Utilities have long been accused of slow-walking innovation. Fair or not, the advent of artificial intelligence (AI) tools, and the subsequent wariness from those responsible for maintaining our most critical infrastructure, has only intensified that perception.
But make no mistake: we’re in the midst of the most-rapid technological evolution of utilities in the power grid’s history. Unprecedented load growth from electrification and data centers paired with the scaling of distributed energy resources has required utilities to embrace advanced digital tools for planning, grid management, and customer engagement. And despite those long-held perceptions about risk-averse utilities, they’re chomping at the bit to explore how AI can support their mission of providing safe, clean, reliable, and affordable power to customers.
COMMENTARY
I’m in these conversations seemingly every day. Utilities, especially executive leadership teams, are not opposed to AI innovation. But they are skeptical that general-purpose applications from Big Tech and startups alike can stand up to the demands placed on the power grid. While the opportunities presented by AI are vast, so too are the risks. Utilities already have near-zero margin for error—so, in their view, they can’t get AI wrong. The current marketplace meanwhile is filled with AI that is not built with the grid’s complexities in mind, nor does it possess the data needed to ensure the grid meets today’s rising demands. Utility stakeholders are simply waiting for AI tools to emerge that fit their standards on reliability and flexibility.
It’s Incremental Change, Not Radical
The energy sector is vital, with utilities of all sizes responsible for providing essential services to homes and businesses, even more importantly, human welfare. The critical importance of their role in everyday functions has utilities slowly adopting lower-risk AI implementations that do not jeopardize operations. This could be anything from customer service chatbots, billing automation and forecasting tools, to administrative process improvements and energy demand monitoring.
The organizational structure of utilities also plays a role in the, sometimes slow, process. Cross-functional committees design strategies, pilot programs, review regulatory risks oversight, and stress-test cybersecurity and reliability protocols. For some utilities, the technology vetting process takes years. With AI, there’s a collective sense of urgency to take advantage of new functionality and efficiency, but skipping steps also isn’t an option. These teams and stages of research and deployment are important, too.
Urgency isn’t only linked to the art of the possible excitement–the grid’s increasingly-decentralized structure is leading to a wholesale overhaul of utilities operate, both internally and external. It’s more interconnected and complex than ever before and utility resources are, as they’ve always been, stretched thin. For an industry that loves a buzzword, flexibility is the topic de jure. But executing flexibility–true flexibility–requires real-time situational awareness, data sharing, and coordination between the grid edge, utility operations, and energy markets like we’ve never seen before. This is one of the great opportunities that AI is unlocking for utility leaders.
Forget the grid of tomorrow; the grid of today necessitates an unprecedented rate of innovation. Utilities aren’t just adapting to a new wave of technology—they’re searching for tools that can provide near-term operational support to meet exceedingly high customer expectations.
Policies are Mandating Modernization
Federal and state regulators see the writing on the wall. They are hearing the outcries from ratepayers nationwide, with 77% reporting they’re concerned bills will increase further this year. They are also seeing the impact AI is having in the private sector, with millions if not billions of dollars being poured into the build out of data centers across the U.S. Taken together, the recent initiatives and policies making their way through agencies are beginning to shape the role AI technologies will play in the utility sector.
The U.S. Department of Energy (DOE) has been the catalyst of this recent policy push with programs like the Grid Modernization Initiative and Speed to Power Initiative, which is injecting $1.9 Billion into the buildout and implementation of these new technologies into the grid ecosystem. The funding was included through the passage of the Infrastructure Investment and Jobs Act back in 2021. Also, the DOE has since rebranded its GRIP program to SPARK, which is set to allocate up to $10.5 billion in competitive funding over the next five years. This funding will support states, tribes, electric utilities, and other eligible entities in enhancing grid resilience and promoting innovation.
The push by lawmakers and regulators doesn’t stop at the federal level. The country’s largest states – California, New York, Texas, and Florida–are all beginning to mandate grid modernization in their own distinct ways.
For example, the California Public Utilities Commission (CPUC) has implemented intense wildfire prevention measures, enhanced outage notification protocols, mandates for distributed energy integration, and standards for reliability performance. Right now, the CPUC and California ISO, the state’s grid and market operator, are working through how to better coordinate to realize the value of distributed energy resources, and avoid the reliability risks they present.
In New York, the Department of Public Services deployed its Reforming the Energy Vision (REV) program back in 2016, leading the way in requiring enhanced grid analysis, management of distributed energy and adaptive load balancing. All of which has intensified over the past decade following the immense load growth demand and overall impact we are seeing climate change have on utility assets and infrastructure.
While not every regulatory proceeding is focused on AI, it’s notable that many no longer avoid or outright prevent the exploration of AI as a potential solution to the challenge. In fact, we’re seeing mounting requests for AI use cases in utility proposal requests and regulatory dockets. Utilities and grid stakeholders are feeling the pressure from all sides, especially on rising consumer rates and data center infrastructure buildout costs. But with reliability paramount, any old AI won’t do the trick.
Generic AI Doesn’t Understand the Grid
The most effective AI implementations for utilities and grid stakeholders typically arise from solutions that are energy-native, with a clear understanding of how the grid operates. As an example, we developed an energy-native AI workflow agent that was implemented within CAISO’s current operational framework.
This AI agent shaved down CAISO’s daily outage processing time from six hours to just 30 minutes by proactively processing outage notes to uncover any discrepancies, and then applying necessary operating procedures as needed. The agent also is able to sort through 30 years of historical grid data, in addition to CAISO’s, to anticipate reliability risks, providing a tailored report that allows operators to quickly assess any grid strain. While the agent handles the analytical aspects, operators maintain full control over all critical decisions, applying their expertise where it counts.
This level of knowledge and processing is not possible with generic AI tools. In fact, generic AI can produce incorrect outputs that are presented in an authoritative manner, better known as AI hallucinations. Utilities must look for solutions that possess the knowledge of power flow physics, reliability standards, interconnection procedures, power congestion management, and even outage restoration logic.
Utilities Are Not Slow, They Are Selective
It bears repeating: utilities are not slow, they are selective. They aren’t resisting AI innovations, they are simply filtering out technologies that cannot sustain the reliability and rigorous accountability it takes to uphold current grid operations.
As AI adoption continues to scale, utilities that turn to solutions with energy-native architectures will ensure optimized operational reliability. By pairing these solutions with human oversight, utilities and grid stakeholders will successfully transition into the AI era.
—Dr. Sasan Mokhtari is a renowned electrical engineer and the founder, president, and CEO of Open Access Technology International, Inc. (OATI). With more than 40 years of industry-defining contributions, he has led innovations in grid operations, market systems, and distributed energy resource management.