Elon Musk says he gave up on cutting the U.S. national debt from inside Washington. In a clip played on Moonshots with Peter Diamandis, taken from his interview with Walter Isaacson released October 20, 2026, Musk said he had concluded there was "no way we were going to solve the deficit from within the government." His alternative: "the only way we're going to solve the national debt is with AI and robots."
The clip opened a long exchange on the show, recorded October 20 and published October 21, 2026. Guest Chamath Palihapitiya, the venture capitalist who founded Social Capital and co-hosts the All-In Podcast, joined the regular panel to work out how machines could plausibly shrink a debt that politicians will not touch. The panelists mostly agreed with Musk's direction and disagreed on the mechanism. Every route they described is a forecast.
Why Musk says cuts won't work
Musk's argument in the clip is about politics more than economics. He had worked on the Department of Government Efficiency (DOGE), the Trump administration's cost-cutting push, and said that if you try to stop wasteful spending, "you make a lot of enemies." He said President Trump "likes to be liked," and that the appetite for cuts was missing in the presidency, the House, the Senate and the judiciary alike. "Basically, no one in authority in the government whatsoever really wants to stop, reduce spending," he said.
In the full interview transcript, Musk estimates that digital superintelligence alone could add roughly 20 to 30% more economic output. He describes much larger gains from robotics and expansion into space as possibilities.
Four routes from robots to a smaller debt
Peter Diamandis, the XPRIZE founder who hosts the show, started with the size of the problem. By his count, the national debt was about $40 trillion as of October 5. It grew $2.44 trillion over the past year, or about $6.5 billion a day, and interest alone costs $3 billion a day.
Diamandis named four ways AI and robots could help. The first is growth. A country's debt burden is usually measured against its gross domestic product (GDP), the total value of what the economy produces, so a fast-growing economy makes the same debt smaller in relative terms. He added, in passing, that GDP is "a lousy measure." The second is tax revenue: a bigger economy brings in more tax without higher rates. The third is cheaper government. Health care is the biggest driver of federal spending, he said, so AI doctors, AI-discovered drugs, robotic care and longevity treatments could keep people healthy for longer and cut costs. The fourth is deflation: if robots make goods and services much cheaper, everything the government buys gets cheaper too.
He also cited Anthropic co-founder Jack Clark, who spoke at the IMF meetings in Bangkok that week. According to Diamandis, Clark said Anthropic's internal economic models project U.S. GDP could reach $44 trillion by 2030, up from $30 trillion today. Diamandis called that nearly 50% growth, about 9 to 10% a year, against the 2 to 3% the U.S. has averaged for decades.
Palihapitiya: the debt stops mattering
Palihapitiya accepted the direction of the argument but changed its terms. AI "doesn't get us out of debt as much as it just makes the debt not that important," he said. His main lever was health care. With the next generation of weight-loss drugs plus some cancer vaccines, he asked whether health care could fall below 10% of GDP. "You probably could," he said. Later he estimated that those treatments together would probably cut the cost of health care in half.
The drug he had in mind is retatrutide, Eli Lilly's investigational once-weekly medicine, which acts on three hormone receptors: GIP, GLP-1 and glucagon. In May 2026, Lilly reported results from the TRIUMPH-1 obesity trial, which randomly assigned 2,339 adults without diabetes to 4 mg, 9 mg or 12 mg doses or a placebo for 80 weeks. Assuming people stayed on treatment, average weight loss was 19.0%, 25.9% and 28.3% across the three doses, compared with 2.2% on placebo. Counting everyone regardless of whether they kept taking the drug, the figures were 17.6%, 23.7% and 25.0%, against 3.9%. Stomach and gut side effects were common, and 11.3% of people on the highest dose stopped because of side effects, compared with 4.9% on placebo. The trial measured weight loss; the savings to the health system are Palihapitiya's own estimate.
He also explained why he does not expect reform from inside the system. The hunt for fraud, he said, missed the bigger problem of everyday waste. People were looking for "somebody's hand in the cookie jar," when the system was "just like a little bit more leaky than that." His example was upcoding, where a hospital bills for more than the visit involved: "You come in for your knee, you know, it tags you for the sore throat," and if nobody pushes back, the claim gets paid.
Nobody inside has a reason to stop it, he said. Insurers that cut waste would also book less revenue. So he expects any fix to come from "discontinuous improvements so that the systems aren't used": treatments that keep people from needing care, so the waste never has a chance to happen. He added robots on the productivity side and said he could see the debt growing more slowly while GDP ends up much higher.
Wissner-Gross: a "sleight of hand"
Alexander Wissner-Gross, a computer scientist and founder of Reified, looked at where the debt is growing: entitlement programs such as Medicare, and defense. He described "sci-fi scenarios" in which spending that would normally flow through those channels shifts over to robotics and AI.
His analogy was the encyclopedia. In the 1980s, having knowledge at your fingertips meant buying the Encyclopedia Britannica. Now Wikipedia is free for everyone. He thinks defense and entitlement spending could follow a similar path, though it might take "a forcing function or a crisis" to get there. He pointed to ephemeralization, Buckminster Fuller's term for technology doing more with less, and pictured new spending in those areas falling to near zero through automation from one year to the next.
In hindsight, he said, it would look like "a rebasing of the currency, not a debasing." He expects it to take a few more years of technical progress. He labeled the idea a prediction, "not financial advice," and said it would look like a sleight of hand that erases "a good chunk of the national debt increase."
Why Musk dropped DOGE, as Dave Blundin tells it
Diamandis noted that Musk's own growth forecasts have shifted. When the show interviewed Musk last December, he recalled, Musk predicted double-digit economic growth within 12 to 18 months and triple-digit growth within five years. At the Abundance Summit in March, Musk talked about a tenfold larger economy in 10 years, which Diamandis described as pulling back a little while staying very bullish.
Dave Blundin, founder of Link Ventures, offered his own reading of Musk's change of course. When Musk first got involved in politics, Blundin said, the world was in the GPT-4 era, and the consensus around MIT and Stanford put artificial superintelligence (ASI) maybe 10 to 15 years away. Google DeepMind chief Demis Hassabis, he said, was arguing that two or three big new ideas were still needed. Fixing the government before the country went bankrupt looked like the necessary chore.
By the time Musk was working on DOGE, Blundin said, progress was coming much faster than anyone expected. If AGI was arriving eight or 10 years sooner than predicted two years earlier, AI itself became "the escape valve." Blundin asked whether Musk would get involved in DOGE again if he could rewind his life, and answered himself: "No way. Almost certainly not."
As evidence, he pointed to Musk's other moves. Musk "massively doubled down" on his AI company xAI, Blundin said, and "bought every freaking Vera Rubin chip," Nvidia's latest AI processor, that could come off the assembly line, while praising Nvidia chief Jensen Huang to get a bigger allocation. "It seems like he changed his mind," Blundin said. "No, the world changed very, very quickly."
Ismail: shrink the debt by growing the economy
Salim Ismail, founder of Open ExO, said Musk is making "a denominator argument." Debt is judged as a share of the economy, and AI and robots attack the bottom of that fraction by driving huge productivity gains. Combined with Palihapitiya's and Wissner-Gross's point about technology making existing goods and services far cheaper, Ismail said, that "will go a long way to solving the debt issue." He called debt the biggest problem facing the world, not only the U.S., and pointed to France as an example.
An aside on unrest
Diamandis raised a darker possibility. He recalled Musk predicting civil unrest within the next few years, and worried that the government would respond with COVID-style checks of $2,000 or $3,000 a month per American. Palihapitiya answered with banter. He said there was "a case to be made for giving everybody" retatrutide, joked about "forced hugging," and listed the drops he imagined in social media, alcohol, cannabis and pornography use, adding, "GDP would drop." Wissner-Gross called it "a brave new world" with the drug as soma, the pacifying pill from Aldous Huxley's novel.
What about countries without the AI lever?
Blundin pointed out that the whole discussion was about the U.S. A European listener, he said, faces worse debt without "the AI lever." If Anthropic's wet labs start finding new medicines quickly, which he called likely, the U.S. will export them, and "that money is going to come pouring into the U.S." He said he would be "worried sick" in continental Europe without its own AI models, though he acknowledged he was not part of those conversations.
Palihapitiya went further. He said every country without a resource worth trading to the two powers with ASI, the U.S. and China, "essentially becomes a vassal state." He pictured a handful of remora fish around each of the two big sharks, with everyone else "in a very difficult spot." His example was Italy, to which he said he has "tremendous emotional allegiance" because his wife is Italian. Italy has valuable critical resources that Liechtenstein and Belgium do not, he said, so it may have to ask why it is aligned with them, an alliance he called "a historical artifact" of decisions that made sense at the time.