Over the past several months, the shift in U.S. EV policy has attracted increasing attention. The public narrative has largely focused on politics: changes in leadership with President Trump taking office, the dismantling of EV policies, the reported influence of the oil industry, tariff implications and consumer concerns around the affordability of EVs.
However, there is a deeper story unfolding here: grid capacity and constraints and making room for the rise of artificial intelligence (AI). The U.S. is rapidly entering a period in which grid resources must be allocated carefully. And EVs, once a top policy priority, are now in direct competition with the rise of AI and its energy needs.
The Power Demands of EVs Were Real and Rising
Under the Biden Administration’s finalized vehicle GHG standards, automakers were expected to reach up to 56 percent BEV sales by 2032. If that target had been met, it would have resulted in roughly 45 million electric vehicles on U.S. roads by that year.
Each EV typically consumes between 3,000 and 4,000 kilowatt-hours (kWh) annually, depending on driving behavior and efficiency (and without other assumptions on managed charging). Using a conservative estimate of 3,750 kWh per vehicle per year, those 45 million EVs would have required approximately 169 terawatt-hours (TWh) of electricity annually by 2032. That amounts to about 3.5 to 4 percent of total projected U.S. electricity demand in that year.
In the California-led states that adopted the Advanced Clean Cars II (ACCII) program, the grid implications were even more pronounced. The ACCII policy mandated 100 percent zero-emission light-duty vehicle sales by 2035. The Section 177 states that followed California’s lead represented roughly 40 percent of all U.S. new vehicle sales. Late last month, Congress overturned California’s waiver from the U.S. Environmental Protection Agency (EPA). Had ACCII been fully implemented and successful, these states would likely have had around 60 million EVs in operation by 2035.
At 3,750 kWh per vehicle, those 60 million EVs would have required approximately 225 TWh of electricity per year, shown in the figure below.
That is equivalent to more than 28 percent of California’s current electricity demand. On a national basis, the ACCII states alone would have consumed the equivalent of 4.5 percent of total U.S. power. Incidentally, this does not include additional demand on the grid that would have been created by electric trucks under the California’s Advanced Clean Fleet and Trucks programs (ACF, ACT, respectively) that other states had also planned to implement.
These projections are not marginal. EVs were on track to become one of the largest new sources of electricity demand in the U.S. over the next decade, rivaling industrial sectors and building electrification.
AI Enters the Picture and Alters the Trajectory
Until recently, these projections were considered ambitious but achievable. Then came the AI infrastructure buildout boom. In just two years, demand for AI computing power has outpaced grid planners’ expectations. In 2022, data centers consumed roughly 4 percent of total U.S. electricity. By 2030, that figure is expected to double, reaching 8 to 10 percent, with some forecasts going even higher depending on cooling intensity, chip architecture, and continued AI scaling.
The graphic below from Visual Capitalist shows that, in absolute terms, U.S. data center energy demand is set to jump from 224 terawatt-hours in 2025 to 606 terawatt-hours in 2030. Though the U.S. has experienced relatively flat power demand since 2007, data center growth alone could account for 30-40% of all net-new electricity demand through 2030.
If current growth trajectories hold, AI data centers could consume more power than the entire light-duty EV sector by 2032, even under Biden-era policy assumptions.
Already, utilities are reprioritizing interconnection queues to serve AI loads first. In some cases, EV charging infrastructure is being delayed or scaled back, not just because of technology failure or policy reversal, but because the available grid capacity is being reallocated.
Behind the scenes, data center developers are lobbying for priority access to grid resources. Some federal and state regulators are sympathetic. The narrative around AI has shifted from one of technological novelty to one of strategic necessity. From a national security perspective, the urgency to build AI capacity before China closes the gap has become a central concern. Electricity is no longer just an environmental issue—it is a critical asset in the AI race.
Strategic Electrification: A New Framework for Energy Policy
The implications are clear. The U.S. is moving toward a model of strategic electrification, where not all electric loads are treated equally. Sectors that are considered vital to national security, technological dominance, or economic leadership are being prioritized. EVs, once seen as the flagship of the energy transition, are increasingly viewed as flexible or deferrable.
To be clear, this does not mean that EVs will disappear. But the pace of adoption is slowing. Public charging deployment under NEVI has stalled. The Inflation Reduction Act’s tax credit framework has become more complicated and is under legislative threat. Several states that had initially adopted California’s ACCII rule are now reconsidering or delaying implementation. The Trump Administration plans to roll back both the Biden-era vehicle GHG and fuel economy standard programs.
We are entering a period in which AI and EVs are no longer complementary infrastructure bets. They are competitors for the same electrons.
AI Driving U.S. Energy Policy?
Increasingly, this reprioritization is being reflected not just in market behavior but in formal energy policy. Federal and state regulators are now linking the need for rapid AI infrastructure growth with expanded access to firm, dispatchable power. Nuclear energy, long viewed as politically and economically difficult, is being repositioned as a critical source of electricity for AI data centers. Companies like Microsoft, Amazon and Google are exploring direct procurement of small modular reactors to support their compute loads.
Hydrogen is also being reframed by the U.S. Department of Energy as a tool for grid resilience, especially in regions where AI-related electricity demand is surging. Even grid planning mandates from the Federal Energy Regulatory Commission (FERC) and state commissions are beginning to cite AI as a strategic load that must be prioritized.
So, this is no longer about decarbonization. It is about restructuring the grid to support a new generation of infrastructure critical to national security and global technological leadership. In that emerging framework, electrifying every vehicle is no longer the primary objective right now. Winning the AI race is.
What This Means for the Fuel Landscape
For clean fuels stakeholders, the implications are significant. The idea that electrification would serve as the sole pathway for decarbonizing transportation is starting to give way to a more complex reality. This shift creates space for a more adaptable and resilient approach to transport energy. Electrification will remain important, but it will need to coexist with low-carbon fuels, hybrid vehicles, and other solutions that require less infrastructure and can respond more flexibly to regional and political constraints.
Need help understanding how these shifting dynamics—and the legal and regulatory risks behind them—impact your business?
I provide the strategic insight companies need to navigate change, avoid costly missteps, and protect investments. 👉 Learn more here.
