JD Vance on AI Regulation and H-1B Fraud

In 1954, an official at the Atomic Energy Commission promised that nuclear power would soon make electricity "too cheap to meter." The line became a punchline, a symbol of a technological promise that never arrived. Seventy years later, Vice President JD Vance borrowed the phrase for a different fuel: the power needed to run data centers and artificial intelligence. "We have to have this vision in the United States of electricity that is too cheap to meter," he said (11:50). It is a strange thing to hear a sitting vice president reach for a broken promise as an aspiration, and it says something about the mood of this conversation: a government eager to claim it is building at scale, even while the numbers it cites show how far behind it still is.

Vance's appearance covers a lot of ground: immigration, manufacturing, Iran, inflation, the deficit, AI regulation, H-1B visas, entitlement fraud, and Israel. But one idea runs through all of it. The last two years, in his telling, have been about finding where money, workers, and power are leaking out of a system, and plugging the leaks.

Start with immigration, the achievement Vance leads with. He says illegal border crossings are three million lower than when the administration took office (02:00), and he argues that no postwar American president has actually reversed the flow of illegal immigration rather than merely slowing it (02:13). It is a big number, delivered without much room for cross-examination in this format, but it sets the tone: the administration wants credit for closing a gap, not opening an opportunity.

The manufacturing story runs the same way. Vance says the country has reversed "a 40-year trend" of industrial decline over the past 18 months (03:15), tying it to what he calls a bipartisan consensus that let the US drift toward a services-and-finance economy. He hedges immediately: "Rome wasn't built in a day" (03:22). That qualifier matters. Much of this interview works the same way, a bold claim followed by an admission that the underlying problem is still large.

Nowhere is that pattern clearer than on the deficit. Vance says the federal government inherited a $2 trillion deficit (08:26) against a $7 trillion budget (07:43), and that anti-fraud work alone has saved about $250 billion (07:22). Set against that budget, $250 billion is real money and also a small fraction of the whole. Vance's own math makes the point: the goal is not to balance the budget through cuts, but to grow the economy until the deficit falls to around 3 percent of GDP, or gross domestic product (08:45), a target that depends as much on future growth as on today's spending discipline.

Energy is where his argument sharpens into something closer to a story about national decline. Vance says China now generates three times as much electricity as the United States (11:39), a gap he traces to decades of failure to build power generation, with the environmental movement carrying much of the blame in his account (12:06). That same failure, in his telling, produces the backlash against AI data centers: not enough power built for a generation, so any big new user of electricity becomes a target for public anger over rising bills (10:16).

The Frankenstein Problem

That argument about generation slides straight into a fight over who is responsible for AI, and here Vance is at his most pointed. Some AI companies, most visibly Anthropic, have argued that their own models are becoming dangerous enough to require a new layer of international governance. Vance rejects the premise. "So if you're gonna create Frankenstein, don't come to the government and say, 'We need regulation,'" he says (13:26). His argument has two parts: if a company genuinely believes its technology is dangerous, it should stop building it, and if it will not stop, it should focus on giving customers the defensive tools to fight the very risks it created, rather than asking Washington to write the rules. He is careful to say he does not think Anthropic's chief executive, Dario Amodei, is being cynical: "Everybody that I know tells me that he's very earnest... it's not about regulatory capture" (12:59). But sincerity, in Vance's framing, does not settle who bears the cost of the risk.

The same instinct shows up in his account of H-1B visas, the program that lets US employers sponsor foreign workers with specialized skills. Vance says Congress will not reform the program legislatively, so the administration acted on its own: a new $100,000 fee (17:03) meant to filter out companies using the visa to hire cheaper foreign labor instead of genuine high-value talent. His test case is blunt: a company that says it cannot find workers while it has just laid off five thousand employees should not be allowed to import replacements (17:48). "The H-1B should not exist to replace American workers with low wage foreigners. It should exist to enrich the American economy," he says (18:00).

Fraud is where the enforcement theme becomes almost obsessive. Vance, who has taken on an informal role as the administration's fraud watchdog, cites an estimate from Elon Musk, who led the Department of Government Efficiency, or DOGE, that as much as 20 percent of federal spending, roughly $1.4 trillion, is lost to fraud and waste (19:10). He points to Medicare claims tied to people he says are not verifiable, real individuals (19:39), and to the Supplemental Nutrition Assistance Program, or SNAP, where the federal government wires California roughly $9 billion a month with what he describes as almost no eligibility check (21:03).

"It's not an exaggeration to say that we have no ability to know who is even on the SNAP program, whether they're a real human being, whether they're a legal resident of the United States of America." (JD Vance, [21:03])

It is a striking claim, delivered with the confidence of someone who has read the paperwork, though it also rests on the cooperation of state governments that Vance says have refused to help verify eligibility. Whether the number survives closer scrutiny is a separate question from whether it lands as politics.

The same instinct to draw a hard line runs through his foreign policy answers. On Iran, Vance defends military strikes as consistent with, rather than a betrayal of, a promise to avoid foreign entanglements, arguing that pulling the US presence from the Middle East would risk a "worldwide energy crisis" given Iranian attacks on commercial shipping (05:24). On Israel, he argues the administration has been more willing than any in 40 years to publicly diverge from Prime Minister Bibi Netanyahu's government when American and Israeli interests do not line up (22:32), framing this as evidence of independence rather than distance.

What ties all of this together, and what makes the interview worth hearing beyond its news value, is a story Vance tells about the presidency itself. Asked what it is like working for Donald Trump, he describes a man who barely sleeps and calls aides at all hours.

"He'll call us at 2:30 in the morning and he'll say, 'Were you up?' And it's like, 'No, sir, I wasn't, but I am now.'" (JD Vance, [15:00])

It is a small, human anecdote, and it does real work in the interview. It reframes everything else Vance has said, the border numbers, the fraud estimates, the visa fee, as the product of a management style built on constant pressure rather than steady process. Vance says he prefers this to what he calls "caretaker presidents" (15:33), leaders who manage decline carefully rather than risk change. Whether that preference is wisdom or a rationalization for moving fast and arguing about the numbers later depends on which of his claims survive contact with the next audit, the next court challenge, the next election.

That last question, the midterms, is where Vance ends. His closing pitch is not really about any single number. It is a binary: stay the course on a program of enforcement, or hand power to what he calls political "radicalism" (26:39), citing a New York congressional candidate's comments about July Fourth as evidence. It is the kind of ending built for a campaign, not a policy seminar, and it is a useful reminder that everything before it, the deficit target, the visa fee, the fraud estimate, was offered in service of that argument, not apart from it.