Scott Wiener, the California state senator who authored AI safety law SB 53, won his race to succeed Nancy Pelosi in the U.S. House, campaigning explicitly on AI regulation5. He arrives in a Congress where Democrats now control both chambers after a midterm sweep that included the first Black woman governor in Georgia, the first Native American woman governor in New Mexico, and a Democratic flip of a Maine Senate seat held by a Republican for 30 years1. The combination of a regulation-forward lawmaker entering Congress and a Democratic majority that ran, in part, on holding AI companies accountable creates the most concrete legislative threat the AI industry has faced at the federal level. The question is no longer whether Washington will try to regulate AI but which bills move first, how fast, and whether the industry's political spending apparatus can do anything about it.
The industry's main vehicle for shaping that answer failed its first real test. Leading the Future (LTF), the biggest pro-AI super PAC network, entered the cycle with a war chest of around $100 million, backed by Andreessen Horowitz, Palantir co-founder Joe Lonsdale, and OpenAI's Greg Brockman3. It ended the midterms with just over $20 million still in the bank and a roster of dozens of successful congressional candidates from both parties6. That sounds respectable until you examine where the money went. Donations went overwhelmingly to safe candidates, and the results in the small handful of riskier races that LTF got involved with indicate that its plans to shape the election fell flat, as Wired's Tess Owen reported. Two AI-focused PACs, American Mission and Building the Future, collectively spent $4.8 million to back South Carolina Senate candidate Darline Graham, who lost2. American Mission also spent more than $500,000 apiece supporting Republican House candidates Aaron Flint in Montana and Laurie Buckhout in North Carolina, both of whom lost. LTF's Democratic arm, Think Big, had better luck backing Florida Rep. Jared Moskowitz, who won reelection. The structure had been designed for "maximum flexibility" and drew comparisons to Fairshake, the pro-crypto super PAC network that dominated in the 2024 cycle. The comparison now cuts the other way: Fairshake's 2024 wins came in competitive races, while LTF's 2026 wins came mostly in races that were never in doubt, leaving the network without demonstrated ability to move a contested outcome.
On the other side of the ledger, groups advocating for stronger AI safeguards saw their own mixed results but landed a marquee win. Public First, backed by Anthropic, contributed $3 million to a PAC supporting Wiener, who was leading his race at the time of publication. Wiener's victory gives the safety-regulation camp a sitting House member with an actual legislative track record on AI, something the industry's opponents lacked before this cycle.
The market read the results as a constraint on AI infrastructure buildout. Democrats running on AI regulation and anti-data center platforms won key races, including Sherrod Brown in Ohio, as leadership signaled aggressive AI oversight9. In Texas, Democratic governor-elect Gina Hinojosa proposed a statewide moratorium on new data center permits, though projects already under construction would reportedly be exempt11. The Philadelphia Semiconductor Index plunged 11 percent, the Nasdaq Composite tumbled nearly 4 percent intraday on Thursday, and neocloud players CoreWeave, Nebius, and Iren fell between 9 and 13 percent12,13. Barclays analysts said the blue-wave elections for governor created more of a logistical challenge and sentiment headwind, adding that Oracle and CoreWeave were more exposed because of the capital raises they had undergone7. "They need these projects to turn over quickly and begin producing revenue or else questions are going to rise around the servicing of that debt they've taken out," a CNBC correspondent reported.
ANALYSIS That debt exposure is not hypothetical. Vector Wire's deal desk tracked a $40 billion SpaceX debt raise in late October and a $60 billion Broadcom debt issuance. Nebius raised $4.5 billion in August with Nvidia as an investor. Our prior coverage flagged credit markets already repricing AI infrastructure risk before the election, with Meta facing a downgrade and OpenAI absorbing a reported $10 billion funding cut. The midterm results layer political risk on top of credit risk that was already tightening.
Yet the sell-off contained a notable split. Shares of the largest tech giants actually jumped on Wednesday, the first full trading day after the results, as investors bet that a Democrat-controlled Congress would slow AI capital expenditure that had been draining cash flow8. Ben Thompson argued that the midterms outcome was a win for hyperscalers and AI labs, a mixed bag for Nvidia, and an advantage to China10. Japanese investment bank Nomura said the bigger impact may hit from 2028, with much of the capacity due to be available from 2027 already under construction and later projects more exposed to delays. JPMorgan's James Sullivan told CNBC that existing capacity may become more valuable if new supply is constrained, while debt-funded neoclouds face much higher borrowing costs than hyperscalers.
This is the K-shaped trade that CNBC named explicitly: companies that already have capacity in the ground, chiefly the hyperscalers, benefit from scarcity, while companies still raising debt to build, chiefly the neoclouds, face a double squeeze of tighter permitting and higher financing costs. The political calendar reinforces the asymmetry. Gubernatorial moratoriums and permitting reviews can take effect within months; federal legislation takes longer but sets a ceiling on what states can allow.
The counter-case is that Congress alone cannot kill a project. As one analyst noted on Bloomberg TV, Trump remains in control of the DOJ, FTC, SEC, and FCC. Executive-branch agencies still set enforcement priorities, and a presidential veto can block legislation. The South China Morning Post reported analysts warning that slowing U.S. AI development could help China's ambitions in the technology rivalry4. That geopolitical framing could give moderate Democrats pause on the most aggressive proposals, particularly any measure that would cap compute buildout without a corresponding restriction on Chinese capacity. But the election results showed that voters in swing states responded to anti-data-center messaging, which gives Democratic leadership a domestic mandate that may outweigh the China-competition argument in the near term.
Most major AI company CEOs have remained publicly silent since the results. LTF entered the cycle with around $100 million and ended with just over $20 million remaining. That spending, concentrated on safe candidates rather than contested races, left the network without a demonstrated ability to shape outcomes where it mattered. Anthropic, which backed the safety-regulation side through Public First, is the notable exception among major labs. It just completed an IPO at a $100 billion valuation. OpenAI closed an $11 billion round days before the election. Both companies now face a Congress where the lawmaker most identified with AI safety legislation holds a seat and the majority party ran on accountability.
What to watch: whether Wiener introduces a federal analog to SB 53 in the first 100 days of the new Congress, which would signal the timeline for model-licensing requirements. Whether Hinojosa's proposed Texas data center moratorium survives legal challenge and, if so, whether it triggers similar actions in other newly Democratic states. CoreWeave and Nebius report earnings next Wednesday; guidance on future capital raises will show whether the political risk is already priced into their financing plans. And whether LTF deploys its remaining $20 million-plus toward lobbying rather than elections, a pivot that would confirm the PAC's own assessment that its electoral strategy failed.