America’s power producers face growing pressure to do more with less. A rapidly evolving grid, increasing demand, aging infrastructure, and policy uncertainty have created a system where traditional approaches to reliability are no longer enough.
The North American Electric Reliability Corporation (NERC) recently issued its 2025 RISC report, highlighting the leading risks facing America’s power sector. NERC warns that outdated planning assumptions, infrastructure interdependencies, and unpredictable weather are stretching the system beyond what it was designed to handle.
COMMENTARY
This is also coming at a time when President Trump has emphasized the grid’s importance to America’s artificial intelligence (AI) industry and national security.
And yet, some grid operators are relying on legacy methods, making dispatch decisions in silos without seeing the larger picture, and need to spend resources inspecting aging infrastructure. This approach must change to meet the growing power demand from building electrification, data centers, factories, and electric vehicles.
Power producers have a generational opportunity to use AI to drive intelligence-driven, anticipatory operations.
That’s where knowledge graph–based systems come in. Unlike traditional analytics tools that operate within single datasets or pre-programmed rules, graph-based AI connects the dots across a wide array of operational inputs (sensor data, outage records, fuel logistics, weather forecasts, maintenance logs, market pricing data, etc). By adopting trustworthy AI systems, leaders can identify subtle patterns, anticipate stress points, and support predictive decision-making.
NERC’s recent report emphasized the rising risk of interdependent infrastructure failure, such as a gas supply interruption that could cascade into generation outages. AI platforms that understand these cross-sector linkages help plant operators see around corners, identifying vulnerabilities and enabling smarter resource allocation before risks materialize.
Or consider predictive maintenance. AI systems that fuse real-time inputs with historical failure trends and environmental conditions can flag issues with accuracy and direct human crews to where they’ll have the most impact. This has been used effectively in the oil and gas industry for offshore drilling to help managers know when their equipment needs to be replaced.
This is not about replacing human expertise. If a plant entirely outsourced its thinking to an AI model, NERC would face an entirely new set of risks around AI hallucination-enabled power crises.
Instead, the best AI platforms are those that let users explore the “why” behind a recommendation, run accurate scenarios based on available knowledge, and compare the AI’s data with their own experience.
This is another advantage of the knowledge-graph-based approach. Analysts can dissect exactly how their AI reached a particular conclusion or recommendation by following the literal connections between data that the AI used. In comparison, a traditional database could spit out an answer without a source, and the user is unable to verify if it is accurate.
Large Language Models that keep their data in a “black box” have had a history of prominent incidents where the models have been blamed for potentially hallucinating everything from scientific papers that do not exist, to fake medical codes for hospitals, to made-up quotes submitted in a legal brief, among many other examples.
Blindly trusting AI is not the answer, but the power generation industry can leverage the energy it produces for AI to make the very generators that fuel it even smarter. Without this strategic approach, companies miss out on reaping the rewards of their generators powering massive AI data centers.
We’ve entered an era where siloed systems no longer suffice. If power producers want to stay ahead of aging infrastructure, growing demand, and unpredictable stressors like supply chain disruptions for fuel sources such as natural gas or coal, they need to invest in AI.
Explainable and traceable AI can help grid operators create an environment where human insight and machine intelligence reinforce each other. This symbiotic relationship adds an entirely new layer of resilience to the power grid, transforming how we think about energy generation and management.
—Jon Brewton is CEO of Data Squared.
