an interview. He serves as President of National Grid Partners—the venture arm of National Grid—and Chief Strategy & Regulation Officer at his parent utility.

For roughly two decades, U.S. electricity demand was relatively flat. But now several forces are converging: AI computer farms, massive data centers, the electrification of transportation and heating, and heightened expectations for reliability in a warming climate. Smith cautions that utilities are “in the spotlight and the firing line” as these dynamics intersect.

NGP’s numbers highlight the stakes. Since its 2018 launch, it has deployed roughly $550 million across more than 50 companies. Seven portfolio companies have already achieved successful exits—through acquisitions or IPOs—with more on the horizon.

Not every bet is a winner. While 2–3 investments may not succeed, Smith says the mission remains: to accelerate proven technology into utility control rooms, not just pilot programs.

Real Investments, Real Impacts

In the U.K., National Grid has partnered with Emerald AI, whose “Conductor” software manages data center workloads in real time based on grid conditions. Rather than running computing tasks whenever a data center chooses, the software shifts loads to avoid network stress. In a demo project, this approach cut power usage by 25% during peak demand periods.

AES has teamed up with LineVision to install 42 non-contact sensors along transmission lines in Indiana and Ohio. Using Dynamic Line Ratings, the sensors monitor real-time conditions, allowing AES to safely transmit more power through existing lines. The result: increased capacity at a fraction of the cost and time of conventional upgrades.

Southern Company, through its subsidiaries Alabama Power and Georgia Power, is piloting WeaveGrid’s smart-charging platform. The system automatically schedules home EV charging for enrolled customers, shifting electricity use to off-peak hours. The payoff: supporting more EV adoption without costly infrastructure or long regulatory delays.

At National Grid Partners, Smith describes the approach as straightforward. The strategy? Invest in companies with proven products, deploy them within its utility system, prove them at scale, and then let them expand across the industry. “Our role is to get out of the way of growth. If a company wants to connect to our grid, we find ways to do it faster—two years instead of three or four—using AI and technology.”

The Stakes And The Pay-Off

Utilities can reap big benefits from smart grids. Faster connections for high-demand customers like data centers and factories open up new revenue streams. Plus, smarter asset use helps delay costly infrastructure upgrades. Improved forecasting and adaptable load management reduce outage risks and operational costs. In a world focused on reducing carbon emissions, utilities with the most advanced grids will thrive.

But momentum isn’t a guarantee. The utility business remains regulated, risk-averse, and capital-intensive. Investment in “disruptive” technology within utilities comes with guardrails: ratepayer sensitivity, regulatory scrutiny, and long lead times. The specter of the 2000-era tech bubble looms when energy meets hype. As Smith acknowledged in our interview, not every bet pays off. Utilities must stay disciplined: this isn’t venture capital chasing moonshots—it’s strategic capital seeking operational value.

But, clearly, the demand surge isn’t hypothetical. The tools are increasingly proven, and the business model is evolving. Utilities are no longer just buyers of technology—they are partners, investors, and deployment platforms.

“Innovation isn’t proven until it’s deployed—and utilities need to learn faster than they’re comfortable with,” Smith says. “I’m optimistic. There’s an enormous opportunity here. AI enables us to modernize, expand capacity, and share knowledge across utilities quickly—speed is of the essence.”

For utility executives, corporate-venture teams, and investors watching the grid-software nexus, the lesson is unmistakable: the era when utilities waited for innovation to arrive is over. The era when they help own it has begun.