After decades of wear and tear, the US energy grid is beginning to reach its split point.
Traditional grid systems were designed in an era when electricity demand was stable and not very intense. Today, the grid faces an unprecedented mountain of stressors. Acceleration of electrification, the surge in energy-hungry AI infrastructures like data centers, and the reinforced impacts of extreme weather driven by climate change have pushed grid infrastructure to its limits and destabilized already vulnerable energy systems.
According to the US Department of Energy, these disruptions could cost American businesses up to $150 billion each year, leaving potentially millions of consumers without reliable electricity.
In response, utilities are beginning to carefully embrace artificial intelligence as a tool to stabilize grid operations. Once considered merely an emerging technology, AI has emerged as part of a utility strategy to predict disruption, protect grid equipment, and better provide a rapidly changing energy environment.
AI is stepping up predictive maintenance
AI sharpens the proven tools that the utility uses to prevent failures across the energy grid.
One is predictive maintenance. This maintenance uses sensors and software to monitor the condition of grid equipment, such as transformers and power lines, and find issues to fix before it goes down.
Predictive maintenance is nothing new. However, according to Somjyoti Mukherjee, consulting partner at IT consulting firm Cognizant, the integration of machine learning into the process has made it faster and more accurate to detect failed equipment.
Sensors embedded in circuit breakers, switchgears, and transmission lines send real-time data to an AI system, analyzing patterns and predicting when components are likely to break down.
“Predictive maintenance offers the fastest returns,” Muhaji, who leads the grid modernization efforts in the utility sector in North America, told Business Insider.
Murkherjee pointed to one utility client with an outdated system that left a crew of field techniques wasted hours of daily waste because he couldn't capture the problem in time. After switching to an AI-driven maintenance system, the software will suggest recommended tools, equipment replacements, position defects in real time, allowing crews to work “smart and faster,” Murkherjee said.
American energy provider Duke Energy is also using AI to identify vulnerabilities in the grid. The Fortune 500 Utility Provider has developed a hybrid AI system that blends machine learning and expert diagnostics to flag high-risk devices. The tool is designed to monitor the health of Duke's Transfleet, a connected web of circuits that send electricity from one board to another.
Matt Carrara, president of Doble Engineering, combined Duke's hybrid approach with AI-powered insights that combines AI expertise with AI-powered insights, led to “more consistent identification of problematic equipment” and “improvement of planning decisions.”
Matt Carrara is the president of Doble Engineering. Courtesy of Doble Engineering
Some startups are pushing more AI capabilities.
Rhizome is working with Seattle City Light, the Vermont Utilities and other US grid operators to map climate-driven risks before an attack. Co-founded by CEO Mishal Thadani, the platform uses AI to analyze historical grid data, causes of outages, and environmental threats such as wildfires, storms and vegetation growth, to the level of individual poles and wires.
The result is a digital risk map, which guides where you can invest in upgrades and maintenance for the greatest impact per dollar. For example, one Texas utility used Rhizome's prediction model to identify which circuits in the energy system are at high risk of storm shocks, allowing capital to be invested in improving equipment with vulnerable utility. By doing so, Texas utility reduced storm outages by 72%. According to Rhizome.
With utilities facing tighter budgets, rising insurance costs, increasing pressure from climate change and electricity-hungry data centers, Thadani said platforms like Rhizome will help them make more strategic investments in improving the grid.
“More utilities need to be very aware of the investment they are making,” Thadani told BI, adding that big capital decisions “need to justify data and evidence to show the value of rate wages.”
Mishal Tadani is co-founder and CEO of Rhizome. Provided by Rhizome
Energy providers are exploring new AI tools
Beyond maintenance, the utilities employ new AI tools to better understand and manage physical equipment in the field.
Peter Meaning, principal advisor at Stantec, an engineering consulting group, pointed to one of the company's utility clients who deploy cameras with image recognition to automatically capture, identify and digitize device data. Doing so has improved the quality and speed of data collection, less time spent collecting Intel, improved equipment fleet decisions, and as a result, fewer visits to manual sites.
Implementing computer vision technology in the grid is part of a larger shift towards using AI for pattern recognition and data-heavy tasks, such as forecasting demand, mapping outages, and streamlining upgrades.
“This is where AI shines,” the nearby told BI about the technology's data processing capabilities.
Some utilities rely on generated AI to simplify fieldwork. In March, US renewable energy supplier Avangrid launched First Time Right Autopilot, a Genai tool trained with the company's internal manuals, troubleshooting guides and other internal documents. With access to your mobile device via voice or text, chatbots can answer technician repair questions in real time.
For example, if the wind turbine goes offline, the technician can ask the AI assistant how to fix it. This tool uses the contextual data of the turbine equipment to analyze problems and provide step-by-step instructions.
According to Nelly Jefferson, Chief Information Officer of Avangrid, since implementing AI tools, Avangrid has been repaired faster and reduced downtime.
“It strengthens our workforce by providing real-time access to professional level support for field technicians,” Jefferson told BI.
Still, given the outdated infrastructure of the grid, managing energy demand – especially at peak times, is a feat that AI is hard to deal with. That's why most utilities are still in pilot mode when it comes to AI-driven load management, according to Vivian Lee, managing director of the Boston Consulting Group, with expertise in the energy sector.
According to Lee, some people are experimenting with short-term load forecasts to predict the time or days ahead of electricity demand using real-time data such as weather, usage trends and local events. Others test AI to control distributed energy resources such as smart thermostats, EV chargers and household batteries to slightly reduce or shift energy usage during high demand periods and mitigate strain on the grid.
These tools remain primarily rule-based and only work when given instructions for a particular use case, limiting a wide range of applications. However, Lee sees the long-term potential of managing energy loads with AI.
“The widespread adoption of AI in load management is still in the early days,” Lee told BI.
Near Peter is Stantec's Principal Advisor. Provided by Stantec
Obstacles are preventing AI adoption
Despite growing optimism, an energy expert who spoke to BI said utility companies find AI difficult to employ.
Many still use Legacy IT and operational systems that are not easily integrated, making it difficult for AI to put together clean, usable data to elicit insights.
“Data quality and availability remains a major hurdle,” Lee said.
A lack of talent increases friction. Lack of AI literacy across employees could make organizations more resistant to embracing new technology, according to nearby.
Regulation bottlenecks make that transition even more difficult. The lack of clear guidelines for AI deployment in the energy sector creates hesitancy among utilities, which must navigate legal frameworks to ensure that AI applications are compliant with data protection laws.
The Trump administration's tariffs on imported components such as trans and metals have also skyrocketed costs, further complicating the project's timeline, Carrara said.
Utilities are careful and rely on AI
Still, many of these obstacles do not break the transaction anymore. Mukherjee says the utility has made progress by moving to the cloud, training employees to use AI, engaging with regulators, and driving technological change.
“Regulators are responding,” Muhaji said, pointing to agencies like the US Federal Energy Regulatory Commission, which employs technical experts and “leaning towards innovation.”
Building trust is also important. Mukherjee is nearby, and Lee focused on low-risk, explainable use cases to build small momentum, emphasizing that from day one, he began small, involving frontline workers.
Looking ahead, utility experts say energy providers are keen to continue investigating the possibilities of AI to modernize grids to reduce strain.
But they have a long way to go before they can fully embrace AI with their arms crossed.
“AI does not replace core grid functionality,” Lee said. “But it will become increasingly functioning as an accelerator.”
