President Donald Trump is moving to recast the economics of artificial intelligence, arguing that the companies racing to build massive data centers should also pay to expand the power grid that keeps them running. His administration is pitching a plan to shift the cost of new large-scale power plants away from ordinary ratepayers and onto tech giants whose AI ambitions are driving a surge in electricity demand. The proposal drops the fight over whether data centers are good for the economy and instead zeroes in on who should pick up the tab for the energy buildout they require.
At its core, the White House is betting that voters will accept more fossil fuel and nuclear generation if they believe Silicon Valley, not households, is footing much of the bill. The plan also tests how far the federal government can go in reshaping regional power markets and state utility rules to accommodate AI growth without triggering a backlash from industry or consumer advocates.
The political bet behind Trump’s power-plant push
Trump’s argument starts with a simple framing: artificial intelligence is straining the grid, and the companies that profit from it should pay for the extra capacity. In public remarks and policy documents, the administration has tied the rapid buildout of data centers to rising electricity demand, warning that without new generation, reliability and prices could suffer. The political calculation is that voters are more likely to accept new plants, including gas and nuclear facilities, if they see them as a way to protect their own bills from the energy appetite of AI data centers rather than as a subsidy for Big Tech, a case the White House has been making through detailed energy plans.
In that framing, Trump is not just reviving a build-big ethos for the power sector, he is also trying to reposition Republicans as defenders of residential and small-business customers against what he casts as elite technology firms. The administration has highlighted how hyperscale facilities for cloud computing and AI training can consume as much power as small cities, and it has linked that growth to the need for new baseload plants that can run around the clock. By tying the expansion of large nuclear and fossil fuel units to the AI boom, officials are arguing that the choice is not between clean energy and dirty energy, but between letting tech companies drive up everyone’s bills or forcing them to underwrite the infrastructure they need, a message that has been reinforced in briefings to investors and utilities covered in market reporting.
How the cost-shift would work for AI data centers
The administration’s emerging blueprint leans on a mix of federal pressure and market design to push more of the cost of new generation onto data center operators. Officials have floated requirements that large AI and cloud facilities sign long-term contracts or capacity agreements that effectively guarantee revenue for new plants, in exchange for priority access to power and grid connections. In practice, that would mean tech companies committing to pay for a share of a plant’s fixed costs over decades, rather than relying on utilities to socialize those expenses across all customers, an approach that has been described in detail in financial analyses of the plan.
At the same time, the White House is signaling support for utility and grid-operator proposals that would create special tariffs or cost-recovery mechanisms for very large loads, particularly AI training clusters and cloud campuses. Those tools could include higher connection fees, dedicated transmission upgrades, and bespoke rate structures that reflect the round-the-clock nature of AI workloads. The administration has framed these ideas as a way to avoid cross-subsidies, arguing that a family in Ohio or a small manufacturer in Pennsylvania should not be paying for the transformers, substations, and generation capacity needed to serve a new AI complex, a point that has been echoed in policy-focused coverage of Trump’s push to have tech firms directly finance new power plants.
PJM, governors, and the regional grid squeeze
The political urgency behind the plan is most visible in the territory of PJM Interconnection, the regional grid operator that covers parts of 13 states and the District of Columbia. PJM has warned that a wave of data center projects, especially in places like Virginia and Ohio, is colliding with plant retirements and slow transmission buildout, creating a risk of capacity shortfalls and higher prices. In response, Trump officials have worked with governors in the PJM footprint on a framework that would prioritize new generation and shift more of the associated costs to the large customers driving demand, a strategy that has been described in detail in reporting on the PJM governors agreement.
That regional focus is not accidental. PJM’s markets are a bellwether for how the grid will handle the AI surge, and the states involved span both Republican and Democratic administrations. The administration has pointed to bipartisan concern over data center-driven congestion and local rate impacts, arguing that its cost-shift plan aligns with what state leaders are already seeking. Coverage of PJM’s internal studies has highlighted how concentrated clusters of data centers can force expensive upgrades and capacity procurements, and Trump officials are using those findings to justify federal backing for new large-scale plants whose costs would be anchored by long-term contracts with AI operators, a linkage that has been underscored in accounts of the bipartisan PJM concerns.
Trump’s message to Silicon Valley and the public
Trump has been unusually explicit in his public messaging to tech leaders, casting the proposal as both a demand and an opportunity. In a recent appearance, he argued that companies building AI systems that consume enormous amounts of electricity should “pay their fair share” for the plants that keep their servers running, while also suggesting that long-term power deals could give them more certainty over energy costs. The tone has been confrontational but transactional, with the president signaling that if tech firms want faster permitting and more reliable power, they will need to help finance the infrastructure, a stance he laid out in remarks that were widely shared in video form.
For the broader public, the White House is framing the initiative as a shield against bill shocks tied to the AI boom. Officials have emphasized that without new rules, utilities might seek to recover the cost of serving data centers through general rate increases, effectively turning households into silent partners in the AI gold rush. By contrast, the administration says its approach would lock in private capital from tech companies to support new plants, while still allowing utilities to earn regulated returns on the projects. That message, which blends populist skepticism of Big Tech with a promise of grid reliability, has been central to the administration’s outreach and has been described in policy coverage of Trump’s plan to shift power costs to.
Winners, losers, and unanswered questions
If the plan advances, the clearest winners would be utilities and developers able to secure long-term contracts backed by the balance sheets of the largest technology companies. New gas and nuclear projects that might have struggled to clear traditional regulatory hurdles could become more attractive if they come with guaranteed demand from AI data centers, particularly in regions like PJM where capacity margins are tightening. At the same time, residential and small-business customers could see some protection from the most extreme bill impacts of the AI buildout, since a larger share of the cost of new plants and grid upgrades would be assigned directly to the companies driving the load growth, a dynamic that has been highlighted in analyses of Trump’s effort to make tech firms foot the bill for new capacity.
More from Morning Overview