The Digital Rust: Why Utility Operations are Stalling in the AI Era

For decades, the utility sector has operated on a "reliability-first" mantra, a philosophy that ensures the lights stay on but can often leave internal innovation in the dark. While global headlines focus on the staggering electricity demands of AI data centers, a quieter, more systemic issue is brewing: utilities are failing to integrate artificial intelligence into their own internal business operations at the pace required to manage a modern grid. In order to utilities to continue maintiaining reliability and serving the growing demand, they will need to accelerate the adoption of AI internally.

Despite having access to mountains of operational data, many utilities remain tethered to legacy processes. While other sectors have moved toward seamless automation, the utility industry's internal adoption of AI remains a patchwork of isolated pilots rather than a cohesive digital strategy.

I recently attended Distributech, an annual electric utility innovation conference, where I expected to see the advanced development of AI use cases within the utility sector. But I was disappointed in the relative adoption of many of the nation's largest utilities compared with other industries like manufacturing, retail, and oil and gas. The expo floor was packed with relavant, impactful use cases from the world's largest tech vendors, but the sessions were detail-light in their implementation of AI.

The utility industry is partly limited by the regulation necessary to keep rates at a stable level and to keep utility companies healthy enough for cheap capital. However, the pace of progress will create a turning point for tech companies as they continue to push utilities to move faster to keep up with the larger AI evolution.

The Pockets of Progress: Copilots and Computer Vision

It would be unfair to say utilities have ignored AI entirely. There are clear success stories where the technology has proven its worth in specific niches:

• AI Copilots for Analysts: Forward-thinking utilities have begun deploying AI copilots to help back-office analysts navigate decades of dense regulatory filings, technical manuals, and historical maintenance logs. Instead of spending hours cross-referencing PDFs, analysts use natural language queries to retrieve critical data, significantly accelerating decision-making cycles.

• Computer Vision for Asset Health: One of the most mature applications is the use of image-recognition AI to inspect infrastructure. By feeding thousands of drone and satellite images into machine learning models, utilities can now automatically detect cracks in insulators, leaning poles, or encroaching vegetation on high-voltage wires. This "predictive eyes" approach moves the needle from reactive repairs to proactive prevention.

• Predictive Maintenance: Beyond visuals, AI is being used to analyze sensor data from transformers and gas pipelines to predict equipment failure before it occurs, saving millions in emergency "truck rolls."

The Agentic Gap: From Insight to Action

While these tools are impressive, they represent "Passive AI"—systems that provide information for a human to act upon. The next frontier is Agentic AI, where autonomous agents can plan, reason, and execute complex workflows across different software systems without constant human hand-holding.

Imagine an AI agent that doesn't just detect a faulty transformer but also checks inventory for a replacement, schedules the specialized repair crew, notifies affected customers via text, and updates the regulatory compliance log—all in seconds.

Current adoption of agentic AI within utilities is virtually non-existent. The industry is hamstrung by a silo cultural inertia and a fragmented data landscape where IT (Information Technology) and OT (Operational Technology) systems rarely speak the same language. As a result, most utilities are stuck in the "pilot purgatory" phase, unable to scale these autonomous capabilities across the enterprise

The Tech Takeover: A Catalyst for Change?

The slow internal evolution of traditional utilities has not gone unnoticed by Big Tech. In a series of recent, high-profile moves, technology giants and "hyperscalers" have begun directly purchasing or taking massive stakes in energy companies and power generation assets. Alphabet has recently acquired Intersect Power for $4.75B, Amazon and Google have sold more than $2.7B in electricity on the wholesale market, acting almost as power traders within the industry, and a shift to an "All of the Above" strategy for Amazon, Google, and Meta to acquire capacity for data center infrastructure.

These shifts are driven by a dual motive. While securing energy for their data centers is a primary goal, these tech companies are also bringing their "AI-first" DNA to the utility's internal operations. By applying Silicon Valley’s software-defined approach to the grid’s "iron and wire," these new owners may finally force the digital transformation that the industry has long awaited.

Snapshot of AI Maturity in Utilities

The risk for traditional utilities is no longer just a slow digital transition; it is the risk of becoming obsolete in the face of a more agile, tech-driven competitor. Until utilities treat AI not as a peripheral tool but as the core operating system of their business, the "digital rust" will continue to spread.