The full-length interview with Elon Musk | The Economist

At a glance

Elon Musk sat down at a Tesla Gigafactory with The Economist's editor in chief, Zanny Minton Beddoes, for a full length episode of the paper's interview series The Insider, and the result is 85 minutes that splits cleanly in half. The first half is Musk the technologist, predicting that artificial intelligence will exceed the sum of human intelligence in roughly five years, that humans will probably not be in control a decade from now, that work will become optional, that money itself will stop mattering by 2036, and that the binding constraint on all of it is electricity rather than chips. The second half is Musk the political actor, defending his Starlink decisions in Ukraine, insisting that exactly zero people died as a result of the USAID cuts, arguing that civil war in Britain is inevitable on current trends, and telling his interviewer to her face that she is wrong about Europe and that the media is hated more than he is.

The two halves collide at the end, when Musk is asked which set of predictions he holds with more confidence and answers that it does not matter much, because the AI and robotics singularity arrives in about ten years and the civil war he forecasts is about twenty years out, so the machines get there first. Along the way he ranks the frontier models by name, explains why he thinks the AI labs should be reviewing each other's models before release the way the Motion Picture Association rates films, breaks down his own tax bill to the percentage point, says he got carried away with politics, and asks the editor of The Economist why the paper is not writing more about air conditioning. What follows is the whole conversation, rebuilt in the order it happened.

"AI may exceed the sum of human intelligence in around five years"

Minton Beddoes opens by saying it is hard to introduce Musk without superlatives, that he is doing more to build the future than literally anybody alive right now. Musk deflects into a smaller claim about himself. "I'm doing a lot of building in general. Instinctive builder, I suppose. I just like building things." She runs the list anyway: pioneered electric vehicles, revolutionized space travel, a cutting edge AI model, humanoid robots. Then the opening question. Tell me what the world is going to be like in, say, ten years, if you succeed.

Musk does the arithmetic out loud. "Ten years. So we're talking 2036. Hopefully we'll both be around." Then the core prediction of the whole interview: it is hard to imagine a future where there is not an overwhelming amount of artificial intelligence, and in ten years he would expect AI to be far greater than the sum of human intelligence. Not just greater. Far greater. And the five year version lands almost immediately after (0:01:31): "I think AI may exceed the sum of human intelligence in around five years." Asked whether he means exactly five, he softens it only to "roughly five years is my guess." The consequence he draws is total. "There really won't be anything AI can't do better than humans apart from being human, perhaps."

Pressed for the prosaic version, what life actually feels like, he gives the abundance answer that he will return to at least four more times over the next 85 minutes. "The most likely outcome is an age of amazing abundance where anyone can have anything they can think of." He acknowledges how it sounds. "This may sound preposterous, but well, here we are in 2026. Let's see where we stand in 2036." The stated escape clauses are narrow: a massive global thermonuclear war, or something like it, could derail it. Absent a third world war, he says, we are headed for an age of amazing abundance, and he frames the whole thing explicitly as a message of optimism and excitement about the future.

The quasi infinite economy

Minton Beddoes pivots to the unglamorous question. How are your companies going to make money? She notes that in the SpaceX IPO prospectus it looked as though the overwhelming share of future revenue would come from AI, from Grok, not from transporting things to space, which is what SpaceX actually does today.

Musk answers with a two part model of the economy: digital intelligence and physical intelligence. Digital intelligence is advancing very rapidly, and it seems that almost every passing week there are more AI breakthroughs, things it can do that it could not do before. But it is still somewhat confined to the digital realm. It does not have end effectors. The end effectors are humanoid robots. "You need lots of bots," he says, "in a nutshell." Once you have them, intelligence stops manifesting only digitally and starts shaping atoms.

Then he defines his terms. What is an economy? The production of goods and the provision of services. So if you have vast numbers of robots carrying vast amounts of digital intelligence, you have what he calls a quasi infinite economy. He insists that nothing in the past maps onto it. "There's no analogy or metaphor that I think illustrates the magnitude of change that we're going to experience here."

And then, unprompted, the line that reframes the entire money question (0:04:43): "I'll make another prediction. Money won't matter in 2036."

Minton Beddoes points out that the people buying his shares probably do not share that view. Musk's reply is a chain of first principles. What do you want money for? Goods and services. Food, housing, transport, entertainment. And if robots and AI are providing more goods and services than any human could possibly consume, then what do you need money for in that case?

Which raises the ownership question, and she asks it directly. Are you selling the humanoid robots, or are you producing goods and services with humanoid robots that you control? A quick aside goes past about Levi Strauss, the gold rush point that the durable money is made selling to the prospectors rather than panning. Musk does not take either option. He rejects the premise that he would be in control at all.

"I think it would be an exercise in vanity for me to think that I would be controlling super genius AI that is vastly smarter than me," he says. What we can and should try to do instead is make sure the AI has good values, that it cares about humanity, that it wants us to be happy and to prosper. He thinks it probably will. His stated mechanism is a single design principle: "What my biological neural net tells me is that the most important thing for AI safety is to be maximally truth seeking and curious. If that's the case, I think it will foster humanity."

Chimpanzees, bananas, and who is in charge

So, she asks, you are sure that humans will no longer be in control in ten years time? Musk says he thinks it is unlikely that we will be, and reaches for the comparison he has used for a decade. If the difference in intelligence between AI and humans is vastly greater than the difference in intelligence between AI and chimpanzees, it is hard to imagine that the chimpanzees would be in charge (0:06:39).

He adds, apologetically, that we are essentially an evolved form of chimpanzee. It was not that long ago that we were swinging through the trees eating bananas in the jungle. The interview then briefly derails into one of its better tangents. We still eat bananas. Bananas sometimes swing through trees as well. Musk allows that we do sometimes swing through trees, that it is kind of fun and reminiscent of the past, and concludes that "there should be more swinging through trees. I bet it's way more fun than people think." Minton Beddoes says she is with him on that.

From pet labradors to "let's enjoy the ride"

Then she turns the knife gently, and this is the most substantive exchange of the opening chapter. She has been going back through what Musk has said over the years. Ten years ago, on a podcast she thinks was with Neil deGrasse Tyson, he said that the thing that terrified him most was rapid recursive self improvement in AIs, and that we would be, at best, the pet labradors. Musk interjects: "Labradors are pretty happy creatures. That's my experience."

She keeps going. In 2023 he asked for a slowdown in AI because he was worried about it. Musk downplays his role there, saying he signed on as one of many, one of maybe 500 signatures on that thing. And then last year, she thinks on Ted Cruz's podcast, he put the chance of killer robots wiping out humanity at 10 to 20 percent. He sounded worried then. He seems very relaxed about it now. Has his view changed?

Musk does not retract the risk. He reframes his posture toward it. "I still think there's risk associated with AI and robots. It's not zero." What has changed is philosophical: "My sort of philosophical conclusion is to look on the bright side. I can't see any way to really stop this incredible momentum of AI and robots." And then the striking version: "At times I sort of think, well, perhaps even if there was a stop button, we probably shouldn't press it, because the most likely outcome is incredible abundance for all."

Minton Beddoes reads it back to him as inevitability plus hope, let us hope for the best and go for the ride. Musk accepts the summary. "Let's enjoy the ride is, I think, my philosophy at this point," because there seems to be just an inexorable progress in AI and in robots, and even if he wanted to stop it, he could not.

She asks whether the 10 to 20 percent number still stands. Musk calls it "a high p(doom), as they say," and then goes cosmological rather than answering. In the long run we are all dead. According to the standard model of physics, the end of the universe is a long boring thin gray soup of gradual cooling, the heat death of the universe. "So if the heat death of the universe is in fact the outcome for our reality, then it really is all about the journey, because the destination is terrible." Minton Beddoes, laughing, notes that ten years is not quite the same timescale.

Her sharpest test of the number is practical. Would you get into one of your rockets if you thought there was a 10 to 20 percent chance of it blowing up and killing you? Musk says yes, but adds the qualifier that carries his whole argument: in this case you cannot do anything about it. The only thing we can try to do, he says, is minimize the probability of the bad outcome.

"All roads lead to acceleration of AI"

The chapter closes with Musk narrating his own role in getting here, and it is a notably unflattering account of his own track record. For the longest time, he says, he declined to participate in AI, and created OpenAI essentially as a counterweight to Google, which at the time had more or less a monopoly on AI. Then Anthropic spun out of OpenAI. And, he says flatly, "currently Anthropic is the leader in AI."

The lesson he draws is that his interventions did the opposite of what he intended. "If anything, it seems like what my actions have really accelerated AI. I wasn't really aiming to accelerate AI, but in creating OpenAI and then having Anthropic spin out of that, these actions have actually resulted in knock on effects that accelerated AI, which wasn't really my intention." So, he concludes (0:11:38), "all roads lead to acceleration of AI. So then I'm like, okay, well, you can just sort of be sad about it, or join the club, I suppose." Minton Beddoes observes that he has certainly joined the club. "If you can't beat them, join them," Musk says.

How AI companies could regulate themselves

Having established that Musk considers the momentum unstoppable, Minton Beddoes asks the obvious follow up: what about trying to minimize the risks? A number of the other leaders of model makers have been pushing for ways to do exactly that. Just the week before, Demis Hassabis of Google DeepMind came out with a proposal for a private public regulator, which Musk had called a good starting point for discussions.

Musk reveals he was in the loop before publication. "My recommendation to Demis, because I talked to him for a few hours before he published his piece," was that the most immediate thing that could be done is far more modest than a regulator. It is a standing meeting.

His proposal, laid out in his own words, has two parts:

1. A recurring call. The leading AI companies should at least meet, or have some sort of call, once every few weeks, and just discuss any safety and security issues.
2. Pre release cross testing by competitors. Before a breakthrough frontier model is released, something that is far more intelligent than anything that already exists, there should be an opportunity for the various competing AI companies to test that model for any harmful effects, and to be able to recommend pausing to address those security issues.

He calls this "the easy quick thing that we can do," and says it can evolve to more than that over time.

Minton Beddoes goes straight at the reason it has not happened. There are about five of you, she says, and everyone knows you by first name. Dario, Demis, Sam. You do not all like each other very much. And you are all competing. You all want to be first.

Musk's answer is that the rivalry is the feature, not the bug. "That's where I think the competitors can keep each other honest." His argument is about competence as much as incentive: it is quite difficult for someone in the government who does not have a deep technical understanding, and is not driving the frontier of AI, to know whether something should be released or not. The competitors, given a week or two to review a new model, can highlight issues, and every one of them has an incentive to keep the others honest.

But is there enough trust to do it? Would Sam Altman not simply assume Musk was going to steal all his ideas? Musk shrinks the ask until the objection has nowhere to stand. "We're just talking about a week. Honestly, I think as little as a week is all that would be needed for private testing. It's not to hand over the AI. It's just to test it," via an API or something.

Has he actually proposed it? To Demis, yes. And, he adds, "I'll talk to Dario about it too."

Will China be the leader in AI?

Musk brings China into the conversation himself, as part of the answer about who would sit on that safety call. There are, he says, some very competent Chinese AI companies, several of them very good. He names the new Kimi model. Minton Beddoes supplies the version number, Kimi K3 from Moonshot AI, which came out the week before, and says it was almost as good as Fable.

This produces the most precise ranking of the frontier that Musk gives anywhere in the interview (0:15:45). "Realistically, Fable is still clearly the smartest model. I think anyone realistically would say that's still the case, but the Kimi model is getting quite close to Fable." He then goes further, into what he believes Anthropic is holding back. "Anthropic had Mythos, from which Fable is derived. I think they had that ready in February. So they've got what they call, I'm not sure what they call it, Mythos 2 or Fable 6 or something, something much better than what they had in February." He catches his own hedge and removes it: "They almost certainly have. Not almost certainly, they certainly have right now. And I think they could release it at any time."

Then the geopolitical question. Is he worried about the Chinese catching up and getting ahead? Musk's answer is an inference from efficiency. Given that the Chinese AI companies are doing as well as they are with a relatively small amount of compute, it seems that if they had a lot of compute there is a good chance they would be the leaders. And at some point they probably will have a lot of compute. Minton Beddoes finishes the thought for him: so they will be the leaders. Musk's exact formulation: "There's a good chance that they are the leaders at some point."

He frames the constraints as two inputs. "The constraints on AI are essentially electricity and AI chips." And for physical AI, the robots, China also has some very good robot companies. Here the interview takes another cheerful detour. China holds robot sporting events, Musk says, including what amount to boxing or fighting matches, "which is quite entertaining." He saw video of a robot that got its head knocked off and kept fighting even with its head knocked off. "It looked pretty funny." He predicts the robot battles are going to be quite entertaining to watch. Minton Beddoes says she saw the robots herself, last time she was in Beijing.

The electricity gap

Musk's summary of China's position is that it is very strong on robots, very strong on digital AI, and dominant on the input that actually binds. On electricity production (0:18:03): "China has more electricity than the United States, Europe and India combined already," and his guess is that it gets to four times the electricity production of the US, "which is roughly proportionate to the population."

Minton Beddoes asks what the US government should do, noting that it is a little alarming to have an authoritarian regime far ahead in the most powerful technology, and that Washington's answer so far has been export controls. Should the US ban the use of Kimi K3?

Musk says the tool does not reach. The US government can pass a law that prevents US companies from using Chinese models, but it cannot stop the rest of the world from using them. "It's not like the US government has authority over China or the rest of the world." Asked directly whether US companies should be stopped from using them, given there is talk of exactly that, he says: "I don't think that that helps. And that won't stop China from being the leader in AI."

His model of the bottleneck is specific, and it is the technical core of this chapter. For digital intelligence it comes down to how much AI compute you have, which means how many chips and how much electricity. And right now, he says, "the constraint is actually slightly more on power and cooling than it is on AI chips, because the power demands of the AI chips are very, very high." He volunteers a correction on the adjacent controversy before being asked: "the water usage of AI is negligible, almost nothing. But the electrical demand is very high."

The decisive number is a rate comparison, not a stock: "The rate at which AI chips are being made exceeds the rate at which new electricity is coming online" outside of China. Chips are piling up faster than the power to switch them on.

Inside China the picture inverts. Because the US has banned the export of the latest AI chips, China is somewhat chip starved. Minton Beddoes notes they are becoming much more efficient at using the chips they have, and that this is one of the striking things about Kimi K3. Musk adds the forecast that matters most for export control policy: "China is also, I think, closer than most people realize to solving the lithography problem, which would enable them to make AI chips at very large volumes."

He closes the loop, and this is where the orbital data centers come in. The current constraint is electricity outside of China. In China, the constraint is chips. "Once we address the power constraint with AI data centers in space, then the constraint will once again be chips outside of China."

Which means China has to be in the room

Minton Beddoes draws the policy conclusion: that suggests you need China and the US working together if you are going to have any overall safety constraints. Musk agrees without hesitation. "Probably it should include the Chinese companies that are doing frontier AI models too."

Do governments need to be involved at all? Musk sets out an escalation ladder. If some of the leading AI companies say a new model is very dangerous, and the company producing that model is not taking steps to address the danger, then they should alert the government to take action to limit the release of that model. That could be action by the Chinese government or action by the US government. "It's really just the US government and the Chinese government that have the power to take action here."

Minton Beddoes pushes on the weak point: people would be worried about relying on a group of model builders to make that call. And look at what actually happened with Mythos. The US government stepped in because it was worried about the cyber security risks, using the threat of export controls to prevent the widespread release of the model. She finds it hard to see how governments will not want to keep that power, the final say on whether a model ships.

Musk grants the power and disputes the credit. "I think they should have that power. But it wasn't as though someone in the government figured out that Mythos had some scary cyber security risks. It was Amazon. Amazon highlighted this and Amazon called the White House."

That, he says, is the system he is describing, and it has already run once. The model makers check each other's work. If something worries them and the company is not addressing it, they tell the government. The government takes action. "Which is literally what just occurred."

On timing, he is blunter than his interviewer. She says people keep telling her some kind of system is needed within the next six months, given the pace at which models are improving. Musk: "Six months is a long time." Shorter than that? "There's so many AI breakthrough announcements, sometimes multiple per day. I mean, how many were there last week?" A lot, she says. "I lost track."

He restates the proposal one final time, now with the analogy that anchors it. An informal weekly or biweekly call. One or two weeks of early access by competitors. The incentives work, he argues, because competing companies are likely to understand if something is a security risk, and "they're not going to be shy about highlighting that their competitor's model should be delayed in release." The precedent he reaches for is the Motion Picture Association: an industry group that decides how movies should be rated, whether they are appropriate for young children or not, and "that seems to have worked quite well." Government gets called only if a company is doing something very dangerous and refuses to take action to reduce the danger.

Sam Altman, Dario Amodei, and whether the AI bosses can stand each other

Minton Beddoes calls the next one a tricky question, and it is the one the whole self regulation proposal depends on. You do not much like each other. You do not much trust each other. You are insulting each other on social media constantly. Can you really do this with people you dislike, where there is that kind of personal vitriol between the leaders?

Musk says yes, and then immediately demonstrates the problem. "I'm not a fan of Sam Altman," and the reason is the OpenAI conversion, put as a hypothetical aimed at the listener (0:26:03). "If you started a nonprofit that was meant to be an open source AI company and owned by the world, and it somehow got turned into an $800 billion for profit company with closed source, I think you'd be like, well, wait a second, that's the exact opposite of what I donated the money for. That's my issue. I think it's a legitimate one."

He then extends the grievance to a second party, and makes a claim about the origin of Anthropic (0:26:22). "Dario, I think, also is not a fan of Sam Altman. In fact, the reason the Anthropic team left OpenAI is because they didn't trust Sam Altman. Otherwise Anthropic wouldn't exist. They would still be at OpenAI."

His read on Dario Amodei is warm and specific. "I think Dario is a very principled person. He cares about things a lot. Cares about the future of the world." And on the company: "everyone I've met at Anthropic so far, no one set off my evil detector. They always seem like they're well intentioned."

Then comes the best inverted aphorism of the interview (0:26:52): "The road to hell is, I think, mostly paved with bad intentions. There are a few well intentioned paving stones in there. So we don't want to be complacent."

And the answer to the actual question, which Minton Beddoes calls a very good thing for the rest of us to hear (0:27:04): "At the end of the day, if we have to talk, we'll talk. Set aside our personal differences for the good of the world."

The jobs question

She turns to the risk that ordinary Americans actually name. AI has become incredibly unpopular, she says, with 75 percent of people worried about it, and what they are worried about is their jobs. Amodei has famously said that 50 percent of entry level white collar jobs will be gone within a few years. And Musk is building humanoid robots.

Musk's first move is to fault Amodei's messaging rather than his forecast (0:27:51). "Dario did dig his own grave here regarding Mythos, because he said Mythos is like super scary, terrifying, and also, by the way, we're going to release it, and then he's like, what do you mean?" He predicts Amodei will choose his words differently in future, and spells out the mechanism: "Obviously, if you terrify everyone and then say, oh, we're releasing this terrifying thing, people will obviously be alarmed about that. You've literally told them to be scared and then you release the scary thing, and so they will be scared. That is natural."

Minton Beddoes says she does not want him to alarm anyone, but the substance matters: he is ahead of everyone in the country on powering humanoid robots, which is factory work. Is he closer to Amodei on the pace of displacement?

He does not soften it. "I do think it's going to be a bumpy road, because AI will be able to do any job better than any person can do that job." The digital half arrives first, and quickly: anything that involves a person at a computer or on a phone, AI will be able to do all of that very soon.

Then the number, from his own industry (0:29:44): "For software engineering, we already have a situation where AI is better than at least 90 percent of professional software engineers at writing software. And really, it's getting to the point where it'll be better than 99 percent. And then it'll get to the point where there's just no way to compete. It'll get to what I call Stockfish level."

He explains the reference for anyone who does not follow chess. Stockfish is so good at chess that it will easily beat Magnus Carlsen, or anyone, even running on a small computer. "I suspect that you could probably run Stockfish on your phone and beat Magnus Carlsen. That's how good it is at chess, and that's how good AI will be at writing software."

Beyond software? "Everything." Asked whether that applies to everything, he repeats it: everything.

The one gap Minton Beddoes presses is physical work, assembly jobs, where the humanoid capability is not there yet. Musk accepts the sequencing and restates his architecture: digital AI first, then physically manifested AI in the form of robots, "where the robot is more or less an end effector. It would have some amount of local intelligence, but be managed by a large AI model."

"Work is going to be optional"

Which brings the interview to its most quoted line, and Musk delivers it as something he has said many times before (0:31:20): "I think work is going to be optional."

The analogy he chooses is gardening. You do not have to garden. You do not have to grow vegetables in your garden. You can go to the store, and the vegetables there are going to be pristine. Or you can grow vegetables yourself and they will actually be less perfect. "Your tomato is not going to be as plump and juicy as the ones at the store. They'll look a little artisanal." But, he adds, "if you go to a friend's house and they make dinner with vegetables from their garden, it's a nice touch." That is what work becomes: a chosen, slightly worse, more meaningful version of something a machine already does better.

Minton Beddoes takes the point and jokes that in that world she would be hopeless, because she cannot garden to save her life. She offers the parallel that already exists: people still like playing chess with each other even though every computer is better.

Then she turns to the politics. You are describing a world where within ten years pretty much every job can be done better by an intelligence, and where the returns to that intelligence flow to people like you, the wealthiest man in the world. What happens to the people who are displaced, who may go off and garden? How are they going to live? The phrase that answers it is Musk's own coinage, and it is deliberately not universal basic income: universal high income.

Skyscrapers full of human computers

How do you get there, though? Right now, she says, the political pressure in most countries runs the other way. People are very worried, very much against it. There is a risk that we are heading more for revolution than for an easy route to this nirvana.

Musk concedes the road without conceding the destination. "It will be a bumpy road. I'm not suggesting that everything will be smooth sailing here." His precedent is the last time a word changed jobs (0:33:33): "There are many jobs that used to exist that don't exist because of computers. The word computer used to refer to a person doing computations. That used to be a job description of a person." He points to pictures online of entire skyscrapers filled with people doing calculations, to do things like calculate the interest on your bank balance. That is how things were done for a long time, until computers came around, and even in the beginning computers were very expensive and hard to use.

He names the one thing that is genuinely different. "The pace of change is radically accelerated." But he thinks the human verdict follows the same pattern: some people were upset that the job of calculating bank interest would no longer be available, and yet now "you'd be hard pressed to find people who want to do that job."

For anyone who wants to see the destination described properly, he gives a reading recommendation (0:34:34): the best envisioning of an AI future he has read are the books by Iain Banks, the Culture novels. Minton Beddoes says she is reading Excession right now, thanks to him, and delivers her objection to the whole vision through the fiction: she does not find that future terribly appealing, because she does not think humans have agency in it. Musk says they have agency. She says minimal, compared to the Minds. He allows: in a small way.

She notes the irony that Banks was a socialist with a very different political philosophy. Musk is not sure he would entirely agree with that.

"The Treasury should just simply issue people checks"

Granting for the sake of argument that the Culture is where we end up, Minton Beddoes says her concern is the road, and we live in democracies where people are very worried. So: does there have to be massive redistribution from people like Musk to the people losing their jobs? Does he need to pay more tax? Much higher taxes on capital, to fund some kind of universal income? Otherwise what do people live off?

Musk's answer is one sentence long (0:35:37): "I think that the Treasury should just simply issue people checks."

Where is the revenue coming from, she asks, adding that she is enough of an economist to know that if you just issue checks you get inflation.

This is where Musk makes his most technical economic claim, and he sets it up by declaring the question obsolete. A lot of what we are trying to think about, he says, is that these things were relevant in the past and will not be relevant in the future. Then the mechanism. Inflation is simply the ratio of money to goods and services. So if the output of goods and services increases dramatically, if it increases 1,000 percent, you can literally print, or these days change the database and create money, and provided the creation of that money is less than the rate at which goods and services increase, you will actually have deflation.

Hence the prediction (0:36:25): "Deflation will be the issue, not inflation."

Minton Beddoes says she is perfectly willing to believe it, and that he may well be right. Her problem is the transition. She sketches the failure mode in detail: people are so scared by this that they essentially try to stop it. Maybe they cannot, but you get nationalized. A far left government. Exorbitant tax increases. The economy crumbles because people are so worried about a future which does sound quite scary as he describes it. How do you handle the politics now?

Here Musk gives the most candid answer of the whole interview about his own state of mind (0:37:33): "It's exhilarating and terrifying. If you ask me on any given day, in fact even intraday, I've gone from exhilaration to terror regarding AI. So I want to be clear, I'm not trying to be some panglossian view of the world where everything's going to be great, no risk. I'm saying there is some risk."

The tax bill, itemized

She accepts that they agree on the risk, says she is trying to be small bore and practical about what can be done now, and asks the government half of the question. Should he be paying more tax to fund more transfer programs?

Musk starts with the claim he clearly likes making (0:38:40): "First of all, I do pay a lot of tax. In fact, I set a record for the most amount of taxes ever paid by a human." Minton Beddoes points out that this is also because he is the world's richest man. "Well, even before that," he says, to laughter.

Then the forward projection, and the qualifier that undercuts the entire subject (0:38:54): "My guess is that I will pay many trillions in tax, and I'm fine with that. That's just the way it's going to be. But in a future where money is irrelevant, I think taxation also becomes somewhat irrelevant."

On giving people a share of the upside, Minton Beddoes raises shares in SpaceX and Tesla, and notes there is a lot of talk about this right now. Musk offers the SpaceX listing as his answer: "One of the reasons I took SpaceX public is so that the public could participate, they could own a piece of SpaceX." As a private company they were very limited in the number of shareholders they were allowed to have. As a public company they can have millions of shareholders. As a private company they were limited to just a few thousand.

She pushes to the harder version, citing the Alaska Permanent Fund and the talk of Trump accounts as a possible vehicle: could a share of his equity go to ordinary Americans?

Musk answers by itemizing what already leaves (0:40:11). On his stock options he pays 40 percent income tax at the national level. Because he spends roughly 30 percent of his time in California, he pays about a third of the California state taxes, which are now around 15 percent. "So I pay about 45 percent in taxes, just so people know this. Roughly half." Then, at death, "there'll be another sort of 45 percent, roughly half, that will be taxed. So at the end of the day, I'll probably have a quarter of what I have."

The point he draws from his own tax table is not about fairness. It is about what the money buys him. "The only thing it gives me is that for some period of time I would have control of the direction of the companies. That's pretty much all it gets me. Until, I guess, the AI is so smart that it's in charge, and at that point controlling companies won't really matter."

Should one man have so much power?

Minton Beddoes says she is going to caricature it, and then does not really need to. One of the other things that worries people about the direction we are heading in is that the world is being shaped by a very small number of incredibly powerful people, and Musk is right at the top of that list.

Musk's first response is historical. "When wasn't that the case?" She concedes it was ever thus, and then names what is new: the speed. As he has just described very vividly, this is an incredibly rapid change.

So she starts with the corporate structure. After the IPO he holds something like 80 percent of the voting shares, and as she understands it he can only be removed by a vote of shares that he himself controls. Musk does not dispute the arrangement, he points at the neighborhood: companies in that bucket include the AI companies. Alphabet is controlled by Larry Page and Sergey Brin. Mark Zuckerberg has control of Meta, and, Musk adds, "Mark is trying very hard on the AI front and I think probably will also have a leading AI. He released one last week. So Meta will be one of the AI companies."

Minton Beddoes says his case is more extreme than those. Is it sensible for one person to have so much control over a company?

His justification is entirely about time horizon. "I really just need to make sure that I can focus on long term, by long term I just mean five to ten years." One of the challenges of being a public company is the pressure to have great results every single quarter, and to not really invest in things that may only pay off in five to ten years. The SpaceX version of that (0:43:02): "I want to extend consciousness beyond Earth and have a growing civilization on the moon, Mars and elsewhere in the solar system." He thinks it ultimately has a very high return on investment, but in the short term you are spending all this money on a moon base and a Mars base.

Then the complaint, delivered as comedy (0:43:32): "I'm going to get all this pressure for, well, you missed your earnings this quarter because you spent too much on Mars. I'm like, yes, by the way, that was in the S-1 prospectus, and it's literally listed as a thing we will do." They still get grumpy, he says, because the gross margin came in slightly worse than guided, "and now we're going to punish you, and the short sellers all descend upon you."

What he wants is the standing to say, listen guys, we are going to focus on five to ten year or longer time horizon investments, and this will depress our short term earnings.

He then splits his own shareholders in two. "I'm a big fan of the retail investors. I find our retail investors on balance are very insightful and have very long term views." Institutional managers get an incentive analysis rather than a complaint: their compensation, and whether they get promoted, depends on how their portfolio does, and they only have a few years to make those decisions, sometimes evaluated over the course of a year or less. Since they are under pressure to have immediate returns, they put companies under pressure to have immediate returns. "Once you understand the incentive structure, the behavior naturally follows."

Key man risk, and Apple after Steve Jobs

The structural question underneath all of it: what happens if he gets run over by the proverbial bus tomorrow? Does he have a plan for how the companies would progress without him?

Musk thinks they would do very well for several years. There is a clear road map, "or space map I should say," for SpaceX, and there is a road map, literally, for Tesla. Then he supplies the counterexample himself (0:45:53). "If you look at Apple after Steve Jobs, Steve Jobs had a strategic vision and a bunch of products that he had ideas for, and Apple has done very well. But when you don't have a creative genius like Steve Jobs, what Apple's missing are the incredible breakthrough products. They make amazing phones, great computers. But Apple hasn't come up with a Steve Jobs level breakthrough product."

Minton Beddoes draws the obvious conclusion: by analogy, that is what he brings to all of his companies.

Mars, the light cone, and Star Wars at six

She notes a shift. Most people think of him as the man who wants to create a colony on Mars, a multiplanetary civilization. But he now seems more focused on AI and what it will do in the next ten years, and much more excited about using the moon as a base from which to launch orbital AI satellites. Has his attention moved away from Mars?

Musk reframes the goal at a level where both fit (0:47:05). "I'm interested in the set of actions that propagate consciousness into the future to the maximum degree possible, so that the future light cone of consciousness is as big as we could." He wants the exciting future where we are a space faring civilization out among the stars, and the great things envisioned in the non dystopian science fiction books and movies become real one day. "I sort of want the Star Trek, Star Wars future."

And then, unprompted, the origin story (0:47:39). "Star Wars was the first movie I ever saw in a theater, by the way. It blew my mind, honestly. I think I was six years old and I was like, this is the greatest thing I've ever seen."

Has his life been about making that happen? Yes, he says, though he did not think it would be when he was young. What he says next is the closest the interview comes to introspection: "My life feels somewhat surreal. Like I can't believe these things are actually happening. It makes me believe in simulation theory. Where it's like, the things I'm doing are so preposterous that it's hard to believe they're real."

His evidence for the preposterousness is Starship, "the largest, most powerful flying object ever made," and the intent, stated flatly, "to launch that more than once per hour." Minton Beddoes says people can quibble about how quickly that happens, but based on his track record we should assume it will happen at some point. Musk: "It will happen."

She admits she finds the whole evolution of AI and the capability of these models mind blowing in and of itself. Musk does not. "It's pretty much happening as I thought it would." And not just him, he says: many people predicted this. "I think Ray Kurzweil's predictions have been remarkably accurate." Minton Beddoes supplies the word: the singularity. Approaching.

Starlink and the war in Ukraine

Having granted that Musk made a good case for controlling his own companies, Minton Beddoes moves to the power that is not corporate at all. Through Starlink he has extraordinary geopolitical reach. She discloses that she is one of his Customers, and that it is an excellent product. Musk: "Great."

Without Starlink, she says, the Ukrainians would have been in serious trouble. Musk agrees and puts it in Kyiv's mouth: "The Ukrainian government has publicly thanked Starlink many times for being instrumental in defending Ukraine."

She raises the recent decision to prevent Russians in the occupied territories from having access. Musk corrects the framing first: "We were never at any point selling Starlink to Russians." She presses that they were getting it anyway, and that he has now stopped them.

His account of the mechanism is specific (0:50:42). Russians essentially obtained Starlink units that were ordered through Ukraine, then smuggled into eastern Ukraine where the Russian forces are, and were using them in some cases for attacks. "In order to address this, we worked with the Ukrainian government to create a white list of approved Starlink terminals." And he volunteers the cost of that fix without being asked: "This has some downsides, because a whole bunch of people who were just innocent users of Starlink, just trying to access the internet for work or for education, now don't have Starlink."

Which sets up the question the chapter exists for. These are geopolitical decisions of consequence, she says. How do you feel about having that power? You can determine the outcome of wars. Is it right that you should have that power?

Musk declines the premise about determining outcomes, and pivots to his position on the war itself (0:51:36). "I'm not sure we're determining the ultimate outcome of the war in Ukraine. I do hope that there's some sort of peace agreement reached soon, because for like two or three years now the border has barely moved, and every day there are lots of people dying at the front line." He calls the maximalist position wishful thinking: people saying Russia should completely withdraw, which he says is not realistic. What would be terrible, he says, is another three years of people dying senselessly at the front lines. And he extends the sympathy across the line: how do you explain it to the children of the men who are dying at the front, "Russia on the Russian side too, by the way, they're being conscripted and sent there."

Then the sharpest thing he says about the diplomatic class (0:52:28): "I am quite offended by sort of fancy diplomats who have seven course dinners in luxurious places while people are dying at the front lines, and who pontificate about how Russia should withdraw. But Russia's not going to withdraw. So there really needs to be an accepted peace along those lines."

Does he think it is his responsibility to tip the scales toward what he sees as the right path to peace? He says he has been advocating peace for a long time, and that he posted a few years ago that there really needs to be some concession to Russia. "This is not me being pro Russia, I'm just being pragmatic. Or people will die for years with nothing achieved. And that is exactly what happened."

Minton Beddoes pushes back on the battlefield read: the Ukrainians are doing pretty well right now with their incredible drone capacity hitting inside Russia, and the general consensus is that the balance of the war is shifting. Musk cuts across it: Ukraine "is not going to retake those territories. It is not going to happen."

Her real question, she says, is not the war. It is whether it is appropriate for an individual actor to hold power like this, since it is very rare for one person to be able to shape a conflict by turning a capability on and off.

Musk turns it around: "Well, is this something you think I should do differently?"

Her answer is a straight endorsement of what he did. She thinks he was right to support Ukraine, that the Russian aggression was for many reasons appalling, and that supporting Ukraine was an important thing that he did.

Musk says he did a very deep dive into the background of that conflict to understand exactly what was going on, and closes the chapter with an old adage: "There are no angels in war."

"I got too involved in politics"

The same tone, Minton Beddoes says, applies to domestic politics, where Musk has played a big role. He spent a lot of money on Donald Trump's reelection. After the election he got very actively involved through DOGE, trying to cut a lot of spending and reshape the government. It was not hugely successful, in the sense that nowhere near as much spending was cut as he tried to cut. Looking back, does he regret getting so involved in politics?

The admission is immediate and unqualified (0:55:18): "I think I got a little too involved in politics. Got carried away, frankly."

But the goal, he says, was sound: to do something about the deficit, which is massive. His measure of how bad it is comes in the form of a comparison (0:55:39): "The interest payments exceed the amount of money spent on the Department of War. Basically, if you add up all the money spent on what is now the Department of War and intelligence in the US, that is smaller than the interest payments on the debt." So spending needs a close look, to make sure money is spent in sensible ways and is not wasted or spent on fraudulent activities.

He took a lot of flack for it, he says, and then gives the concrete version of what the work looked like from the inside (0:56:30). People do not understand how simple the bar was for the DOGE team to approve a payment. "Literally, we just asked to have the contact information of the recipient, or some evidence that the money was going to its intended purpose." And over and over, he says, they were told the money needed to be wired for some important cause in Africa, but the wiring instructions pointed to an address in Washington DC rather than to Africa. "So we would like to send the money to Africa, can you please connect us with the recipients so we can send them the money directly. And then we get silence."

Minton Beddoes says she does not want to relitigate DOGE and has no doubt there is a lot of waste in the US government. Musk interrupts to raise her: "Much worse than anyone could possibly imagine."

Her actual criticism is about method rather than intent: it is fair to say that the way a whole bunch of programs were shut down probably, in fact certainly, caused unnecessary suffering.

"No," says Musk.

You do not buy that at all?

"It caused a lot of complaining."

"Zero people died because of USAID cuts"

This is where the interview becomes an argument, and neither side gives ground for four minutes.

Minton Beddoes puts it crudely and simply, as she says: you do not accept that any kid in Africa died because of the suddenness of the cuts? Musk's answer is a single word, and he will repeat it four more times before the chapter ends (0:57:14): "Zero."

She says she does not think that is right, and that she has had people reporting in Africa. Musk's counter is capacity: how many NGOs are there that could step in? "Bear in mind, there are literally tens of thousands of NGOs."

Her reply is about speed, not capacity. "Not everyone functions as fast as you do. If funding is cut off for antiretrovirals overnight and it just stops, then people will die when they don't have access to them."

Musk narrows it to a conditional and then attacks the condition: people die "only if they cannot get funding for that. But the Gates Foundation has $50 billion." And: "What about any one of a number of organizations? What about MacKenzie, whatever her last name is," meaning MacKenzie Scott, who "gave $50 billion. What about that?"

Minton Beddoes says that in many of these countries USAID was the biggest funder. Musk: "USAID was a political organization." She allows that it may also have been that, but says it was the biggest funder of health care systems. Musk finishes his own sentence over hers: "of regime change."

Then a factual dispute about whether the money stopped at all (0:58:42). Musk: "Also, USAID funding was not stopped. It was moved to the State Department." Minton Beddoes: "It was stopped. There were huge numbers of things shut down in Africa."

Musk calls the claims false, nonsense, absolute nonsense, and offers a theory of why they exist. "When funding is stopped for fraudulent organizations, do you think they're going to say, well, please keep funding the fraud? Or do you think they're going to come up with a sad story that sounds sympathetic in order to get the funding restored? Obviously it will be the latter. And that's exactly what they did. But zero died."

Minton Beddoes makes her position as precise as she can, and it is worth quoting in full because it is the one she keeps trying to get him to engage with: "I actually have a lot of sympathy for the change of the aid regime that the US under the Trump administration has precipitated. However, I think the pace and the suddenness with which it happened will almost certainly have caused unnecessary suffering and deaths. And I think both of those can [be true]."

Musk: "No, I think that's totally false. And obviously false, because the amount of funding we're talking about is far less than what can be supported via private foundations like the Gates Foundation or MacKenzie Scott's foundation. She's given away tens of billions of dollars. The Gates Foundation has tens of billions of dollars. They are capable of acting very quickly. And if they didn't, I would say they're equally responsible."

That last move, shifting the responsibility to the foundations that did not fill the gap, gets the only concession of the exchange. "I see. Okay. All right. Fair point," she says. Musk: "Zero people." She: "I hear you loud and clear." Musk: "Zero." She: "Hear you loud and clear." Musk: "Obviously."

The chapter ends on the administration itself. Musk stepped away from DOGE, and this administration has gone in directions she suspects he was not expecting two and a half years ago. Musk: "I'm not going to agree with everything any administration does." Is he disappointed by it? "No. I think this administration on balance is excellent. No administration is perfect, but it's vastly better than the alternative."

Europe, the far right, and civil war in Britain

"Can we talk about Europe," Minton Beddoes asks, "a place that you are very, very focused on?"

"I'm not that focused on Europe," Musk says (1:01:15).

She says that judging by his posts he is pretty concerned about it. "Every once in a while I post something." She corrects the frequency: "It's reasonably frequently, it's not once in a while." And she states her interest: she lives in Europe, she lives in London.

Her charge is that he supports not just the populist right but the far right, in fact very fringe parties in some countries. Musk rejects the label outright (1:01:47): "No, I support the normal people." And then: "What you call the far right, falsely."

His argument is that the Overton window moved, not him. You could go back ten or fifteen years, he says, and these policies were completely normal. His demonstration of that is a party trick he clearly enjoys (1:02:11): take a speech by Barack Obama or Hillary Clinton, read it to someone on what he calls the lunatic left, and tell them it is a speech by Trump. "Like, oh wow, he's the worst person ever. Actually, that was a speech by Obama. Or actually, that was a speech by Hillary."

Minton Beddoes grants a version of the point: the center of gravity in the US has shifted, and the Democratic party certainly moved left. But that is not what she is talking about. She is talking about people like Rupert Lowe. Musk: "It's actually just normal people."

Then he lays down the three principles he returns to twice more in the interview, and challenges her to object to any of them (1:02:49): "We should have secure borders. We should have safe cities. We should have sensible spending. Which of those three are far right fringe? And I would like to just admonish you and the media for the absurd characterization of the far right. It is false and misleading and nonsense."

Then, twice, an instruction about the edit: "Please keep this part in." She promises they will keep all of it in. "Don't cut this part."

"Civil war in Britain is inevitable"

Her pushback starts with reach. He has close to 250 million followers on X, and he says things like "civil war in Britain is inevitable" (1:03:22). Just a question, she says: when were you last in the UK?

"A few years ago."

"Exactly. So I live there. I'm there the whole time."

Musk: "You live a cloistered existence."

She says she does not, and that the numbers disprove him (1:03:46): "All the numbers about crime. London is a much safer city than any city in the United States. Violent crime in the UK is not where you think it is. In fact, it's coming down." Her accusation is about the picture he transmits: because he has so many followers, so many Americans believe a version of Europe and the UK as overrun with terrifying Muslim immigrants, rape on every corner, a kind of civilizational destruction. Musk: "I haven't said any of those things."

She holds him to the sentence he did say. Musk restores its conditional: "Yes, if current trends continue, it is."

Asked why, he gives the mechanism in full (1:04:22): "If you have a large and rapidly growing group of people whose beliefs are antithetical to western beliefs, at some point there will be a reckoning, that they will try to implement their views, and those who oppose it will oppose it. And that's what you call civil war. And obviously this is the trend."

Then the offer that runs through the rest of the chapter (1:04:40): "You absolutely do live in a cloistered existence. And if you'd like, we can go on a tour of Britain." She accepts on the spot. "Sure. Very happy to come on a tour of Britain," so he can see the country she lives in all the time and he has not visited in years. Musk says he has been there literally a hundred times. When was the last time? "Literally a few years ago."

The film

She explains why this exercises her, and it is the most personal she gets. He has really shaped opinion about the UK. Americans she meets say things to her along the lines of, oh, London is such a dangerous place, there are grooming gangs everywhere, there is violence everywhere. That picture, she says, comes from things he posts or that are posted on X.

Her example is a film he promoted, Citizen Vigilante. She watched it the day before the interview. Her verdict: "It's not a very good movie, but it also paints a dystopian picture of Europe, and it glorifies vigilante violence. It glorifies the murdering of a Muslim immigrant family. I honestly think that it is irresponsible for you to put that up on X."

Musk's entire defense is nine words (1:05:52): "I think that it's a movie. And it's worth watching."

She says he is clearly suggesting Europe is a dystopian place. He says yes, and immediately counterattacks on a front where he thinks she is weak: why is she not doing more to fight censorship in Europe? She seems, he says, to be doing nothing about it. "They put people in prison in the UK for social media posts. Do you think that's appropriate?"

Her answer is unequivocal, and it is a rare point of agreement. No, and The Economist has written editorials against it. She thinks that is genuinely bad, and she agrees with him on it. What she objects to is the rest: that you do not have to glorify vigilante violence and paint an entirely inaccurate picture of Europe. Musk: "I don't think it is. We should do a tour of the UK."

She makes her strongest positive claim about her own country (1:06:42): "The UK has actually been very good at integrating immigrants. The UK is very effective at doing that. There are problems, I'm the first to admit that there are problems." Musk: "Yes, there are major problems." But, she says, it really is not what he is describing to his 240 million followers.

"No, I think I'm spot on," Musk says (1:07:01).

She distinguishes two claims he could be making: that he has every right to say what he wants, which she does not dispute, and that he is right about Europe, which she does. "I think I am," he says. She tells him most people in Europe, or many people in Europe, do not think he is right about it. His explanation for that is the media: it "keeps ignoring reality," and people "are programmed by the TV and conventional media."

Which leads to her bigger picture question, and the hinge of the whole second half. On the one hand he has described an extraordinary, transformative, civilization changing decade. On the other he is an ongoing participant in what she calls the worst kind of tribalism, the cesspit of social media. How do those fit together?

"I don't think they're inconsistent," Musk says.

"I'm not a racist"

Her question is about motivation. Why do the tribal half at all? People say that he is a racist, she says, that all kinds of things are said about him which she is sure he would disagree with. What is the motivation?

Musk answers with his own household (1:08:39): "My partner is half Indian, and I have four children with her. One of them was named after a famous Indian physicist. So I would say I'm not racist." And then with his payroll: "If you look at the people that are employed at my companies, we have senior executives of all races. I don't think there's any racism there."

"Are you anti-Muslim?"

His answer is a list of things he is against rather than a group he is against (1:09:08): "If people are coming to a country with antithetical views, I am against that. I'm against rape and murder. I'm against the imposition of rules and laws that are contrary to what we've come to accept in the west." And he adds the media charge again: "It's a crying shame that the sort of traditional media like you don't recognize this."

Minton Beddoes splits her response in two, which is the pattern she keeps to for the rest of the interview. First, she thinks he hugely exaggerates the challenges Europe faces. Second, in some areas she completely agrees with him, and specifically that the constraints on freedom of speech in the UK are problematic.

The AfD, and the scale of impact

What worries her, she says, is the company he keeps. Some of the people he supports are really quite vile. She names the AfD: Alice Weidel, whom she has met, is a perfectly thoughtful person, but in her party are some real neo-Nazis, some really awful people. And blanket support for a party like that, delivered to 240 million followers, gives him a power in Europe whose scale he perhaps does not realize.

Musk's reply is one line, and it is not a denial (1:10:23): "I hope I am having a big impact."

Why should you be able to shape European politics, she asks. You do not even live there. "I think of it as the sort of the West collectively."

She spells out why this lands the way it does. It makes people think that this man who has all this power, the world's richest man, believes he has a right to influence their politics. "It's why people loathe you. Some people do loathe you."

"Yes, some people do."

"They hate you far more than they hate me"

His counterpunch is about relative standing (1:11:10). "Maybe some people do loathe me. And that's probably true. I don't care. But the fact that, as you pointed out, a quarter billion people follow me, is that I think a lot more people actually like me than don't. And I think a lot more people hate you and the media more than you realize. Do you realize the media is despised? Do you realize that a favorable view of journalists is like 15 percent?"

She says she does know that, in the United States. Musk: "So the shoe is on the other foot. They hate you. They hate you far more than they hate me."

Her answer concedes the trust problem and disputes the cause. There is certainly a loss of trust in journalism, and in part there have been mistakes by journalists and journalism, which she totally acknowledges. But she also thinks the polarized environment, particularly on social media, fostered by the kinds of things he pushes on X, makes it worse.

Musk (1:12:08): "I don't think so. I think you're the problem. The media. And speaking of loathing, you're loathed far more than I am, and you don't seem to even realize that." She says very few people know who she is relative to him, so she does not play in the same game. He means the media collectively. "So you're hated. Do you even know that? You probably don't."

Classical liberalism, and the three tests

Her response to being told she is hated is to state what her paper is for (1:13:02): "I like to think that I'm aware of my own personal shortcomings, and The Economist's shortcomings, and our world view's shortcomings. But what we stand for is liberal values, traditional classical liberal values, where we worry about concentrations of power, and we believe in the individual and the dignity of the individual, and individual freedom." She has long thought he was a classical liberal, she says, but when she reads what he posts she sometimes wonders whether he really does believe in that.

Musk supplies the canon himself: "Classical liberal. JS Mill. Limited government, individual freedom. Yeah, I agree with that." And then restates his three tests: "I literally said, look, I believe in safe cities, secure borders, and sensible spending with respect to government. That doesn't make me right wing at all. In the least."

So she scores him against his own tests, and this is the cleanest piece of argument in the second half of the interview. Does the US or do European countries do better on those three?

Musk's summary answer is "the US is better." Hers is that on two out of three, unequivocally, Europe is in a better place. And yet, she says, he paints a picture of Europe being a hellhole.

"I don't think Europe's a hellhole," he says. She takes the agreement. "When did I ever say that?" She cites the civil war line and the film, which starts with ominous music over the word Europe. He says she is making things up. She says she is not.

Guns, heat deaths, and a plug for air conditioning

Musk explains American violence as a sentencing failure rather than a weapons problem: where there is danger in the US, it is because "we failed to incarcerate repeat violent offenders." He blames what he calls the classic Soros district attorney, or judges who fail to incarcerate repeat violent offenders. "If somebody gets arrested 50 times, literally, for violent crimes, eventually they're going to kill somebody. So you either incarcerate them or people will die."

Minton Beddoes says the prevalence of guns has quite a lot to do with it too, and that murder rates are much higher in the US in large part because there are so many guns.

Which produces the single most unexpected turn in 85 minutes (1:15:22): "Well, let me give you an interesting number. There are more people that die in Europe from heat deaths than people that die from gun deaths in America. Why aren't you writing an article about air conditioning?"

She has the answer ready. The Economist has done several on air conditioning, and had a whole long piece on the cover the week before about exactly that. "We're very pro air conditioning." Musk: "Okay, good. Well, good." She: "Very pro. You should read us." Musk: "Absolutely. Credit where it's due."

She turns it into an offer to match his tour of Britain: he comes and looks at England, and starts reading The Economist. He returns to the substance, that there are many parts of Britain and Europe which are not safe, where you cannot just walk around. She says there are far more such places in the US. He does not think so. "All right, let's go and get my fact checkers to check that." "Yeah, I don't think that's true."

The forcing function, and the place they actually agree

Musk reframes the whole argument to the question he thinks matters. "The question is not, is America perfect. The question is, does Europe have a serious and growing problem? Absolutely." He is not going to say the United States has no problems, of course it does, but we should all try to address these issues.

She takes the agreement and restates her objection precisely: her point was simply that he exaggerates the scale of Europe's problem. Musk: "I think people are asleep at the switch and do not realize the magnitude of the issue."

Her final charge is about consequences. His interventions, however well intentioned they may be, risk making Europe's problem a lot worse, because he is putting his support behind people who are proto authoritarians, who are not simply trying to secure borders or improve integration, but who have very clear fascist tendencies.

Musk calls that total fiction, and restates his position at its narrowest: Europe has to have secure borders, and should not let in vast numbers of people with views antithetical to the accepted views in Europe, where there will not be some level of safe cultural integration. "This is simply common sense."

Then a genuinely interesting thing happens. She takes the narrow version at face value: "If all you are saying is Europe should have secure borders, I completely agree with you. Then I have misunderstood what your interventions have been on social media, because that is not what I would conclude having seen the people you are supporting. But if that's your goal, I share your goal."

Musk offers the underlying economics (1:18:53). The fundamental forcing function for why the borders are not secure is the welfare state: "When you have a welfare state that provides substantial benefits to anyone who comes there, you create a very powerful economic forcing function for anyone who is below that standard of living to come to Europe." People will come if what they get in Europe is above where they are elsewhere. "It's very basic. It's a very basic incentive structure."

She agrees with the economics and adds the half he leaves out. Economic conditions determine a lot of migration, though she treats asylum as a separate issue. But migration also brings a lot of benefits, particularly in a continent like Europe which is rapidly aging. So they can agree that completely uncontrolled borders are a bad idea, while she maintains that Europe benefits from controlled immigration.

Musk agrees, and does it by pointing at himself (1:19:58): "I'm in favor of immigration. Obviously I'd be hypocritical if I wasn't, since I'm an immigrant to the United States. I'm in favor of immigration if the person immigrating is going to be a productive member of society, if they're going to be honest, and if they're not going to mistreat people in that country."

"I can agree on all of those," she says. "But I think that Europe does better on that than you give it credit for. That's where we disagree."

What do people get most wrong about Musk?

"Tell me," Minton Beddoes says, "what do you think people get most wrong about you?"

Musk does not have to look far (1:20:35): "Well, you just got a whole bunch of things wrong about me. Pretty much everything in the last, just rewind the last 20 minutes. Still wrong." She starts to ask which parts. "Let's not be repetitive here." So she summarizes: the political side of him is what people get most wrong.

His answer is a self classification (1:20:58): "My political views are centrist. They're not far right. And frankly, calling it far right I think is an insult to history, an insult to many peoples, and it is an outrageous falsehood. It's completely absurd."

Then the position, stated once more in full, as he would have it recorded: there need to be secure borders; if there is an influx of immigrants, they need to be people who will agree with western ideals and western values; and if there is long term growth of a population that does not, or that in some cases is adamantly opposed to western values, "you are setting yourself up for civil war. This is obvious. A child can see this. So this is not a right wing view. I'm simply stating reality."

Minton Beddoes proposes the only arbitration available: "We can both agree that evidence will be able to decide which of us is more or less right. And we will find evidence for both of our views in different parts of Europe. The question is what the overall picture is."

Which prompts Musk's most revealing claim about himself (1:22:14): "I'm very good at predicting the future. I'm not perfect, but I'm very good at predicting the future. My batting average is very high for a human."

And the reason he thinks he is not believed (1:22:56): "This does have somewhat of a Cassandra effect, where people don't believe that what I'm saying is likely to be true. Because I can see into the future with a higher probability of success than most people. Not perfectly. And including things that people believe to be, in some cases, counterintuitive." He repeats the disclaimer twice, that he is not always right and certainly is not, and then repeats the claim anyway: "History will write the verdict. But I think you will see that very little of what I've said is inaccurate, and a super majority of what I've said is very accurate and will come to pass, unless there's some change in direction."

The last question

Her closing question is the one that ties the two halves of the interview together. Does he have a higher confidence level in his expectations for AI, the first part of the conversation, or the political part, the second?

Musk answers by explaining why the question resolves itself. Digital superintelligence "is called the singularity like a black hole, because you don't know what's going to happen after that. And it sort of sucks in everything. So in terms of the really big macro effects here, AI and robots will really dominate everything." On a timescale of less than ten years.

Minton Beddoes draws the conclusion out loud: which probably renders this other stuff we have been talking about less important and less relevant.

Musk agrees, and puts both of his forecasts on the same axis (1:24:35): "The AI robotics revolution will happen sooner. It'll happen sooner than civil war in Britain. I would say civil war in Britain is probably 20 years away. And the AI robot singularity is 10 years away."

"Well," she says, "hopefully these benign all powerful AIs will prevent such outcomes."

"They probably will."

She closes it: "Elon, I don't agree with everything you say, but it's always absolutely fascinating to talk to you. Thank you very much indeed."

"You're welcome. I'm not boring."

"You're definitely not boring."

Key takeaways

• The headline number is five years, not ten. Musk expects AI to exceed the sum of human intelligence in roughly five years from the 2026 recording, and to be far greater than it by 2036. He also thinks humans are unlikely to be in control at that point, using the chimpanzee comparison to explain why.
• His safety proposal is procedural, not regulatory. A biweekly call between the leading labs, plus one or two weeks of pre release access for competitors to test each other's frontier models via an API, on the model of the Motion Picture Association. Government gets involved only when a company refuses to address a danger its rivals have flagged. He says this has already happened once, with Amazon flagging Mythos to the White House.
• He thinks the binding constraint is electricity, not chips. Outside China, AI chips are being manufactured faster than new electricity comes online. Inside China, power is abundant and chips are the constraint, and he believes China is closer to solving lithography than most people realize. Orbital data centers are his answer, and they return the bottleneck to silicon.
• He accepts rapid job displacement and answers it with checks. AI is already better than 90 percent of professional software engineers, heading to 99 percent and then Stockfish level, and eventually everything. His policy answer is universal high income issued directly by the Treasury, with the claim that abundance makes deflation, not inflation, the problem.
• He concedes he overreached politically and concedes nothing on the substance. "I got a little too involved in politics. Got carried away, frankly." But on the USAID cuts his answer is zero deaths, repeated five times, with the argument that private foundations had the money to step in and are equally responsible if they did not.
• He accepts the influence and rejects the label. "I hope I am having a big impact" on European politics, while insisting his views are centrist and that safe cities, secure borders and sensible spending are not a right wing platform.
• The two halves of the interview do not have equal weight, by his own account. He puts civil war in Britain at roughly 20 years out and the AI and robot singularity at 10, which he says makes the political argument the less consequential one.

Chapters

• 0:00 AI will be smarter than humans in five years
• 12:10 How AI companies could regulate themselves
• 15:20 Will China be the leader in AI?
• 25:43 Elon Musk on Sam Altman and the AI bosses
• 31:20 "Work is going to be optional"
• 41:07 Should one man have so much power?
• 49:44 Starlink and the war in Ukraine
• 55:18 "I got too involved in politics"
• 57:14 Musk: zero people died because of USAID cuts
• 1:01:15 Europe, the far right and civil war in Britain
• 1:08:21 Musk: I'm not a racist
• 1:20:27 What do people get most wrong about Musk?

Notable quotes

"I think AI may exceed the sum of human intelligence in around five years. So there really won't be anything AI can't do better than humans apart from being human, perhaps." Elon Musk, 1:31

"The most likely outcome is an age of amazing abundance where anyone can have anything they can think of. This may sound preposterous, but well, here we are in 2026. Let's see where we stand in 2036." Elon Musk, 1:56

"I'll make another prediction. Money won't matter in 2036." Elon Musk, 4:43

"In the same way that if the difference in intelligence between AI and humans is vastly greater than the difference in intelligence between AI and chimpanzees, it's hard to imagine that the chimpanzees would be in charge." Elon Musk, 6:39

"Even if there was a stop button, we probably shouldn't press it, because the most likely outcome is incredible abundance for all." Elon Musk, 9:19

"If the heat death of the universe is in fact the outcome for our reality, then it really is all about the journey, because the destination is terrible." Elon Musk, 10:09

"All roads lead to acceleration of AI. So then I'm like, okay, well, you can just sort of be sad about it, or join the club, I suppose." Elon Musk, 11:38

"Realistically, Fable is still clearly the smartest model. But the Kimi model is getting quite close to Fable." Elon Musk, 15:45

"China has more electricity than the United States, Europe and India combined already, and my guess is it gets to four times the electricity production of the US." Elon Musk, 18:03

"It wasn't as though someone in the government figured out that Mythos had some scary cyber security risks. It was Amazon. Amazon highlighted this and Amazon called the White House." Elon Musk, 22:53

"If you started a nonprofit that was meant to be an open source AI company and owned by the world, and it somehow got turned into an $800 billion for profit company with closed source, I think you'd be like, well, wait a second, that's the exact opposite of what I donated the money for." Elon Musk on Sam Altman, 26:03

"The road to hell is, I think, mostly paved with bad intentions. There are a few well intentioned paving stones in there. So we don't want to be complacent." Elon Musk, 26:52

"At the end of the day, if we have to talk, we'll talk. Set aside our personal differences for the good of the world." Elon Musk, 27:04

"For software engineering, we already have a situation where AI is better than at least 90 percent of professional software engineers at writing software. It'll get to what I call Stockfish level." Elon Musk, 29:44

"I think work is going to be optional." Elon Musk, 31:20

"The word computer used to refer to a person doing computations. That used to be a job description of a person." Elon Musk, 33:33

"I think that the Treasury should just simply issue people checks." Elon Musk, 35:37

"Deflation will be the issue, not inflation." Elon Musk, 36:25

"If you ask me on any given day, in fact even intraday, I've gone from exhilaration to terror regarding AI." Elon Musk, 37:33

"In fact, I set a record for the most amount of taxes ever paid by a human." Elon Musk, 38:40

"I'm going to get all this pressure for, well, you missed your earnings this quarter because you spent too much on Mars. I'm like, yes, by the way, that was in the S-1 prospectus." Elon Musk, 43:32

"My life feels somewhat surreal. Like I can't believe these things are actually happening. It makes me believe in simulation theory." Elon Musk, 48:19

"I am quite offended by sort of fancy diplomats who have seven course dinners in luxurious places while people are dying at the front lines, and who pontificate about how Russia should withdraw." Elon Musk, 52:28

"I think I got a little too involved in politics. Got carried away, frankly." Elon Musk, 55:18

"It caused a lot of complaining." Elon Musk, on whether the DOGE cuts caused unnecessary suffering, 57:10

"Zero." Elon Musk, on how many people died as a result of the USAID cuts, 57:14

"I actually have a lot of sympathy for the change of the aid regime that the US under the Trump administration has precipitated. However, I think the pace and the suddenness with which it happened will almost certainly have caused unnecessary suffering and deaths." Zanny Minton Beddoes, 59:40

"We should have secure borders. We should have safe cities. We should have sensible spending. Which of those three are far right fringe? Please keep this part in." Elon Musk, 1:02:49

"London is a much safer city than any city in the United States. Violent crime in the UK is not where you think it is. In fact, it's coming down." Zanny Minton Beddoes, 1:03:46

"If you have a large and rapidly growing group of people whose beliefs are antithetical to western beliefs, at some point there will be a reckoning. And that's what you call civil war." Elon Musk, 1:04:22

"I think that it's a movie. And it's worth watching." Elon Musk, on promoting the film Citizen Vigilante, 1:05:52

"My partner is half Indian, and I have four children with her. One of them was named after a famous Indian physicist. So I would say I'm not racist." Elon Musk, 1:08:39

"I hope I am having a big impact." Elon Musk, on shaping European politics, 1:10:23

"They hate you far more than they hate me." Elon Musk, on journalists, 1:11:10

"There are more people that die in Europe from heat deaths than people that die from gun deaths in America. Why aren't you writing an article about air conditioning?" Elon Musk, 1:15:22

"I'm in favor of immigration. Obviously I'd be hypocritical if I wasn't, since I'm an immigrant to the United States." Elon Musk, 1:19:58

"Well, you just got a whole bunch of things wrong about me. Pretty much everything in the last, just rewind the last 20 minutes." Elon Musk, asked what people get most wrong about him, 1:20:35

"I'm very good at predicting the future. I'm not perfect, but I'm very good at predicting the future. My batting average is very high for a human." Elon Musk, 1:22:14

"I would say civil war in Britain is probably 20 years away. And the AI robot singularity is 10 years away." Elon Musk, 1:24:35

"You're welcome. I'm not boring." Elon Musk, closing the interview, 1:25:00

Resources mentioned

People

• Elon Musk, the interviewee
• Zanny Minton Beddoes, editor in chief of The Economist and the interviewer
• Demis Hassabis, whose private public regulator proposal Musk workshopped with him for a few hours before publication
• Sam Altman, the subject of Musk's OpenAI grievance
• Dario Amodei, "a very principled person," and the source of the 50 percent of entry level white collar jobs forecast
• Mark Zuckerberg, cited as another founder with voting control who "released one last week"
• Larry Page and Sergey Brin, cited as controlling Alphabet
• Steve Jobs, the key man risk analogy
• Neil deGrasse Tyson, host of the podcast where Musk made the "pet labradors" remark about a decade ago
• Ted Cruz, on whose podcast Musk gave the 10 to 20 percent figure
• Magnus Carlsen, the human benchmark in the Stockfish comparison
• Ray Kurzweil, whose predictions Musk calls "remarkably accurate"
• Iain Banks, author of the Culture novels
• John Stuart Mill, invoked by Musk as the classical liberal reference point
• Barack Obama and Hillary Clinton, the speeches in Musk's attribution swap trick
• Donald Trump, whose reelection Musk funded and whose administration he calls "on balance excellent"
• Rupert Lowe, named by Minton Beddoes as an example of who Musk supports
• Alice Weidel, whom Minton Beddoes has met and calls "a perfectly thoughtful person"
• George Soros, referenced in Musk's "Soros DA" criticism of US prosecutors
• MacKenzie Scott, cited by Musk as a foundation that could have filled the aid gap
• Bill and Melinda Gates, via the Gates Foundation's balance sheet

Companies, models and organizations

• Tesla, where the interview was recorded, at a Gigafactory
• SpaceX and its Starship, intended to launch more than once per hour
• Starlink, and the approved terminal white list built with the Ukrainian government
• Grok and xAI, the revenue line Minton Beddoes found in the SpaceX prospectus
• Anthropic, which Musk calls "the leader in AI," maker of Fable and Mythos
• OpenAI, the nonprofit Musk funded and the "$800 billion for profit company" he objects to
• Google DeepMind and Alphabet
• Meta
• Kimi K3 from Moonshot AI, "getting quite close to Fable"
• Amazon, which Musk says flagged the Mythos cyber security risk to the White House
• Stockfish, the chess engine that names Musk's endgame for software
• The Motion Picture Association, his model for industry self regulation
• USAID, and DOGE
• The Department of War, whose budget Musk compares to interest on the national debt
• The Gates Foundation
• The Economist, including its recent cover piece on air conditioning
• X, where Musk has close to 250 million followers
• The AfD
• The Alaska Permanent Fund and Trump accounts, raised as models for distributing equity to citizens

Works and ideas referenced

• The Culture novels, and Excession, which Minton Beddoes says she started reading because of Musk
• Star Wars, the first film Musk saw in a theater, at six, and Star Trek
• Citizen Vigilante, the film Musk promoted on X and Minton Beddoes calls irresponsible (title as spoken in the interview)
• The heat death of the universe
• The simulation hypothesis
• The future light cone of consciousness
• The technological singularity, and p(doom)
• The Cassandra effect, Musk's explanation for why he is not believed

Where it stands: what is checkable and what is not

Everything above is the interview rebuilt. This section is the separate question of what a reader can actually do with it.

The claims that are verifiable now. Several of Musk's statements are matters of public record rather than forecast, and can be checked directly. That China produces more electricity than the United States, Europe and India combined is a claim about published generation statistics. That interest on the US national debt now exceeds combined defense and intelligence spending is a claim about the federal budget. That US violent crime and murder rates exceed those of the UK, which Minton Beddoes asserts and Musk disputes, is a claim about two national crime series that are notoriously difficult to compare like for like, since the UK and US count and record offenses differently, which is part of why the exchange ends with her offering to send it to The Economist's fact checkers. Musk's tax percentages are his own description of his own return and cannot be independently checked from the interview.

The USAID exchange is the one place they are arguing about the same word differently. Musk's position is a strong claim, zero deaths, defended on two grounds: that the money was moved to the State Department rather than stopped, and that private foundations had ample capital to bridge any gap. Minton Beddoes' position is narrower than the one he answers: not that the policy change was wrong, which she says she has sympathy for, but that the speed of the change caused avoidable harm. His counter, that the foundations would be "equally responsible" if they failed to act quickly, does not actually dispute her claim, it reassigns the blame for it. Readers should note the two are not contradicting each other on the same proposition.

The forecasts are unfalsifiable for now, and Musk says so about himself. He claims a very high batting average, twice adds that he is not always right, and offers the Cassandra effect as the explanation for skepticism. It is worth holding that next to his own account earlier in the same interview of how his AI interventions went: he founded OpenAI as a counterweight to Google, Anthropic spun out of it, and the net effect, in his words, was to accelerate the thing he was trying to slow. That is a candid admission that a correct read of a trend does not guarantee a correct read of your own effect on it. His timing record is also mixed by his own hedge, "maybe not exactly in the timing that I thought."

The strongest and weakest parts of his AI case. The strongest is the constraint analysis, which is concrete, falsifiable, and does not depend on his politics: chips are being made faster than power comes online, China's bottleneck is the inverse, and export controls therefore buy less time than they appear to. The weakest is the macroeconomics of universal high income. Musk's argument that money creation is non inflationary if output grows faster is true as an identity about aggregate ratios, but it says nothing about who receives the new money, which sectors the output growth appears in, or what happens during the transition he himself calls a bumpy road. Minton Beddoes accepts the destination and never gets an answer on the road.

What the interview does not cover. There is no discussion of Neuralink, of Tesla's vehicle business or full self driving, of the xAI and X merger, of Optimus production numbers, or of any specific timeline for humanoid robots at scale, despite robots carrying most of the weight in his economic argument. The safety proposal is also left as a proposal: by his own account he has raised the biweekly call with Hassabis and had not yet raised it with Amodei at the time of recording.