Fix the AI Electricity 'Crisis' by Unleashing Market Forces
The AI boom is showing the limits of our regulated monopoly model for generating electricity.
The rapid growth of artificial intelligence is causing an AI power crisis, as the electricity demands of massive AI data centers are straining existing power grids and outpacing the development of new energy sources. This has led to concerns about grid stability, rising electricity costs, and significant environmental impacts. If there's not enough juice, then I'll have to write my own ledes—as those two sentences were penned by AI and powered by the local grid.
Excuse my snarkiness, as I'm not feeling too kindly toward AI this morning after being stuck on the most insane call with a "virtual assistant" who was supposed to help me get my internet service up and running but was oddly aggressive and incapable of fixing my problem. Nevertheless, I know AI is a remarkable advancement that offers huge potential. A big issue, as raised by the robot above, is whether our state has the electricity capacity to handle it.
California remains the national hub for AI development, and it's one of the main bright spots in our economy. The state's budget—now awash in an $18-billion deficit—is increasingly dependent on it. In its recent fiscal analysis, the Legislative Analyst's Office noted that income-tax collections are the one bright spot in the state economy. They are "growing at double‑digit rates" and are "driven by enthusiasm around AI, which has pushed the stock market to record highs and boosted compensation among the state's tech workers."
Obviously, the state has a vested interest in assuring that there's plenty of electricity available to power this major tax-generating sector. Meanwhile, lawmakers keep proposing—and sometimes passing—regulations that would hobble the industry. Gov. Gavin Newsom has so far rejected the worst measures, but has signed some new transparency requirements. With the new legislative session comes renewed efforts to regulate more—plus a possible ballot measure in November to impose a billionaire's tax, which will mostly affect tech titans.
If the state wants to remain competitive in the burgeoning AI world, then it needs to resist these efforts. If it gives in to them, then it will watch AI momentum shift toward Texas, Virginia, and other states that also have built significant AI infrastructure. The wrong choice will be costly in terms of the state budget. But the state's AI-related electricity challenges are a different matter, as that issue is all about planning for the future.
Of course, California doesn't have the best track record for planning for future infrastructure. For decades, state policymakers have failed to upgrade our water systems, freeways, and electricity grid. California hasn't built significant new water storage since the 1970s. Anyone who drives around the Los Angeles basin understands the inadequacy of the freeway system. The state's struggling electricity grid has been the subject of much discussion, given that state energy policy is all about moving toward EVs and other "green" electricity-dependent technologies.
My AI assistant reminds me of one emblematic example: "In September 2022, during an extreme heatwave, California's grid operator (CAISO) asked residents, including EV owners, to voluntarily conserve energy by avoiding charging during peak evening hours … to prevent grid strain and potential blackouts." And the latest reports suggest that EV charging is nothing compared to the massive demand that new data centers will impose.
Furthermore, the push for new generation runs up against the state's environmentally oriented goals. As a CalMatters report on the data-center controversy reveals, despite its 2045 total carbon-neutrality goal, the state "still depends heavily on natural-gas plants during hot summer days." It points to a recent report showing that "data-center carbon emissions nearly doubled from 2019 to 2023—largely from gas-fired generation—underscoring how even a relatively clean grid may struggle to absorb AI-driven load without higher emissions."
Something has to give—either the state budget or our aggressive climate goals. Simply put, California needs more electricity competition. My R Street Institute colleagues gave the state a C+ grade in that area—not as bad as its usual F grades for most things, but we can do better. Competition is the best way to incentivize new capacity (and improve the environment).
Communities are rebelling against the construction of massive data farms. Some opposition is based on land-use concerns, but it's also driven by fear of inadequate electricity and higher energy prices. As the Competitive Enterprise Institute's Clyde Wayne Crews explains, this is a symptom "of far deeper structural problems rooted in legacy approaches to infrastructure—approaches that tether data centers to coercive public utility price- and access-control models."
Perhaps it's time for an approach that unleashes market forces, as the AI boom is showing the limits of our regulated monopoly power model. If California continues down its usual regulated path, it's unlikely to have the capacity to build power plants that are needed to drive the industry—and pay for all that state government spending.
This column was first published in The Orange County Register.
