Two Hundred Sixty Five Million Dollars Into the Midterms: The AI Industry Meets the Data Center Backlash, September 6, 2026

0:00Welcome to the DX Today podcast, your weekly deep dive into the AI ecosystem. I'm Rick Spare and joining me as always is Laura. Today we are following the money. I have been waiting all week for this one, Rick, because the story is not a model launch or a benchmark score. It is an election and the numbers are frankly staggering. Then set the table for us when we say the artificial intelligence industry is entered American electoral politics. What scale of spending are we actually talking about heading into these midterms and how does it compare to what we normally see? Roughly $265 million pledge to super political action committees and allied political groups this cycle. That is campaign money, not lobbying money. Andrieson Horowitz alone accounts for more than $115 million in disclose federal contributions, which reportedly makes the firm the single largest owner of these entire midterms. Four years ago, this industry barely had a political operation at all.

1:00Let us name them because listeners deserve to know who is actually behind the advertisements they will see on their televisions between now and November. And where that funding originates. The flagship is a super political action committee called leading the future. It has raised roughly $140 million, launched with more than 50 million committed on day one and the funding traces to a remarkably small circle of people. That small circle is one of the things that makes this different from ordinary industry political spending, where you usually see hundreds of mid-sized corporate donors rather than a handful of individuals. Andrieson Horowitz put in about $50 million. Greg and Anna Brockman gave $25 million combined. And Greg Brockman is of course the president of OpenAI, which puts a frontier laboratory executive directly inside the operation. There is a second vehicle as well. And I think the distinction between the two is actually the most interesting wrinkle in this entire story.

2:02So walk us through that one carefully. That is public first action sitting on a war chest of roughly $100 million. Anthropic has committed at least $40 million to that effort and their stated posture is notably different from the flagship. Different how though because from the outside $100 million political vehicle funded by frontier laboratory looks an awful lot like the other $100 million vehicle funded by frontier laboratory investors. Reporting indicates anthropic restricted its contribution to policy education rather than candidate attack advertising. And the company has publicly argued that governments should retain the power to block genuinely dangerous systems, which is not a universal industry position. Then let us get to the actual policy ask because money and politics only becomes interesting once you know precisely what it is trying to buy in terms of legislation. One thing above all others, a single national framework for artificial intelligence that preempts state law.

3:04So companies face one rule book instead of 50 different and occasionally contradictory ones across the country. And the reason that ask has become urgent is that state legislatures have not been sitting quietly. The volume of state level activity over the last two years has been genuinely remarkable by any measure. More than 1,000 artificial intelligence bills were introduced across the states in 2025 alone. From the industry seat that is not a patchwork, it is a minefield and every session adds more. I want to steal man that position because I do not think it is obviously unreasonable. A genuinely fragmented compliance landscape does impose real and asymmetric costs, particularly on the smallest companies. That is the strongest version of their case and it deserves a hearing. A well capitalized laboratory can staff 50 compliance teams without blinking. A 12 person startup in Ohio simply cannot and quietly gets squeezed out of the market. The second argument I hear constantly is the competitive one.

4:05The claim that regulatory fragmentation slows American development and effectively hands a strategic advantage to China. How much weight does that actually carry? Some, but it is doing an enormous amount of rhetorical work. That argument gets deployed against essentially every proposed guardrail regardless of content, which over time makes it very hard to distinguish from a reflex. And Congress has already weighed in on preemption once in a vote a lot of people missed at the time, but which really explains why all of this money suddenly exists. The Senate rejected federal preemption by 99 to 1, not a close call, not a partisan split, essentially unanimous. When the legislative door closes that hard, you go looking for an electoral one instead. That is the pivot that turns a lobbying story into an election story. If you cannot win the vote with the current Congress, you spend money to change who sits in the next Congress. Precisely. And here's where the plan collides with something the spreadsheets did not price in,

5:07which is that American voters have developed extraordinarily strong feelings about data centers in their own communities. This is the part I find genuinely fascinating because the backlash is not really about chatbots or job displacement at all. It is about land, water, electricity bills, and neighborhoods. And the polling is brutal. Only about 20% of Americans support new data center construction, while roughly 80% oppose it. That is not a partisan number. That is a rare national consensus number. Break the partisan internals down for me anyway, because I suspect they tell a more useful story than the top line about where exactly this fight is going to be waged. 14% of Democrats say data centers are good for America. 31% of Republicans say the same thing. So the ceiling is low everywhere and the floor is genuinely under water in both parties. What is driving that? Five years ago, most people would have told you a data center was a boring warehouse on a highway that nobody had any particular opinion about whatsoever.

6:13Water and power, mostly. Two thirds of the data centers built since 2022, sit in areas of high water stress, and the construction pace was up around 60% year over year in July. And that that pressure is now producing actual policy at the state level, which is exactly the kind of thing a preemption campaign is designed to stop from spreading to other states. New York issued the first statewide moratorium on new data center construction in July of 2026. Lasting one year, that is a state government hitting the brakes on this industry's physical footprint. So the industry responds the way industry is always respond, which is with an advertising campaign. Tell us about the group formed specifically to sell data centers to a skeptical public. It is called build American AI affiliated with leading the future. And it is running advertisements in Kansas, Ohio, and Wisconsin to persuade voters of the virtues of having data centers in their own backyards. Across the whole cycle, how much money has actually gone into data center themed political advertising?

7:20I want a sense of whether this is a serious sustained campaign or merely a gesture. About $31 million on data center related political advertisements this year. And separately, MetaOr announced a $1 billion community fund that it has branded futures for everyone. Critics will read that $1 billion fund less as generosity and more as the going price of local acquiescence. Though I think the harder charge to defend is astroturfing, manufacturing what looks like grassroots support and hoping that nobody bothers to check the funding behind it. There was open AI stated that groups advocating on artificial intelligence should be clear about their policy views, be honest about whom they represent, and not use tactics like astroturfing. That is a useful standard to hold everyone to, including the people who articulated it. Let us test it against what has actually happened in a real congressional race this year. The best case study is New York's 12th Congressional District,

8:20where roughly $27 million of artificial intelligence related political money landed on a single house primary. That is an extraordinary concentration of resources. $27 million in one house race exceeds what many statewide campaigns spend. Who were the candidates and what made that particular seat worth that kind of investment? Alex Borra's who authored New York's raise act on artificial intelligence safety lost to Michael Lashir. More than $13 million supported Bores and more than $8 million opposed him in that contest. So the author of a state safety law becomes the test case and the message sent to every other state legislator watching is completely unmistakable regardless of who ultimately won the seat. That is the real product being purchased here, not one seat, deterrence. Every state representative now understands that authoring an ambitious artificial intelligence bill can attract eight figures of opposition spending almost overnight.

9:21There is also a transparency dimension that I do not think has received enough attention. Involving a formal complaint filed with federal election regulators earlier this year against parts of this network. The Campaign Legal Center filed a complaint on May 5th against two super political action committees. American Mission and Think Big, both part of the broader leading the future network of vehicles. And the allegation is not about the size of the spending, which is entirely legal under current law, but about whether the public can actually see where that spending went. Is that right? Correct. The complain alleges the committees routed nearly all their disbursements through newly formed shell companies that are not genuine commercial vendors, which would obscure the actual spending from federal reporting requirements. And the network in question had raised how much at the time that complaint was filed. I want listeners holding the scale and the opacity questions together in their heads simultaneously. More than $125 million.

10:23So the disclosure question is not a technicality about some small committee. It concerns a nine-figure operation and that complaint remains pending before regulators. Let us close by looking forward. Where should people actually be watching between now and election day if they want to know whether this enormous bet is working or quietly failing? Three races carry the most signal. Texas governor, where Greg Abbott faces Gina Hinegosa, Pennsylvania governor, Josh Shapiro, against Stacy Garrity, and the Ohio Senate contest between Sherrod Brown and John Houston. Texas is especially interesting because the challenger there has made data centers an explicit part of her economic argument rather than treating the issue as a narrow environmental complaint. She has Gina Hinegosa said data centers are the physical manifestation in Texas of government that is working for the moneyed interests instead. That single line reframes infrastructure as a fundamental fairness question. And that framing travels well. It works in a suburb

11:26worried about electricity bills just as effectively as it works in a rural county. Worried about groundwater, which is a dangerous combination for this industry. Nicole Turner, Lee of Brookings, put it well when she said that trying to bypass communities and developing these data centers is not working and that they are going to take it to the polls. So what is your honest read? Does 265 million dollars buy federal preemption or does it buy a very expensive lesson about the limits of advertising against local sentiment? My read is that money moves primaries far more reliably than it moves general elections and the data center backlash is fundamentally a general election issue. The bet is real, but I think it is overconfident. I would add one thing. The five largest technology companies are projected to spend around $560 billion on infrastructure this year. So $265 million is essentially a rounding error to them, which is exactly why this does not stop after November.

12:27Whatever happens in these particular races, the political apparatus built during the cycle is permanent and it will be back considerably larger next time. That is the sentence I want people to sit with, not the outcome of any single race, but the arrival of a permanent electoral machine attached to the most consequential technology of our era. And the counterweight is genuinely unusual because for once the opposition is not an advocacy group in Washington. It is homeowners, county commissioners and utility ratepayers who notice their monthly bills climbing, which may turn out to be the most durable form of accountability this industry has faced yet precisely because it did not originate in Washington and therefore cannot simply be lobbied away. That is the story. $265 million of the smartest money in America running directly into 80% public opposition and we find out in November which one of them bends. That's all for today's episode of the DX Today podcast. Thanks for listening and we'll see you next time.