AI Super PACs in the 2026 Midterms: Heavy Spending, Mixed Results
American Mission and Building the Future together spent $4.8 million backing Darlene Graham in the South Carolina Senate race, and Graham lost her bid in the 2026 midterms, according to TechCrunch. American Mission is the Republican spending arm of Leading the Future, an AI-focused super PAC network (a political committee that can raise and spend unlimited sums on campaign advertising, as long as it works independently of candidates) backed by OpenAI president Greg Brockman and Andreessen Horowitz.
The same committee put more than $500,000 apiece behind two Republican House candidates, Aaron Flint in Montana and Laurie Buckhout in North Carolina. Both lost. On the Democratic side, Think Big, Leading the Future's Democratic arm, backed Florida Rep. Jared Moskowitz, who won reelection.
The opposing committees were active as well. Guardrails Alliance spent more than $1 million apiece opposing California Rep. Jimmy Gomez and Montana candidate Aaron Flint, and both lost. The same group spent $300,000 supporting Angela Gonzales-Torres, who defeated Gomez. Public First, an Anthropic-backed group, contributed $3 million to a PAC supporting California state Sen. Scott Wiener.
In the primaries, Leading the Future spent $27.38 million, including $8 million opposing New York Assembly member Alex Bores. Bores lost his bid for Congress in June 2026. The network's general-election activity was considerably smaller. Fortune reported that as of October 26, 2026, Leading the Future had spent only $7.4 million on general-election races, and that it held $34 million in cash on hand during the general election period. Wired reported that the network finished the midterms with just over $20 million still remaining.
Fundraising totals run higher than spending figures. Reuters reported in August 2026 that Leading the Future had raised $140 million for the midterms. A separate Yahoo News report puts total AI super PAC spending in the 2026 cycle at $79 million, most of it before November, across 68 House and Senate seats.
The election outcome itself was shaped by larger forces. Time reported that Democrats won control of Congress, and that voter opposition to AI data centers helped shape key races.
In my view, the spending record does not support a simple reading of influence in either direction. The committees that spent heavily lost as often as they won in the races they targeted most visibly, and the largest pool of money was held back from the general election. The data center question is the more revealing signal. Voters who object to local power demand, land use, or water consumption are not persuaded by national PAC advertising, and a campaign that treats AI infrastructure as a side issue may find that local opposition arrives without any help from the industry's critics.
The broader context here is that AI policy has become a line item in campaign finance, with committees on both sides buying visibility in primaries and hoping to shape how members of Congress vote on chips, model rules, and infrastructure permitting. How much of that effort translated into legislative outcomes will take sessions, not ballots, to show. For engineers and product leaders, the practical lesson from this cycle is that the political environment around compute buildouts is now shaped as much by county commissions and utility rate cases as by federal money.
