Why Republicans Lost the 2026 Midterms

On the night of the 2026 midterms, four tech investors and a comedian pulled 500,000 live viewers to a web stream, more than most cable news coverage managed that evening (02:07). It was a strange kind of victory lap, because the story the results told afterward had nothing to do with livestream numbers and everything to do with a country in a foul mood. Democrats swept both chambers of Congress and won at least 29 governor's races. Republicans lost states that had not gone blue in a generation: Texas, Kansas, Iowa, South Carolina (03:37). Something had broken. The question the hosts spent the episode chasing was what, exactly.

The first answer, offered by Republican strategist David Sacks, is simple: the war in Iran. Inflation had fallen to 2.4% in January, before the war began, down from 9% under President Biden (05:42). Then gas and diesel prices spiked, mortgage rates climbed back to 7.5%, and a party that had been forecast to hold the Senate by a wide margin lost roughly eight or nine seats (06:25, 34:06). Sacks points to a poll showing 74% of voters called the war either crucial or a major factor in their vote, and that bloc broke heavily toward Democrats (34:53). He also notes a defection among "podcast bros," influential right-leaning media figures like Joe Rogan and Tucker Carlson, who turned on the war and took younger male voters with them, a group whose votes swung 36 points compared with the 2024 election (09:23, 27:35).

Entrepreneur David Friedberg traces the failure further back, to the collapse of DOGE, the Department of Government Efficiency that Elon Musk left in June 2025. Friedberg argues that once Congress made clear it had no appetite for cutting the $7 trillion federal budget, the chance to address the cost of living structurally was gone. "China has just won," he tweeted that day, by his own account (14:14). He points to a Gallup survey in which 93% of Chinese citizens believe AI will make their lives better, compared with just 36% of Americans, fourth from the bottom worldwide (15:54, 16:19). In his reading, that gap primed American voters to see AI wealth as something flowing to a narrow class of winners while everyone else fell behind.

Venture investor Chamath Palihapitiya pushes past both explanations to something broader. He had argued before the election that cultural issues were fading and that a bloc of voters, neither reliably Democrat nor Republican, was simply rewarding whoever seemed to solve the affordability problem and punishing whoever did not. He offers three pieces of evidence. Money spent on advertising in contested races failed to predict winners. Gerrymandered districts, redrawn specifically to protect Republican seats, flipped anyway, with Democrats winning 11 of 15 redrawn seats (21:50). And the swing voters who elected Biden, then Trump, then this Congress were the same people each time, simply discharging their anger at whoever held power.

"The structural capacity of the American economy to support and grow wage earners is broken. It's clear." — Chamath Palihapitiya [20:26]

Chamath's diagnosis is that asset owners, people who hold stocks, real estate, and businesses, have profited in every interest rate environment since COVID, while wage earners have not. His prescription, echoed in part by Jason Calacanis, is to eliminate federal income taxes for anyone earning under $150,000 a year, paired with closing loopholes elsewhere, so the wage-earning class gains some stake in the system (24:39). Calacanis adds a package of his own: pardon the entire Trump administration now to defuse future investigations, tax margin loans over $10 million at half the capital gains rate, raise the minimum wage to $15 an hour over four years, and ban members of Congress from trading individual stocks (46:10 to 48:21). Sacks's counterargument is narrower and, in his telling, more urgent: end the war, and Brent crude could fall from roughly $110 to $70 a barrel almost immediately, pulling gas, diesel, and grocery prices down with it (36:15).

The Data Centers Nobody Wants Next Door

That same affordability anger is now aimed at AI infrastructure. Stocks across Nvidia, SpaceX, Tesla, AMD, and Micron lost close to $4 trillion in combined market value in a single week, the worst day in stock market history by total dollars (53:22). Prediction market odds on Anthropic's IPO, tracked by the platform PolyMarket, fell from 95% before the midterms to 71% after (54:06). Part of the trigger is political. Texas elected Gina Hinojosa, a Democrat, as governor, unseating three-term incumbent Greg Abbott in what the hosts call the biggest surprise of election night (39:00). Hinojosa has not blocked existing data center projects, but she has signaled no new permits until new regulations are in place, framing data centers as the reason local electricity bills and water use are rising while the companies get tax breaks (39:55, 75:47).

Chamath describes a video of two SpaceX employees presenting a six-gigawatt data center proposal to residents of Cochise County, Arizona, explaining promised cuts to electricity costs and property taxes while the audience shouts them down, evidence, he says, that the backlash is emotional rather than fact-based (55:28). Sacks is less sanguine about how this resolves. He separates the risk into two buckets: federal AI safety regulation, which he worries could create a surveillance apparatus, and state and local permitting, where he expects a near-total freeze on new projects through 2027 because no politician wants to be blamed for approving the next unpopular data center (74:46, 77:03). Calacanis's prediction is that capital simply moves. Data centers, he argues, can be built in Canada, Mexico, Singapore, or the Middle East once the war ends, turning the United States into a net importer of AI processing power, or "tokens," rather than an exporter (61:24, 62:41).

Buried in that fight is a contract detail that Chamath thinks could reprice the entire industry. SpaceX's compute deal with AI company Anthropic includes a mutual 90-day termination clause (59:29).

"If SpaceX says, 'I'm gonna terminate this compute in 90 days,' that is the king-making move." — Chamath Palihapitiya [60:06]

If SpaceX exercises that option, it could resell the freed-up computing capacity at the open market rate, which Chamath estimates runs two to three times higher than what Anthropic currently pays. Anthropic, heavily dependent on that slice of computing power, would likely have no choice but to accept the higher price. Friedberg is less certain SpaceX founder Elon Musk pulls that lever, given his working relationship with Anthropic, but both agree the clause gives Musk leverage few other players in the industry hold.

A Proof Nobody Expected

The episode's final turn is less about politics and more about what the technology can actually do. Friedberg walks through a question that has sat at the center of computer science since 1999: can a particular category of hard problems, called NP problems, always be solved as quickly as their answers can be checked? A puzzle like Sudoku is an NP problem, slow to solve but fast to verify. In 1999, researchers proposed the Exponential Time Hypothesis, which held that one class of these problems, called 3SAT, can never be solved in anything but exponential time, meaning the time required explodes as the problem grows (86:13). That hypothesis effectively closed off a central open question in math and computing for 27 years.

In a 175-page paper, OpenAI claims to have disproved it, showing that 3SAT problems can in fact be solved in sub-exponential time (88:01). The implication, still being checked by outside mathematicians, is that the deeper question, whether every efficiently checkable problem is also efficiently solvable, remains open rather than settled. Friedberg notes the paper was produced end to end in under ten days, with some reports suggesting a matter of hours (88:08). Whatever the political mood around AI, this is the kind of result that keeps the industry's sharpest believers convinced the technology is still underpriced, even as voters in Texas and elsewhere decide, for now, that they want less of it in their backyard.