The worst price outcomes are not inevitable—they result from policy choices
As Americans search for someone to blame for rising electricity bills, researchers at Columbia University's Center on Global Energy Policy offer a quieter, more demanding truth: the power grid is straining not under the weight of any single innovation, but beneath decades of deferred decisions, regulatory friction, and infrastructure left to age. AI and data centers are part of a growing demand story, but the deeper crisis belongs to the system itself — its planning failures, its permitting bottlenecks, its inability to keep pace with the complexity of modern life. The finding is both a corrective and an invitation, asking policymakers to set aside convenient narratives and engage with the harder, structural work that actually lies ahead.
- Public anger over electricity bills has found a ready target in AI and data centers, but that narrative may be directing political energy away from the real problem.
- Columbia researchers found that aging infrastructure, permitting gridlock, and weak grid planning are the true engines of price acceleration — a system under strain from multiple directions at once.
- Paradoxically, in some regions large new electrical loads have actually pushed prices down, complicating the story further and demanding more nuanced policy responses.
- A roadmap for relief exists: near-term grid optimization, medium-term regulatory reform, and long-term risk management for storms, wildfires, and cybersecurity threats.
- The research signals that the conversation can now shift — from assigning blame to doing the slower, less dramatic work of fixing what is structurally broken.
The public debate over rising electricity prices has found a familiar villain in artificial intelligence and the data centers that sustain it. But new research from Columbia University's Center on Global Energy Policy complicates that story considerably.
Drawing on a year of roundtable discussions with regulators, utility executives, academics, and technology industry representatives, Columbia researchers found that price increases stem from a tangle of structural problems: aging infrastructure, permitting and grid-planning bottlenecks, mounting costs for system upgrades, and broadly rising demand across the economy. Data centers and AI do contribute to demand growth, the research acknowledges — but they are not the principal driver of recent price increases. In some regions, large new electrical loads have actually brought prices down.
Founding director Jason Bordoff described the findings as a corrective to oversimplified public discourse, noting that policymakers need to understand what they are actually trying to fix before they can fix it.
The research produced two publications. The first, by Robin Millican, Douglas Arent, and David Sandalow, maps solutions across three time horizons: near-term optimization of existing infrastructure through grid-enhancing technologies; medium-term regulatory reforms to utility incentives, cost allocation, and permitting processes; and long-term strategies for managing risks from extreme weather and cybersecurity threats. The second publication summarizes the roundtable findings, where participants broadly agreed that price inflation is accelerating and will likely continue — driven not by any single technology, but by a system stressed from multiple directions simultaneously.
The implication for policymakers and ratepayers is uncomfortable but clarifying: there is no single lever to pull. Bringing electricity prices under control requires sustained attention to infrastructure, process, and incentives across the entire grid — the kind of unglamorous, systemic work that rarely makes headlines but ultimately determines what people pay.
The conversation about why Americans are paying more for electricity has centered on a familiar villain: artificial intelligence and the data centers that power it. But a new examination of the problem from Columbia University's Center on Global Energy Policy suggests the real story is far messier, and far less about any single culprit.
Researchers at the center, drawing on a year of roundtable discussions with regulators, utility executives, academics, policymakers, and technology industry representatives, found that electricity price increases across the country stem from a tangle of structural problems in how the power system is built, planned, and managed. Aging infrastructure, bottlenecks in the permitting and grid-planning processes, mounting costs for system upgrades, and simply more demand for electricity across the economy—these are the forces actually driving prices up. Data centers and AI do contribute to growing electricity demand, the research acknowledges. But they are not the principal driver of recent price increases, and in some regions, large new electrical loads have actually pushed prices down.
Jason Bordoff, the center's founding director, framed the finding as a corrective to oversimplified public debate. "High electricity prices have become a front-of-mind energy issue around the world, with much of the public discourse on the cause of this focused on the energy demands of AI and data centers," he said. "The Center on Global Energy Policy's new research finds that the real answer is more complicated, and takes a larger-scale look at the driving factors behind the recent rise in power prices." The timing matters. Policymakers are searching for answers, and they need to understand what they're actually trying to fix.
The research produced two major publications. The first, authored by Robin Millican, Douglas Arent, and David Sandalow, examines the structural roots of the problem and proposes solutions across three time horizons. In the near term, utilities can optimize existing infrastructure through grid-enhancing technologies and demand flexibility programs. Over the medium term, structural reforms are needed: changes to utility incentive structures, new approaches to how costs are allocated when large new electrical loads arrive, and streamlined interconnection and permitting processes. Over the longer term, the system needs to manage risks from storms, wildfires, and cybersecurity threats more effectively. The authors argue that the worst price outcomes are not inevitable—they result from policy and regulatory choices that can be addressed at both federal and state levels.
The second publication summarizes findings from the roundtable series itself. Participants from across the industry and government broadly agreed that multiple factors are affecting electricity prices simultaneously, and that large electrical loads have not been a principal driver of price increases. In fact, in some locations, they have led to lower prices. Yet participants also acknowledged that price inflation is accelerating nationally and will likely continue to rise as demand grows, including demand from large electrical loads. The picture that emerges is one of a system under strain from multiple directions at once—not broken by any single new technology, but stressed by deferred maintenance, regulatory friction, and the sheer complexity of planning for a grid that needs to serve more people and more machines than it was designed to handle.
For policymakers and ratepayers alike, the implication is clear: there is no quick fix, no single lever to pull. The work of bringing electricity prices down requires attention to infrastructure, process, and incentives across the entire system. The research suggests that conversation can now move beyond the question of whether AI is to blame, and toward the harder work of actually fixing what's broken.
Notable Quotes
The real answer is more complicated, and takes a larger-scale look at the driving factors behind the recent rise in power prices.— Jason Bordoff, Founding Director of the Center on Global Energy Policy
The Hearth Conversation Another angle on the story
So the research is saying AI data centers aren't the problem. But they do use a lot of electricity, right?
They do, and they're contributing to demand growth. But here's the thing—the roundtable participants found that large electrical loads haven't been the principal driver of price increases. In some places, they've actually made prices go down.
How does that work? More demand usually means higher prices.
It can, if the grid is constrained. But if you have spare capacity, a large new load can actually help spread fixed costs across more users, which lowers the per-unit price. The real problem is that the grid has aging infrastructure, permitting bottlenecks, and planning weaknesses that existed long before AI became a household word.
So we're blaming the wrong thing?
Not blaming—the research is saying we're looking at one tree when we should be looking at the forest. Aging infrastructure, weak grid planning, rising system investment needs, permitting delays. Those are the structural issues. AI is just the latest demand on a system that's already struggling.
What can actually be done about it?
The research lays out a three-part approach. Near term: optimize what you have through grid-enhancing tech and demand flexibility. Medium term: reform utility incentives and permitting processes. Long term: manage risks from storms, wildfires, cybersecurity. It's not sexy, but it's fixable—if policymakers treat it as a system problem rather than an AI problem.