Email)">
One hundred percent of utility leaders are using AI. The question is no longer “if” but “how” and “where.” In September and October 2025, we sat down with leaders from 18 utilities, from Tucson to Charlotte, and Denver to Cincinnati, to understand how artificial intelligence is reshaping one of America’s most critical infrastructure sectors. What we discovered should capture the attention of every utility executive, board member, and operations leader: the transformation is already underway, but most organizations are at a crossroads.
Every utility leader we interviewed is personally using AI tools. Not 80%. Not 95%. 100%. And many are not just dabbling. Nearly half (44%) use AI daily or multiple times a day. The most popular tools? ChatGPT leads at 72%, followed closely by Microsoft Copilot at 67% and Google Gemini at 44%. Many leaders are using three or more tools, experimenting to find which works best for what.
What are they doing with generative AI? The top personal applications:
Real organizational implementations are emerging. One utility’s IVR system now handles 30% of incoming calls autonomously. Another has automated email responses, managing 60% of routine customer inquiries. A vendor working with utilities has reduced the time required for pervious area delineation and digitization by 90% using AI-assisted analysis.
It’s worth noting that many utilities have been utilizing machine learning for years in applications such as predictive maintenance and SCADA optimization. What’s new is generative AI, tools that can write, reason, and create content in ways that directly augment knowledge work. And they’re just getting started. When we asked about the next six months, 92% expect to increase their AI usage. Some significantly.
While 78% of leaders express positive to cautiously optimistic views about AI, they’re also deeply thoughtful about the risks. This isn’t blind enthusiasm, but informed optimism tempered with pragmatic concern.
The top perceived benefits reflect real business needs:
But leaders are equally cognizant of the challenges:
Data security and privacy are the top concerns, mentioned by 82% of respondents and ranked as the #1 concern overall. “We’re a public utility,” one executive reminded us. “We can’t afford to get this wrong.”
Accuracy and reliability concerns follow closely. Leaders understand that AI can “hallucinate” or provide confident sounding but incorrect information. In an industry where decisions impact public health and safety, this matters deeply.
Organizational culture and resistance rank third. “The technology isn’t the hardest part,” one GM observed. “It’s bringing people along on the journey.” More than half the utilities we spoke with identified cultural resistance as a moderate to major barrier. And a major cultural issue is the perceived lack of authenticity provided by AI.
Rounding out the top concerns were fear of over-reliance and loss of critical thinking, competing priorities, unclear policies, and anxiety about job displacement.
While individual leaders race ahead, their organizations are taking more measured steps, and that creates both opportunity and risk.
The maturity breakdown:
Only one organization in our entire sample is actively using multiple AI tools across the enterprise. Most are in pilot or experimental phases.
The most common organizational use cases reflect a “start safe” mentality:
Notably, applications in critical operational systems, such as SCADA, water treatment, and plant operations, remain primarily in the “considering” stage, with seven organizations expressing significant concerns about deploying AI in critical infrastructure.
Many organizations are making progress with AI, but formal guidance has not yet fully caught up.
Similarly, for strategic roadmaps:
Without clear policies or roadmaps, staff may explore AI tools independently, including for tasks that involve sensitive data, without shared guidelines on what’s appropriate, secure, or restricted. One leader described it as “shadow AI usage”: employees experimenting on their own, sometimes with free tools, uploading potentially sensitive utility data to public platforms.
Who’s leading AI initiatives? In most cases (50%), it’s the IT department. However, they often do it part-time, without dedicated resources or clear mandates.
Only seven organizations have part-time assignments dedicated to AI. Four have no dedicated resources at all. Yet organizations are reticent to hire directly and would prefer to leverage staff with subject matter expertise to drive adoption.
The leadership style varies widely:
But some executives are merely “monitoring” or have “delegated” responsibility entirely
Which organizations are showing the most progress? Those where executive leadership actively champions AI initiatives, where cross-functional teams bring together business units and IT, and where appropriate guardrails are established and a culture of experimentation is encouraged.
Leaders are moving ahead in their personal adoption of AI, while their organizations lag behind. Contrary to past technology trends, senior staff are more likely to use AI tools than their more junior counterparts. This creates knowledge gaps, unrealistic expectations, and potential frustration. Organizations need to catch up, but thoughtfully.
The barriers aren’t primarily technical. They’re cultural, organizational, and personal. Success requires change management, training, clear communication, and patience, not just better tools.
The utilities seeing real results aren’t rushing headlong into AI. They’re taking time to understand their processes, identify high-value use cases, and build appropriate guardrails. This measured approach saves staff time and avoids costly missteps that come from hasty implementation.
The utilities succeeding with AI share common characteristics:
One executive captured the realistic timeline perfectly: “I think it’s going to be a significant tool, absolutely. Transformative? Maybe. It’s like when the internet first became available to us… It took a while to see the full picture. I see this playing out over the next 3-5 years in a big way.”
This research represents the most comprehensive look at AI adoption in public utilities to date, but it’s just the beginning of the conversation. We’re planning to share the complete findings, including detailed benchmarking data, policy examples, and implementation frameworks, with all participants and interested utilities. Participants have also been invited to join us in March of 2026 at the Utility Management Conference in Charlotte, NC.
Want to dive deeper? We’re offering guidance on assembling an AI Use Policy to utility leaders who want to understand the guardrails and guidance surrounding AI use that make sense for their organization’s unique situation. We can also provide assistance with developing an AI strategic road map or create a tool for your organization like our in house AI tool, Rafio, to prevent utility data from going into training AI models.
The AI transformation isn’t coming to public utilities. It’s already here. The question is: will your organization be ready?
Based on in-depth interviews with 18 public utility leaders conducted in September-October 2025. To participate in the AI user group, for more information about our AI Advisory Services, or to discuss your AI Use Policy, contact Chris McPhee at cmcphee@raftelis.com.
