In 2023, U.S. data centers consumed around 176 terawatt-hours (TWh) of electricity, accounting for roughly 4.4% of the country’s total electricity consumption. With the rapid increase in artificial intelligence (AI) usage, this figure is expected to skyrocket, possibly doubling or even tripling to exceed 10% of total U.S. electricity demand by 2028, according to the Department of Energy. This rapid onset of AI demand is forcing a major power shift in the U.S. and more broadly in North America. The scaling of AI, coupled with the buildout of data centers, is overwhelming a grid nearing peak demand.
However, the culprit behind the demand for load growth could also be its hero. While not all AI applications are helpful, understanding the difference between energy-intensive models that strain the grid versus energy-native solutions will be key to unlocking monumental benefits for infrastructure resilience and lower ratepayer bills.
It’s a delicate dance to be certain. As this situation evolves, transitioning from challenge to opportunity is crucial. It’s vital for all grid stakeholders and ratepayers alike to know how AI can be a helpful tool in the toolshed, not one that causes an additional burden.
Grid Operators are Embracing AI
The grid is now increasingly complex, with more data streams and variables than ever before. It’s a potent one-two punch that is pushing grid operators and utility executives to explore the potential efficiencies gained by deploying AI tools. They’re less certain about where AI will have the greatest impact and which tools stand up to the rigor of managing critical infrastructure.
Having to weave power from multiple sources, such as distributed energy resources (DERs) like solar, wind, geothermal, and even standard gas and coal, has created an energy ecosystem that is the most diverse in human history. The output and data that each energy source provides also cause backend logistical challenges and information overload inside control rooms, making it difficult for current grid operations systems to keep up. According to Wood Mackenzie, DER market growth is anticipated to double between 2022 and 2027, despite recent policy headwinds, with the market reaching approximately $68 billion in the United States alone. This projected growth is only adding to the already existing complexity grid operators must manage, especially when combined with unpredictable demand cycles.
As utilities modernize, they must move toward a more coordinated model that connects “grid edge” activity with operational systems and energy markets. The premise of planning and operating the grid holistically is not new—but it’s past time that the “One Grid” theme graduates from conference breakout session fodder into utility boardrooms. Executing that vision is not possible with traditional processes, however, as the demands facing grid operators are too great. There’s little doubt that AI will play a critical role in optimizing resources and streamlining decisions in the not-so-distant future.
AI on the grid vs. AI for the grid
Many of the most advanced AI systems were built for mass market adoption, not critical infrastructure. For an industry that thrives on acronyms, opaque processes, and inconsistent terminology, general-purpose AI offerings often fall short of expectations or are relegated to low-risk use cases. On the other hand, energy-native AI is built from the outset for the industry—trained and tuned to understand and anticipate the needs of grid operators.
As you likely know, AI is only as good as its underlying data. Energy-native AI doesn’t just recognize keywords or phrases but incorporates historical operational evidence into alerts, suggestions, and automated tasks.
'This is your grandmother’s grid.'
The grid won’t undergo the massive infrastructure overhaul it desperately needs anytime soon. The costs, manpower, and time required to overhaul a system comprising over 5.5 million miles of local distribution lines and roughly 160,000 to 450,000 miles of high-voltage transmission lines would be astronomical. Pair that with lean budgets and the urgency of mitigating short-term load growth challenges, and the wait to overhaul the grid is inherently impossible. Advanced digital solutions can make the grid immediately more reliable and efficient.
So, unfortunately, “this is your grandmother’s grid,” and it won’t change structurally anytime soon. But what can change is the prioritization from utilities around integrating energy-native AI solutions that target both the modern demand of data centers and the complexity that grid operators are facing as more energy sources come into play.
