There Are a Thousand Ways to Be Smart: Jeff Bezos on Amazon and the Road to Space

Jeff Bezos spent every summer from age four to sixteen on his grandfather’s working ranch in Texas. His mother had him at seventeen; the early summers, he says, were really his grandfather giving her a break, because a four-year-old is a burden, not a help. As he got older he actually became useful: fixing windmills, laying fences and pipelines, vaccinating animals, doing the jobs any rancher would do. After his grandmother died, when he was about twelve, it was just the two of them. His grandfather was “completely addicted” to the soap opera *Days of Our Lives*, and they would go back to the ranch house around 1 p.m. and lie on the floor to watch it.

The gift he took from those summers was resourcefulness. His grandfather made his own veterinary tools, heating a scrap of wire, pounding it thin, drilling a hole, and sharpening it into a needle to suture cattle. He bought a 1955 Caterpillar D6 bulldozer for about $5,000 because it was completely broken down; a new one would have cost more than $100,000. The two of them spent an entire summer repairing it, mail-ordering transmission gears too heavy to move and then building a crane to lift them. “He didn’t pick up the phone and call somebody,” Bezos says. “He would figure it out on his own.”

Bezos fell in love with space at five, watching Neil Armstrong walk on the Moon. Looking back at the 1957–1969 space race, he quotes Wernher von Braun: “I have come to use the word impossible with great caution.” Going to the Moon had been the stock analogy for something that would never happen — “you’ll do that when men walk on the moon” — until it did. He thinks Apollo was pulled forward in time by geopolitics and by spending that peaked at two or three percent of U.S. GDP. In that sense it is the twentieth-century version of building the pyramids, and it belongs in the pantheon of great human achievements.

Lex Fridman needles him for naming rockets after Americans rather than Yuri Gagarin. Bezos admits an American bias, then calls himself a big fan of Gagarin and of the first words he is purported to have said in space: “My God, it’s blue.” No one had seen Earth from outside. Alan Shepard’s suborbital flight — the one New Shepard is named after — was estimated by the Americans at about a 75 percent chance of success, a 25 percent chance of disaster. Shepard is less famous than John Glenn, the first American to orbit. Glenn sent Bezos a letter about a week before he died, grateful that New Glenn would carry his name, signed as a note “about New Glenn from the original Glenn.” Bezos has it framed on his office wall. “John, wherever you are, we got you covered.”

Asked what he hopes for hundreds or thousands of years out, Bezos does not hedge. He would love a trillion humans living in the solar system. At that population, he says, you would have a thousand Mozarts and a thousand Einsteins alive at any given time, and the solar system would be full of life, intelligence, and energy. Planetary surfaces are too small for that vision unless you turn the planets themselves into stations. The path he describes is O’Neill-style colonies, spun up for Earth-normal gravity, built from lunar material, near-Earth objects, and the asteroid belt. Most people, especially early on, will want to live near Earth so they can go back — the way you might visit Yellowstone — rather than give the planet up altogether.

He does not see spaceflight as a rival to taking care of Earth. Robotic probes have already visited the other planets; this one is “the good one.” Literacy, poverty, and infant mortality have moved in the right direction for decades, and antibiotics and modern medicine make life better than it was for our grandparents. The thing moving backwards is the natural world. Five hundred years ago, pre-industrial, it was pristine; we traded some of that beauty for the gifts of an advanced society. He thinks we can have both, but only by going to space, because rising energy use per capita — the metric he treats as most fundamental — is not ultimately compatible with a finite planet. “You could argue about when you have to do that, but you can’t credibly argue about whether.”

Blue Ring is the piece of that infrastructure he rarely discusses in public. The spacecraft is designed to take up to 3,000 kilograms of payload to geosynchronous orbit or lunar vicinity, with both chemical and electric propulsion. A customer can crawl to GEO over 100 to 150 days on electric power and save the chemical propellant for fast orbit changes, or reverse the order. What he is really selling is a set of services — thermal management, electric power, compute, communications — so payload teams do not have to solve radiation-tolerant compute themselves. He analogizes it to a set of APIs, “a little bit like Amazon Web Services, but for space payloads.” Blue Ring is not for people; crew is the job of the lunar lander.

Bezos went to Princeton intending to become a theoretical physicist; computer science was something he did for fun. In quantum mechanics he watched classmates do effortlessly what was brutally hard for him, and he decided he was going to be a mediocre theorist. The person who settled it, without meaning to, was a student from Sri Lanka named Yosanta — “one of the most brilliant people I’d ever met,” and also the most humble. One night Bezos and his friend Joe spent three hours on a partial-differential-equations problem and got nowhere. They looked at each other, said “Yosanta,” and walked to his dorm.

Yosanta stared at the problem for about ten seconds and said “cosine.” Bezos thought he was kidding. Yosanta wrote three pages of cancelling terms; the answer was cosine. Had he done it in his head? “Oh no, that would be impossible. A few years ago I solved a similar problem and I could map this problem onto that problem.” Bezos took the lesson: maybe theoretical physics is not what the universe wants you to be. He switched to computer science. Walter Isaacson has compared his creativity to Einstein’s; Bezos’s own label is blunter. “If you want to boil down what I am, I’m really an inventor.” He generates a hundred atypical solutions, ninety-nine of which die under scrutiny, and hunts for the one that might work.

He cannot explain how the search works, only that it requires wandering. Efficiency and real invention, not incremental improvement, are at odds: if you already knew how long the meeting would take, you would already have a straight line to the answer. When an idea survives his first private filter, he warns the next person that objections will be easy and asks them to work with him anyway — “there’s something there.” He still looks for the way each person is smart rather than measuring everyone on one axis. “There are a thousand ways to be smart.”

New Glenn is a heavy-lift vehicle, about 45 metric tons to low Earth orbit, with a little more than half the thrust of Saturn V — about 3.9 million pounds at liftoff. The booster flies seven BE-4 engines, each a little more than 550,000 pounds of thrust, burning liquefied natural gas and liquid oxygen in an oxygen-rich staged-combustion cycle that Bezos says the Russians pioneered. The same engine will power the first stage of United Launch Alliance’s Vulcan. The upper stage uses two BE-3U engines, a hydrogen variant of New Shepard’s BE-3, at 160,000 pounds each. Hydrogen’s high specific impulse is a gift on an upper stage and a liability on a booster: the tanks get geometrically huge. Delta IV did all-hydrogen, including the booster; it was operationally capable, he says, and never very cost-effective.

“Rockets love to be big.” Guidance and control weigh about the same on a giant vehicle as on a tiny one, so that parasitic mass is fatal when you are small and trivial when you are large. Turbo pumps, like all rotating machines, get more efficient as they grow: the gap between housing and impeller is an absolute width, and on a small pump it is a large fraction of the flow. The giant exception is manufacturing. Small engines can be assembled on a table by one person; large ones need cranes, fixtures, and civil works. Cape Canaveral pads sit on beaches because you launch over water, which means driving dozens of pilings 50 to 150 feet deep so the sand can hold the monument.

He plays a game with other rocket people: what would amaze a 1960s engineer? Most of the greatest hits have not changed. Carbon composites have. Blue Origin’s tape-laying machine builds fairings that are lighter and stiffer than metallic structures of that era. Aluminum-lithium plus friction stir welding is the other shock. A rotating pin stirs two plates together at a precise speed; the joint is barely warm to the touch, and the weld is as strong as the parent metal, so you do not have to thicken the tank at the seam. The fairing’s job is almost perverse: 100 percent structural integrity until the moment it must have zero. The robust way to get there, he says, is a little explosive.

New Glenn’s first stage is reusable and lands downrange on an ocean platform; the hydrogen upper stage is expendable. They know how to make the second stage reusable, Bezos says, and they also know how to make it cheap enough to throw away. Which path wins on cost is not obvious. Then he shrugs in a way Fridman calls infuriating: getting into orbit is a solved problem from the 1950s and 1960s. The only interesting problem is dramatically reducing the cost of access, so that startups can do new things in space. Inventing a better way — the plow is his example — is how the world gets richer.

The first launch is scheduled for 2024. He is optimistic it will happen that year; he is not sure whether NASA’s ESCAPADE Mars probes will ride it or something else will. The larger challenge is the factory. Twenty-four New Glenn flights a year means an expendable upper stage every two weeks and, at two BE-3U engines per stage, an engine every week — machine tools, fixtures, inspection, acceptance tests, and people all running at that cadence. “It’s one thing to build a first article,” he says. “The hard part is everything that’s going on behind the scenes to build a factory that can produce New Glenns at rate.” Do it inefficiently and you have not solved cost, and you have not moved the state of the art. If you just want to make money, he jokes, start a salty-snack company. Fridman says he is writing that down. Bezos suggests Lex Fridman potato chips.

When he started Amazon in 1994, credit cards, the postal service, and dial-up piggybacking on the long-distance phone network already existed. Without that heavy infrastructure, the capital expense would have been hundreds of billions of dollars; no startup could have done it. Two kids in a dorm room could start an internet company because the road was already there. He wants to take Amazon winnings and build the equivalent for space, so his children’s generation can stand on it. The marker of success is specific: a really valuable space company started in a dorm room. An inventor’s greatest dream, he adds, is that the invention becomes so successful people take it for granted. Nobody thinks of customer reviews or 1-Click as inventions anymore. That is a compliment.

Is he nervous about the first flight? “Are you kidding? I’m extremely nervous about it.” He is nervous at every New Shepard launch he attends. To feel nothing on a first flight, he says, would be “some sign of derangement.” Ground testing and simulation have killed many problems; some you can only find in flight. “Cross your fingers. I guarantee you, you’ll have fun watching it no matter what happens.”

Strangely, sitting inside New Shepard on its first crewed flight was less nerve-wracking than watching other people ride. Telling his mother was the hard conversation, and he was taking his brother too; her reaction was “both of you?” The most emotionally powerful moment came before the flight. At 4:30 a.m., Lauren was going to fly them to the West Texas launch site by helicopter. They walked out of the ranch house and found their kids, their brother’s kids, both sets of parents, and close friends assembled outside. The family was saying goodbye in a way that looked, from their faces, like forever. It felt, Bezos says, almost like attending your own memorial service: you could feel how loved you were.

He went because he knew the vehicle and the people who built it, and because of the escape system. A solid rocket motor in the base of the crew capsule can push the capsule off the booster if anything goes wrong on ascent. Blue Origin put as much effort into that system as into the rest of the vehicle combined; it is why he is willing to let anyone fly. Historically, human-rated rockets had escape systems — Gemini, Apollo — and only the Space Shuttle did not. Most designs use a tower that must be ejected even on a nominal flight. New Shepard’s is a pusher, reusable if unused, because cost matters. The trigger is fully automated; it can react faster than a person. His private test was simple: if he was not ready to go, he would not want anyone to go.

Zero gravity felt “very natural.” The overview effect hit hard. He quotes Jim Lovell looking back at Earth: you do not go to heaven when you die; you go to heaven when you are born. Everything else is blackness. There is one gem of life, and it is Earth. Fridman asks whether part of him is frustrated by Blue Origin’s pace. “Blue Origin needs to be much faster,” Bezos says, and that is a primary reason he left the Amazon CEO job. As CEO of a public company he believed it had to get his full attention. He is still executive chair; he handed over the CEO role so he could put urgency into Blue Origin.

At Amazon he wanted the company to become the world’s most customer-obsessed, in any industry. The analogous ambition at Blue Origin is to become the world’s most decisive company — technically ambitious, quick to take appropriate risk, willing to study five approaches fast and then pick. He returns to a distinction he has taught for years. Most decisions are two-way doors: walk through, spend a little time, and if you are wrong you come back and pick another door. Those should be made by a single person or a tiny team deep in the organization. A few decisions are one-way doors, consequential and hard to reverse. Those should go to senior people who slow down. As Amazon CEO he often found himself playing “chief slowdown officer,” asking for three more analyses because the choice would be costly to unwind. The common corporate failure is a one-size-fits-all process that puts the heavyweight ritual on two-way doors too.

Some technical choices are quick-drying cement. Selecting LNG for New Glenn’s booster and hydrogen for the upper stage has turned out well; changing it would be a huge setback, so you scrutinize it. Amazon Prime is the kind of business bet that is painful to unwind. For everything else, speed plus the option to change your mind is the point. “Disagree and commit” is how he stops the arguing. A lieutenant wants to do it one way; Bezos thinks it is a bad idea; they go back and forth; then he often says he does not think they are right, but he will gamble with them because they are closer to the ground truth. After that he will not snipe, will not say I told you so, and will actively help it work.

He names two resolution mechanisms he considers terrible. Compromise: Fridman says the ceiling is twelve feet, Bezos says eleven, they split the difference at eleven and a half instead of fetching a tape measure. Compromise is low-energy and does not lead to truth; when the truth is measurable, you should measure. The other is stubbornness — a war of attrition between executives until one is exhausted. That does not find truth either, and it is demoralizing. Escalate instead. Amazon has a million and a half people and is still fast, he says, because the culture supports high-velocity decisions at every scale.

Blue Origin’s MK1 lander is designed to put 3,000 kilograms of cargo on the lunar surface and stay there. It is expendable, hydrogen-fueled, and sized to fly on a single New Glenn, which keeps the architecture simple. MK2 is the NASA Human Landing System vehicle, part of what the agency calls the sustaining lander program, and it is meant to be reusable. Apollo’s lander was effectively two stages: the descent stage remained on the surface and only the ascent stage returned to lunar orbit. MK2 is a single stage that carries enough propellant to bring the whole vehicle back up, over and over, to drive down cost. Artemis, unlike Apollo’s six landings and a program that became too expensive to continue, is supposed to go back and stay.

Hydrogen’s disadvantage as a deep cryogen is boil-off. As part of the lunar work they are developing solar-powered cryocoolers to make hydrogen a storable propellant for the Moon, Mars, and the outer planets. On the surface they have already built a solar cell entirely from lunar-regolith simulant. It is only about 7 percent efficient, far below terrestrial cells, but a factory that can land and then feed on the dirt underfoot would keep pouring out power. They are also working on extracting oxygen from the oxides in the soil, which is energy-intensive and pairs with those cells. Permanently shadowed polar craters hold water ice that electrolysis could split into oxygen and hydrogen — a high-efficiency propellant. All of that still has to pass one gate: landing cargo and humans at an acceptable cost.

Will he walk on the Moon or Mars? Unlikely in his lifetime, he thinks; those missions will probably stay with professional astronauts, though he asks Fridman not to count him out yet. Factories on the Moon may not need human tenders. On Blue Origin and SpaceX, he does not want a two-horse race. The internet produced giants, mid-sized firms, and small profitable companies at every scale; space is big enough for the same dynamism. He does not know Elon Musk well and refuses to judge anyone by a public persona, but “judging by the results, he must be a very capable leader.” Tesla and SpaceX do not happen otherwise. Fridman hopes they will hang out. Bezos says they are very like-minded, if not identical, and he loves the idea.

An early home video, his father holding the camera, shows Bezos explaining the genius of an orange extension cord that powers a desk and a CRT. He had left a good New York job anxious and excited, telling early investors he put the odds of getting their money back at 30 percent — not the outcome that actually happened. Startup base rates are grim, and you still have to be 100 percent sure it will work; you hold both thoughts at once. From 1994 to the present he has found Amazon a joy and a privilege, which is not the same as comfort. In 1997 he started writing about Day One in the shareholder letters. In his last letter as CEO he defined Day Two: “stasis followed by irrelevance, followed by excruciating painful decline, followed by death.” That is why it is always Day One.

Day One is renewal. Every day you decide what to do, untrapped by who you were, even by self-consistency. Amazon program teams write tenets, more tactical than principles, and in parentheses they add “unless you know a better way,” so dogma and history cannot lock the door. His starter pack for fending off Day Two is customer obsession, a skeptical view of proxies, eager adoption of external trends, and high-velocity decision making. Proxies are the subtle failure. Someone invents “customer contacts per unit sold” because they understand its link to happiness; five years later the metric is treated as the thing itself, the world has shifted, and people are managing a number they no longer understand. You do need metrics. You also have to remember that you do not care about the metric. You care about the customer.

Fridman asks what it takes to be the person who says the proxy is stale. Bezos generalizes: truth-telling. Humans are not really truth-seeking animals; we are social animals. Ten thousand years ago, going along got you through the village; the truth-teller might get clubbed at midnight. Important truths are uncomfortable, awkward, exhausting, and they make people defensive even when that is not the intent. High-performing organizations need mechanisms that support telling them anyway. One small one: in every meeting he attends, he speaks last. If he speaks first, even strong-willed people wonder whether they were wrong. Ideally the most junior person goes first, then up the seniority ladder, so you hear unfiltered judgment. Junior people should be able to overrule seniors when they have the data.

He has a saying: when the data and the anecdotes disagree, the anecdotes are usually right. That does not mean you slavishly follow stories; it means you inspect the measurement, because you are usually measuring the wrong thing. Early Amazon had a metric showing 1-800 customer-service wait times under sixty seconds, and a pile of complaints that it was longer. In a weekly business review he picked up the phone, dialed the number, and the room sat in silence. It was more than ten minutes. The data collection was wrong. The moment was uncomfortable on purpose. Truth has to get people’s attention, win their buy-in, and energize the fix.

Amazon internally calls tiny experience defects “paper cuts.” The big things still get most of the energy, as they should. His advice to anyone running a business is to find what will not change in ten years. In retail he is sure customers will still want low prices, fast delivery, and big selection; no one will say they love Amazon but wish the prices were higher. Those stable goods are worth sustained investment. Even the obvious big things are astonishingly hard to focus on. Paper cuts never make it down the list of the teams doing that work, so Amazon staffs dedicated teams to fix them. 1-Click, he says, probably eliminated a whole series of them at once. Fridman still gets a surge of happiness pressing the button. Bezos’s reply is almost aesthetic: in the perfect invention, at the perfect moment, in the perfect context, “there is real beauty,” and the inventor, the team, and the customer can all feel it.

They co-evolve with tools. He still remembers closing his eyes after Tetris and seeing the blocks rotate; the brain had been rewired. Phones, he fears, are attention-shortening devices because most of what we do on them is short-form, and convenience makes us do more of it. He is not even sure we know that is bad, but he thinks it is happening, which is why long-attention work — books, Fridman’s long conversations — still matters. Then the conversation turns to generative AI. Large language models in their current form, Bezos says, are not inventions. They are discoveries. The telescope was an invention; seeing that Jupiter had moons was a discovery. A 787 is an engineered object whose behavior we know and do not want to be surprised by. Models keep surprising us.

Even specialized systems, not general AI, could do real damage as weapons. He is still optimistic: these tools are “much more likely to help us and save us” than to hurt and destroy us. Humans already have many ways to go extinct; the models might help us not. People who are, in his view, overly concerned may be missing how useful the tools will be. His favorite stretch of *Oppenheimer* is the one some viewers found boring: Robert Downey Jr.’s bureaucrat, petty and vengeful over an insult that, in the last scene, never happened. We invented nuclear weapons, Bezos says, and we are not really mature enough to wield them. Before anyone gets to machine agency, the nearer-term yield is better medicines and better tools. He is excited about the next ten and twenty years.

The human brain does remarkable work on about 20 watts; today’s techniques burn many kilowatts on equivalent tasks. A sixteen-year-old learns to drive on far fewer miles than a stack that needs billions. There are tricks left to learn, and it is not “just” scaling — he puts *just* in quotes, because scaling is itself hard — even though scaling also pays huge dividends. Models are very good at sounding true and often ungrounded; with thin data they concoct accurate-sounding narratives. “They need to be taught to say I don’t know more often,” he says, “and I know of several humans who could be taught that as well.” On the product side, Alexa’s hundreds of millions of devices are about to get a lot smarter. AWS is building Titan, its own foundation model, and Bedrock, so enterprises can use powerful models with their own data without accidentally contributing that data back. Will people fall in love with Alexa, as in *Her*? Given the spectrum of human sexual variety, “the answer to your question has to be yes.” He does not know how widespread it will be. It will happen.

A productive day does not start at full speed. “I get up in the morning and I putter.” He drinks coffee, reads his phone and the newspaper, chats with Lauren, and moves slowly for the first couple of hours; he wakes early by nature. Then he exercises most days. On the hard days the motive that actually works is near-term: he will feel better in about four hours, with more energy for the rest of the day. Wanting to be mobile at eighty is real and too far away to pull him through a grim morning. A good session is about thirty minutes of cardio and forty-five of weights, rotating push, pull, and legs. A trainer he likes will look at him when he says he cannot add weight and say, “Yeah, I think you can.” Of course the trainer is right. Since leaving Amazon’s CEO role he has never worked harder, mostly on Blue Origin, including architecture and technology meetings on vehicles and engines. The technology is his favorite part; building organizations is his least. “That’s why they call it work.” They hired Dave Limp, whom Bezos has known for close to fifteen years, as CEO.

He does not keep a strict schedule, because he believes in wandering. The perfect meeting starts with a crisp document — “written with such clarity that it’s like angels singing from on high” — and then a messy discussion of questions nobody knows the answer to. Amazon and Blue Origin meetings often open with a six-page narratively structured memo and thirty minutes of silent “study hall,” people reading together and writing in the margins. Pre-reads fail: busy people skim or bluff, like undergraduates pretending they did the assignment. PowerPoint is a sales tool; internally you want truth-seeking. Slides are easy for the author and hard for the audience; a good memo may take two weeks to write, rewrite, and have holes poked in it. Executives interrupt slideshows with questions answered on the next slide; reading to the end, Bezos often crosses out every margin question because the memo already answered them. The author has to go first and be vulnerable. Sitting there while the room reads your work is “terrifying in a productive way.” Complete sentences with topic sentences and narrative structure make sloppy thinking hard to hide. Weekly business reviews, by contrast, can start and end on time: same metrics, incremental improvement.

The 10,000-Year Clock is a physical clock about 500 feet tall, inside a twelve-foot-diameter chamber carved into a West Texas mountain. Danny Hillis conceived it in the 1980s as a symbol of long-term thinking: it ticks once a year, chimes every hundred years, and the cuckoo comes out every thousand. It is fully mechanical, designed to last 10,000 years with no human intervention, remote both for protection and so visitors have to make a pilgrimage. Over centuries it is supposed to take on the patina of age and lengthen humanity’s horizon. Bezos thinks that matters because we now affect the planet and one another, with weapons of mass destruction and with side effects like carbon in the atmosphere — an unintended consequence of an industrial revolution that also delivered huge benefits. “Long-term thinking is a giant lever.” Five years is already a stretch for most institutions; we probably need ten, fifteen, twenty, twenty-five, for the sake of children and grandchildren. Will humans still be here when the clock runs out? He thinks so. The United States will not. No nation-state has lasted anything close to 10,000 years, and accelerating technology makes any particular form even less likely.

On mortality he is blunt. As a young person he was scared of death and did not want to think about it. At fifty-nine that fear has mostly gone. He would like to stay alive as long as possible, but he is more focused on healthspan. “I want that square wave. I want to be healthy, healthy, healthy, and then gone. I don’t want the long decay.” Curiosity keeps him around — he wants to see how things turn out — along with family and close friends. Mortality no longer has the grip it had in his twenties. Fridman closes with Bezos’s own line: be stubborn on vision, but flexible on the details.