Democrats won majorities in both chambers of Congress in the November 3, 2026 midterm elections. The Associated Press reported that Democratic victory was assured in the Senate on election night and in the House the following morning. Democrats flipped seven Senate seats from Republicans, reaching at least 54 seats by CBS News' estimate, and NBC News projects they will secure 258 House seats. In response, semiconductor-related stocks were sold off in Tokyo on November 5, and the Nikkei Stock Average fell 1.79%.
However, a Democratic majority in Congress does not mean tighter AI regulation will advance immediately. In addition to President Trump's veto power, the Senate requires 60 votes to overcome a filibuster, so passing comprehensive federal AI legislation would require bipartisan cooperation. Rather, the AI industry may feel the effects first from other channels: the congressional investigative powers Democrats can exercise from January 2027, a harder path for efforts to override state AI rules through federal law, and moratoriums on data center construction by state and local governments.
Federal AI Legislation Faces Long Odds, but Congressional Investigations May Bite First
Enacting a federal law in the United States requires passage by both chambers and the president's signature. A bill can still become law without a signature if the president does not return it within 10 days (excluding Sundays) of receiving it, provided Congress has not adjourned in a way that prevents its return. If the president vetoes a bill, it must be passed again by a two-thirds vote in each chamber. In the Senate, ending a filibuster, which can block consideration of ordinary legislation, generally requires 60 votes.
Before the election, the Senate consisted of 53 Republicans, 45 Democrats, and 2 independents who align with the Democrats. Democrats flipped seven seats from Republicans in this election and secured at least 54 seats by CBS News' estimate, though results in Alaska and Mississippi have not yet been finalized. In the House, they are expected to far exceed the 218 seats needed for a majority, though the final seat count has not been determined. Even with control of both chambers, Democrats are not in a position to override President Trump's veto without Republican cooperation.
An analysis by Goldman Sachs economists, cited by NBC News, says the election results are unlikely to significantly change the outlook for federal data center or AI regulation, and that bipartisan support will still be needed for regulation to materialize. TIME likewise reported that, because a comprehensive federal AI law would require either Trump's signature or enough support to override his veto, broad regulation by Democrats would be difficult.
That said, room for a bipartisan deal remains. TIME noted that backlash against AI cuts across party lines, making cooperation between Democrats and Republicans possible. Brendan Steinhauser, a Republican strategist and CEO of The Alliance for Secure AI, also said AI regulation could be a chance for President Trump to achieve a bipartisan accomplishment. A federal AI law is not impossible, but it would require agreement from both parties.
There is also a timing constraint. The 120th Congress, with its Democratic majority, begins on January 3, 2027, and until then the lame-duck session continues with the existing membership. Any policy shift using the new Congress's powers will therefore get under way in earnest only after the new year.
The main channels through which the AI industry could be affected fall into the following four categories.
| Channel | Required means | When it could start | Key constraints and uncertainties |
|---|---|---|---|
| Federal AI regulation law | Passage by both chambers and the president's signature, or a veto-override vote | Depends on bipartisan agreement | Presidential veto; Senate 60-vote threshold |
| Congressional investigations and subpoenas | Committee hearings, investigations, and issuance of subpoenas | From the new Congress on January 3, 2027 | Targets and timing of testimony not yet determined |
| Preemption of state AI laws by federal law | Enactment of federal legislation, or lawsuits and administrative action under Executive Order 14365 | Legislation is at the discussion-draft stage; action under the executive order proceeds separately | Difficulty of enacting federal law; court rulings on administrative action |
| State and local data center construction pauses | Gubernatorial executive orders, state legislation, or local government decisions | Already implemented in some states; new governors' pledges take effect after inauguration | Targeted facilities and implementation vary by state |
Of these, state data center pauses and challenges to state AI laws under the executive order were already under way before the election. Congressional investigations led by Democrats become possible in January 2027, while comprehensive federal AI legislation still faces high political barriers. The election's impact on the AI industry is likely to show up first in state regulatory measures and stronger congressional oversight rather than in new federal law.
Heavy AI Industry Spending Failed to Prevent Losses, and Investigative Power Shifts to Democrats
According to Tech Influence Watch figures cited by Fortune, AI-related super PACs (political action committees) spent a total of $79 million this election cycle targeting 68 House and Senate seats. Leading the Future, which supports looser regulation, received $25 million each from Andreessen Horowitz co-founders Marc Andreessen and Ben Horowitz, and a combined $25 million from OpenAI President Greg Brockman and his wife. Fortune reported these amounts based on filings with the Federal Election Commission (FEC).
The AI industry's biggest general-election investment in a competitive race was in the South Carolina Senate contest. According to Fortune, American Mission and Building the Future, both affiliated with Leading the Future, spent $4,801,981 supporting Republican candidate Darline Graham. But Democratic candidate Annie Andrews, who called Graham "Data Center Darline" and criticized her ties to the data center industry, won. CBS News reported that this was one of the seven seats Democrats flipped from Republicans.
The spending also had distinctive features. According to TechCrunch, Leading the Future put $27.38 million into primaries, about $8 million of which went to opposing Alex Bores, co-sponsor of New York State's AI safety law, the RAISE Act. Meanwhile, an analysis of FEC filings by The Intercept, cited by Fortune, found that general-election spending stood at only $7.4 million as of October 26. On the pro-AI-safety side, according to TIME, Anthropic contributed more than $40 million to Public First Action, though that money could not be used for candidates' campaigns, and the company's employees separately donated $5 million. Brad Carson of Public First Action, in WIRED reporting cited by The Next Web, called the results a major blow to the AI industry's efforts to block regulation.
Still, individual election results alone cannot establish that AI industry donations caused the losses. Even so, losing in tightly contested races where the industry poured in heavy funding suggests that future political lobbying will not be easy. And after the election, the industry will be subject to oversight from a Democratic majority in Congress.
According to The Verge, Democrats are expected to retake committee chairmanships in both chambers in January 2027. Robert Garcia, the top Democrat on the House Oversight Committee, posted on X to "lawyer up," with the Trump administration and corporate misconduct in mind, and Ron Wyden, the senior Democrat on the Senate Finance Committee, posted only "subpoenas." Neither remark named AI companies directly. Rep. Greg Casar of Texas, however, stated outright that AI company CEOs should be required to testify under oath, and noted that opposition to data centers was one of the election's main issues. Frank Pallone, the top Democrat on the House Energy and Commerce Committee, also said he would pursue stronger regulation of Big Tech.
Unlike enacting federal law, Congress's investigative powers do not require the president's signature. Each chamber can hold hearings and conduct investigations, and committees can issue subpoenas, which legally compel testimony or the production of documents, in accordance with applicable procedures. MS NOW's report that Democrats gain subpoena authority through committees, while legislation remains constrained by the veto and Senate procedure, points to this same distinction. Whether AI company CEOs will actually be subpoenaed is undecided, but Democrats will be in a position to use that power only from January 2027.
Efforts to Override State AI Rules With Federal Law Face Headwinds
One policy Republicans have pushed is to override differing state-level AI regulations through federal law. According to the policy research site CASRAI, the budget reconciliation bill that passed the House in May 2025 included a provision freezing enforcement of state AI regulations for 10 years. But the Senate stripped the provision by an overwhelming 99-to-1 vote, and the bill was finally passed in July of that year.
Subsequently, on December 11, 2025, President Trump signed Executive Order 14365, which sets out countermeasures against state AI laws. The order directs the Attorney General to establish an AI Litigation Task Force within 30 days and to bring legal challenges against state AI laws that conflict with federal policy. It also asks the FTC (Federal Trade Commission) and FCC (Federal Communications Commission) to consider measures to limit the effect of state laws. Then, on June 4, 2026, six House members from both parties released a discussion draft of the "Great American AI Act" containing a provision to preempt state AI laws through federal law, but it has not been formally introduced as a bill.
In this way, efforts to override state AI regulation have run from the 2025 reconciliation bill to the executive order to the 2026 discussion draft. But no federal law preempting state laws has been enacted, and in January 2027 a new Congress with a Democratic majority begins. TIME also suggested that even if Democrats cannot pass broad federal AI legislation, they would find it easier to block the preemption of state AI laws that Republicans have sought.
What matters here is that an executive order and a federal statute have different effects on state law. According to CASRAI, an executive order alone cannot directly invalidate state laws; it requires tools such as litigation, agency rulemaking, and changes to the conditions attached to federal funding. Whether such measures can limit the effect of state laws depends on how courts judge their relationship to the underlying federal statutes and the Constitution. Congress could also enact an explicit preemption provision, but that would require new legislation, and any resulting law would be subject to judicial review. State laws therefore do not lose force immediately because of an executive order.
Even as the dispute between the federal and state governments continues, states' own AI regulations are advancing. In California, SB 53, which sets safety measures and transparency reporting requirements for frontier AI models, is in effect. Colorado also enacted SB 26-189 on May 14, 2026, revising its AI regulation, which is scheduled to take effect on January 1, 2027. Scott Singer of the Carnegie Endowment for International Peace has pointed out that if Congress remains deadlocked, California's regulations could shape AI governance not only in the United States but globally.
The push to override state AI laws will not stop entirely, however. Lawsuits and administrative action under Executive Order 14365 proceed independently of the makeup of Congress. If the AI Litigation Task Force sues over a state law and a court accepts its arguments, it remains possible for the law's effect to be limited without new federal legislation. What the election mainly makes harder is the path of passing a law in Congress to override state regulation.
New Governors Who Pledged Data Center Pauses Win Across Several States
The governors' races also made data center construction pauses a major issue. In Texas, Georgia, and Ohio in particular, Democratic candidates who support halting or restricting new construction won. At the same time, construction pauses had been implemented by both Democratic and Republican governors before the election, so it is not simply a case of only Democrats backing regulation.
The new governors' pledges and the measures taken before the election are summarized below.
| State | Person / party | Election result / status | Position or action on data center construction |
|---|---|---|---|
| Texas | Gina Hinojosa (D) | Won, defeating incumbent Greg Abbott | Pledged an immediate pause on new construction |
| Georgia | Keisha Lance Bottoms (D) | Won | Proposed a statewide pause |
| Ohio | Amy Acton (D) | Won | Supports a conditional pause |
| Iowa | Rob Sand (D) | Won | Opposes a pause |
| Florida | Byron Donalds (R) | Won | Promotes data center development |
| New York | Democratic governor | Pre-election action | Ordered a statewide pause on large data centers in July 2026 |
| Texas | Greg Abbott (R) | Pre-election action; lost this election | Restricted new connections in August 2026 and expanded the permit halt in September |
According to Georgia Recorder reporting, New York's Democratic governor issued an order in July 2026 imposing a statewide pause on construction of large data centers, the first in the nation. Texas's Republican governor, Greg Abbott, also restricted the connection of new facilities to the power grid in August and ordered environmental regulators to halt permits on September 21. In this election, Democratic candidates who ran on regulation won in several states, while in Iowa a Democratic candidate who opposed a pause also won. Support for data center regulation is not determined by party alone, and the election can be seen as reinforcing a trend toward tighter regulation that was already under way.
Voter reaction is also harsh. In NBC News exit polls, 76% opposed new data center construction in their area, and 56% said AI and data centers were a "decisive or major factor" in deciding their House vote. According to The Verge, local ballot measures to restrict data center construction have also drawn a string of supporting votes. Residents' concerns about electricity bills, water resources, and effects on the surrounding environment are driving state and local policy.
What the market is wary of is the impact such regulation could have on data center construction plans. According to Goldman Sachs analysis introduced by Reuters, about 45% of planned U.S. data center capacity is concentrated in areas where development restrictions could tighten, including states where Democrats won in this election. NBC News and Business Insider also reported Goldman Sachs' view that short-term construction halts by state and local governments could affect about 45% of planned capacity expansion.
However, the 45% is the share of planned capacity that could be affected by regulation, not a statement that construction will stop on 45% of the total. Which projects actually stop, and for how long, will depend on each state's policies and how operators respond. According to Reuters, Goldman Sachs had also said as of October that the short-term impact of political backlash would be limited. Its assessment of the risk has shifted following the election, but the scale of actual construction delays and investment cuts is not yet clear.
It is also unclear whether the new governors' pledges will be carried out as promised. Georgia's Keisha Lance Bottoms is aiming for a statewide pause but has not said whether it could be done by executive order or would require legislation from the state legislature. In other states, the impact will differ greatly depending on whether a halt is limited to new applications or extended to projects whose permits are already in progress. What form the regulation actually takes will have to wait for decisions after the new governors take office.
The Chip Stock Plunge Reflects Market Wariness Over Data Center Regulation
Financial markets reacted to the election results differently depending on the company and sector. In the U.S. market on November 4, according to NBC News, large tech stocks such as Amazon, Microsoft, and Alphabet rose, with some shares reported to have gained about 6%. The actual gains, however, varied by stock. NBC News cited investors' view that if new data center construction is restricted, large cloud operators that already hold ample computing resources would be relatively advantaged.
On the other hand, according to Business Insider, shares of neocloud operators such as CoreWeave and Nebius fell the same day. Neoclouds are emerging providers that mainly offer GPU computing resources for AI as cloud services. The pattern was that large companies with existing facilities rose while operators more dependent on new capital investment were sold. However, NBC News also reported that Apple rose about 6%, so such price differences cannot be explained solely by assessments of data center regulation.
On the Tokyo stock market the next day, November 5, the Nikkei Stock Average closed down 1.79% from the previous day at 61,077.90. Meanwhile, the TOPIX (Tokyo Stock Price Index) rose 1.16%, so the decline was concentrated in AI and semiconductor-related stocks rather than the overall market. During trading, Advantest, a major semiconductor test equipment maker, fell 11.69%, Tokyo Electron fell 9.89%, and SoftBank Group dropped more than 6% at points. CNBC reported that the selling was driven by concern that a Democratic-controlled Congress would tighten regulation on data center development and weigh on demand for AI infrastructure and semiconductors.
If new data center construction is delayed, not only the semiconductors installed there, such as GPUs and memory, but also the timing of orders for the equipment used to manufacture and inspect them could be pushed back. Japanese chip equipment makers are not direct targets of U.S. data center construction restrictions, but they are indirectly exposed through postponed capital spending. At this point, however, no specific reduction in semiconductor orders due to regulation has been confirmed, and the share price declines are thought to reflect concern about future demand.
The AI industry has also begun responding on the policy front. According to Axios, OpenAI plans to step up policy advocacy and public relations immediately after the election, announcing measures such as introducing "speed limits" that impose mandatory safety measures on the most advanced AI, signing agreements with communities where data centers are located, and calling its facilities "compute campuses." On November 5, the company also reaffirmed its policy of fully covering data centers' electricity costs.
According to OpenAI's official announcement, the commitment applies to data centers the company owns and operates going forward. It will pay for the electricity the facilities use and the necessary infrastructure, so that costs are not passed on to nearby residents. This responds to communities worried about rising electricity bills, but the same terms will not necessarily apply to facilities owned by partner companies. Meanwhile, according to the WSJ, Meta President Dina Powell McCormick said the election results will make it harder to build data centers, and some in the industry are reportedly considering moving construction outside the United States.
The milestones that will shape policy are concentrated in January 2027. On January 1, Colorado's revised AI law takes effect, and on January 3 the 120th Congress, with its Democratic majority, begins. Furthermore, once new governors who pledged data center regulation take office, the scope and implementation of any pauses will become concrete. Whether Congress asks AI company CEOs to testify and whether state construction halts extend to existing plans will be the next key factors to watch.
The midterms did not immediately decide regulation of the AI industry. But even without waiting for federal legislation, the likelihood has grown that AI companies' business activities will face constraints through congressional investigations and state and local construction rules. At the same time, if commitments like OpenAI's pledge to fully cover electricity costs are set out in concrete contracts, there is room to ease conflict with local communities. The focus going forward will be whether political pledges turn into actual regulation, and how AI companies revise their construction plans and the burdens they place on communities.