Voters punished the data-centre boom as a kitchen-table energy issue. Now Democrats, preparing to take both chambers of Congress in January, are trying to convert that anger into two tracks of legislation: near-term rules meant to stop AI server farms from pushing household electricity bills higher, and harder AI-safety and liability proposals that run straight into a presidential veto. The first track has political payoff. The second has hearings, subpoenas and must-pass bills as its real leverage. Distinguishing what Congress can force from what it can only symbolise is the central question of the next session.
A wire-service report this week captured the scale of the Democratic plan. Lawmakers, aides and advocates said the party intends to introduce a flood of bills once it controls the committees, aimed at shielding consumers from rising electricity costs tied to data centres, requiring independent testing of advanced AI systems, and clarifying when AI companies can be held liable for harm. That agenda is a sharp break from the Republican-led Congress that largely promoted AI development and declined to pass restrictions even as public opinion turned against the facilities and their power demand.
President Donald Trump has repeatedly dismissed calls for mandatory AI rules. On Friday he said he would work with states that "want to become RICH" by embracing AI, without offering further detail. Any freestanding safety bill that reaches his desk faces steep odds. That is why Democratic strategists talk less about signing ceremonies and more about committee platforms, sworn testimony and the annual appropriations and authorisation vehicles that must pass regardless of who sits in the White House.
The electricity track is where the midterm mandate is clearest. Senator Chuck Schumer, the New York Democrat expected to become majority leader, said on Thursday at a Washington event that protecting consumers from higher energy costs linked to the rapid expansion of data centres would be among his priorities next year. Representative Greg Casar of Texas, chair of the Congressional Progressive Caucus, said this week that AI executives should be called to testify under oath. Casar framed the politics bluntly: "The revolt against data centers was one of the top issues of the 2026 election, and in a few years we will look back on it as just the early tremor of a much bigger AI earthquake upending American politics."
That revolt already produced concrete legislative text. Representatives Rob Menendez of New Jersey and Casar introduced the Preventing Rate Inflation in Consumer Energy Act, known as the PRICE Act, and a companion Data Center Transparency Act. Menendez said New Jersey families had seen electricity bills rise by about 20 percent because of energy-intensive AI data centres. Casar put the fairness argument in a single sentence: "Families should not pay more for utilities so that billionaires can get rich off AI data centers."
The underlying numbers in their press materials are designed for hearings. US electricity consumption hit a record high in 2024, with data centres accounting for more than 4 percent of national electricity use. The Independent Market Monitor for PJM, the largest grid operator in the country, found that data-centre load growth was the primary reason for higher energy prices in the PJM region. The PRICE Act would require data centres to generate the electricity they consume, reach 75 percent renewable generation by 2035 and 100 percent by 2040, and face civil penalties of up to $100,000 per day for violations. The Transparency Act would require the Environmental Protection Agency to publish quarterly environmental data on water, air pollution and greenhouse gases, and the Energy Information Administration to report every six months on energy consumption and household bill impacts.
Those bills do not need to become law in their current form to change the industry's incentives. They set a negotiating floor for Democrats who will soon chair Energy and Commerce and related panels. Parallel efforts already point the same direction. Representative Lori Trahan of Massachusetts, a member of Energy and Commerce, has backed ratepayer-protection measures that would force large-load customers, including data centres, to cover grid-upgrade costs and provide financial guarantees so ordinary households are not stuck with stranded infrastructure bills. The political logic is simple: voters feel the power bill before they feel an abstract debate about model evaluation.
The safety track is broader, harder and more exposed to the veto. Trahan plans legislation requiring independent experts to evaluate leading AI models against safety standards, and separate language clarifying that companies can be held responsible when AI systems act in harmful ways, according to Francis Grubar, her deputy chief of staff. Representative Zoe Lofgren of California, seeking to chair the House Science Committee, said Democrats would focus on "responsible regulations for AI while also fostering innovation." In October, Lofgren and other Democratic House committee leaders asked OpenAI, Anthropic, Meta, Google and SpaceX to detail their security practices and report instances in which their tools took actions designers did not intend.
Incoming senators and House members elected this week have gone further still. Abdul El-Sayed, who won Michigan's Senate race, called during his campaign for an independent AI oversight agency and has argued for a "full-on pause" on frontier AI research, a treaty approach with China, and a ban on recursive self-improvement. Senator Jon Ossoff of Georgia, in his victory speech, said: "The titans of Silicon Valley dig bunkers for themselves and warn that the artificial intelligence they're creating risks human extinction." Scott Wiener, who won the California House seat long held by Nancy Pelosi, has said he will push federal safety standards and catastrophic-risk guardrails. Representative-elect Manny Rutinel of Colorado has cited a roughly 10 percent civilisation wipeout risk from the technology, a figure he attributed in a published interview to his own assessment of the hazard.
Senator Ed Markey of Massachusetts, re-elected this week, previewed the oversight style Democrats intend to use. In a post online he wrote that tech executives "are going to need to plant new forests for all of the subpoenas we'll be printing." Hearings do not rewrite the statute books, but they force disclosure, freeze voluntary commitments into the public record and raise the political cost of dismissing safety incidents. Analysts who follow the sector say summer AI security failures already shifted policymakers from abstract talk to concrete questions about monitoring, incident reporting and release decisions. The hearing room is where that shift becomes visible.
Senator Maria Cantwell of Washington, slated to chair the Senate Commerce Committee, has published a governing framework she has called a precursor to legislation. It would require mandatory safety standards for cutting-edge models, independent testing and audits before release, and prompt reporting of safety incidents. Brad Carson, who leads Americans for Responsible AI and previously served in the House, has argued that every Democrat supports that approach and that five or six Republican votes could be enough to clear a filibuster. Whether those Republican votes materialise once the White House applies pressure is the open variable of 2027.
Markets have already priced some of the risk. Technology shares sold off sharply on Thursday on concerns that a Democratic Congress would slow AI development and that capital spending on the buildout might not earn sufficient returns. The slide continued Friday in Asia and Europe, though Wall Street futures pointed to a modest Nasdaq bounce later that day. Ed Mills, who manages policy research at Raymond James, told a wire service that "the probability of legislation is much higher than people anticipate." He noted that Congress can override a veto with two-thirds majorities in both chambers, and that Democrats could also try to insert AI rules into routine annual bills that must pass.
Override maths is unforgiving. Even with Democratic majorities, assembling two-thirds support for binding safety mandates that the president opposes is a high bar. Insertion into must-pass vehicles is more plausible for consumer-facing electricity protections than for sweeping liability or pause regimes. That asymmetry matters. Democrats can more readily force data centres to internalise grid costs, disclose environmental footprints and face ratepayer-protection standards than they can compel a Trump White House to accept independent pre-release audits of frontier models. Symbolism and substance will travel together in press conferences. They will diverge in conference committees.
Statehouses remain a parallel battlefield. Forty ballot measures to slow or stop data centres passed in this cycle, according to a public analysis of the midterms. An exit poll found that 56 percent of respondents said AI and data centres were a crucial or major factor in their House vote. In Texas, governor-elect Gina Hinojosa has pledged to keep a halt on new data-centre projects until the legislature writes laws that shield residents and give communities a real role in approvals. Those state moves do not replace federal statute, but they raise the cost of shopping for the most permissive jurisdiction and give Democratic chairs ready case studies for hearings.
Public opinion already tilts toward regulation. An October wire-service poll found that most US voters think the Trump administration and Congress are not taking AI risks seriously enough and want new rules. A crisis that produces clear, attributable harm could expand that majority into veto-proof territory. Democrats know this. So do industry lobbyists. The interim strategy for the majority party is therefore not to pretend a safety statute will be signed in March. It is to lock in electricity-cost wins that voters can feel, haul executives into sworn hearings, and keep drafting liability and testing language that is ready if events, markets or a handful of Republican defections change the veto calculus.
None of this displaces the other opening fights of a Democratic Congress, from health coverage to tariffs to foreign policy. AI is a parallel track, not the whole agenda. But the midterm message on data centres was unusually clear for a technology issue: voters treated server farms as power bills, water use and local consent, not as abstract futurism. Democrats now have the committees to act on that message. What they can actually pass on AI depends less on the ambition of their speeches than on which pieces of that agenda can survive contact with a president who still wants states to get rich by racing ahead.