When Jony Ive designed the first iMac, he put a handle on it. Not so people could carry it, it was a desktop computer. Nobody was going to carry it anywhere. Steve Jobs insisted on it anyway. Manufacturing a recessed handle cost a fortune. Every engineer at the old Apple would have killed it. Ive explained why it mattered: "Back then, people weren't comfortable with technology. If you're scared of something, you won't touch it. If there's a handle on it, it makes a relationship possible. It gives you permission to touch." A handle changed how people felt about approaching a computer. Elon thinks the same way. Early Tesla engineers would describe him stopping meetings over a panel gap, a surface ratio, a curve that didn't feel right... Things that had nothing to do with range or performance. He'd just say it's not beautiful enough and refuse to move on. Most people assume Jobs and Musk are opposites, one sold dreams, the other sells engineering. The overlap is that both understood something most builders never figure out: "The way a product makes you feel when you look at it is part of how it works."

Patrick Collison (@stripe CEO) on How to Move Fast:

First Principles Thinking: How Musk and Bezos Rebuilt Rockets and Retail... In October 2001, Elon Musk landed in Moscow to buy a missile. He brought Jim Cantrell, an aerospace fixer who knew the Russian industry, and his college friend Adeo Ressi. The plan was to send a small greenhouse to Mars, and a refurbished Russian ICBM looked like the cheap way to get it off the ground. The first trip ended with nothing. According to Cantrell, one of the Russian chief designers treated Musk as a novice and spat on him. Musk went back in February 2002 with Mike Griffin, who later ran NASA. At the meeting with ISC Kosmotras, the Russians quoted $8 million per rocket. Musk offered $8 million for two. The offer went nowhere. On the flight home he opened a spreadsheet. Accounts differ on which trip it followed, but not on what he did with it. He priced what a rocket is made of: aluminum, carbon fiber, propellant. By his calculation the raw materials came to about 2 percent of what a finished rocket sold for. Early investor Steve Jurvetson remembers the figure as 3 percent. The Russians had answered the question he asked them, which was how much they would charge. The spreadsheet answered a different one: what a rocket is made of, and what that costs. Four months after the second Moscow trip, he founded SpaceX. Two ways to solve a problem: 1. Reasoning by analogy starts with what exists and improves it. Rockets cost $65 million, so you raise $65 million. Competitors charge $99 a month, so you charge $89. 2. Reasoning from first principles discards what exists and works out what is true. What is a rocket made of? What do those materials cost? Why is there a gap between that number and the price? Aristotle defined a first principle as the first basis from which a thing is known. That was about 2,300 years ago, and most people have heard of the idea since. Few use it, and the reason is less about the idea than about how minds work. Why your mind resists it: A brain is built to match patterns. This looks like that, so do what worked last time. The habit is efficient and it keeps you functional, but it carries a cost: reasoning by analogy means inheriting every assumption inside the analogy, and those assumptions feel like facts. “College is expensive” feels like a fact. The first-principles version asks what education is made of, whether that’s access to information, feedback on your work, or a credential that signals competence, and whether each can be obtained differently now. “That’s how this industry works” invites you to adopt someone else’s constraints without examining them. Musk’s Moscow trip was a case in point. The price of a rocket was treated as a given until he priced the parts. What Bezos did with it: On June 9, 2004, Bezos emailed Amazon’s senior leadership team with the subject line “No powerpoint presentations from now on at s-team.” In their place he required a six-page memo written as a narrative, in full sentences. Meetings began with everyone, Bezos included, reading the memo in silence. By his own account in 2018 that took about the first thirty minutes. He explained the silence plainly: executives will bluff their way through the meeting as if they’ve read the memo, because they’re busy, so the time to read it has to be carved out of the meeting itself. Bezos also said that slides carry obscure information, while a memo forces “verbs and sentences and topic sentences and complete paragraphs.” A slide lets you gesture at a problem. A page of prose requires you to say what the problem is, what you know, what you are assuming, and how the solution follows from it. The format makes you separate what you know from what you are guessing. --- A worked example: the battery question In a 2012 interview with Kevin Rose, Musk walked through how he applied this at Tesla. Step 1: State the problem with no solution inside it. Electric cars were expensive largely because the battery pack was expensive. The accepted price was around $600 per kilowatt-hour, and the accepted explanation was that this is what batteries have always cost. That is reasoning by analogy: the price of a battery is the price of batteries. Step 2: List what is verifiably true. A battery pack is made of specific materials: cobalt, nickel, aluminum, carbon, polymers that separate the cells, and a steel can. Each of those trades on commodity markets at a known price. Priced at London Metal Exchange rates, Musk said, the materials came to about $80 per kilowatt-hour. Step 3: Rebuild from that list. If the parts cost $80 and the pack sells for $600, most of the other $520 is what it costs to turn materials into a pack: manufacturing, supply chain, volume. A company can work on every one of those. A price that is “just the way it is” gives it nothing to work on. Musk’s conclusion was that you need clever ways to combine those materials into a cell, and then batteries can be much cheaper than anyone realizes. The $80 was a floor on materials, and no finished pack was ever going to cost that. But the $600 turned out to be movable. Industry surveys put the average pack price at roughly $139 per kilowatt-hour in 2023, under a quarter of the 2012 figure. --- Run it on your own company: You can do the same exercise on the cost your startup stopped questioning: cost per customer served, cost per order delivered, cost to build the product. Write the cost as a single number you pay today. List the raw inputs, meaning what must actually happen or be bought for that unit to exist. Price each input on its own. Compare the total to what you pay today. Take an illustrative case, a delivery startup that pays $9 per order. The raw inputs are a driver’s time, a vehicle, and distance. Price those three at their own cost. The difference between that sum and $9 is dispatch, idle time, failed deliveries, and rework. That difference is where the work is, and it is a list of specific problems to attack instead of an industry norm to accept. The method gives you a gap, and sometimes the gap turns out to be real.

The Facebook board met on a Monday morning in July 2006. There were three people on it: Mark Zuckerberg, Peter Thiel, and Jim Breyer of Accel. Yahoo had offered $1 billion for the company. Facebook was two years old, its revenue was around $30 million, and it wasn’t profitable. Zuckerberg was 22 and owned roughly a quarter of it. Thiel and Breyer both thought they should take it. Zuckerberg opened the meeting by telling them it was a formality. It would take ten minutes, he said, because they were obviously not going to sell. Thiel pushed back. A billion dollars is a lot of money, he told him, and Zuckerberg owned a quarter of the company. There was a great deal he could do with that. Zuckerberg barely entertained it, and the meeting ended with a no. Thiel went along because of the rule his firm runs on, which is to always back the founder. He told SXSW years later that it made him a little worried. Almost immediately, stories appeared questioning the decision. Zuckerberg had a reason he didn’t spell out for them that morning. Facebook was about to launch News Feed, and he figured that if it worked, the company would be worth far more than a billion dollars. News of the rejection wasn’t the end of Yahoo’s interest. Later that year it came back, and the talks stretched into a weekend. In 2013 Michael Arrington reported that he had seen an internal Yahoo analysis from late 2006. According to him, Yahoo was prepared to pay at least 60 percent more than the billion it had formally offered, around $1.6 billion. In 2007 Microsoft paid $240 million for 1.6 percent of Facebook, which valued the whole company at $15 billion. Thiel has called that July morning the most important moment in Facebook’s history.

By 2004, Steve Kessel’s job was keeping publishers calm. He ran Amazon’s physical media business, and much of his time went to reassuring the people who made books that it was fine for used copies to sit next to new ones on the same page. Books were what Amazon had been built on, and Kessel was the person defending them. Then Bezos took that away from him. He had been reading Clayton Christensen’s The Innovator’s Dilemma, which is about how big companies die protecting the business that made them. Sears had missed discount retailing. Bezos had a suspicion about where the threat to Amazon would come from: a device, from Apple or someone else, aimed at the one thing Amazon sold best. So he moved Kessel off physical books and put him on digital, with an order to proceed as if the goal was to put everyone selling physical books out of a job. His reasoning was that someone running both businesses would never go after digital with enough tenacity. When Kessel asked how long he had to build Amazon’s first piece of hardware, Bezos told him he was basically already late. Amazon was a retailer and had never built a device. Kessel flew to Silicon Valley and sat down with hardware people from Apple and Palm and with executives at IDEO, trying to learn what it would take. That fall he hired Gregg Zehr, who had been a vice president of hardware engineering at Palm and before that spent nine years at Apple, where he ran PowerBook engineering. Zehr’s small team started out in shared space in a Palo Alto law library. They called the lab Lab126, with the 1 standing for A and the 26 for Z, the arrow in Amazon’s logo. They called the device Fiona, after a character in Neal Stephenson’s The Diamond Age. The project was kept secret from most of the company. The hardware turned out to be the easier problem. Bezos had decided the store needed 100,000 books on day one. Publishers had converted around 20,000 into digital form, and Amazon couldn’t explain why it needed the rest, because the device didn’t officially exist. J eff Steele, who worked on the effort, later described his job as dragging publishers kicking and screaming into the twenty-first century. Bezos also made a gut call on price. The most popular books would cost $9.99, a number modeled on Apple’s 99-cent song, and he kept it from publishers all the way through the negotiations. He insisted on a keyboard too, inspired by the BlackBerry, so readers could search for titles. The launch was reportedly aimed at Christmas 2006. It slipped a year while the catalog filled up. On November 19, 2007, Amazon put the Kindle on sale for $399 with roughly 90,000 books. According to Amazon’s own account, the team watched the announcement live, waiting to see what would happen. It sold out in five and a half hours, and stayed out of stock until late April. In his letter to shareholders, Bezos wrote that after three years of work they had hoped it would be well received, but had not expected the level of demand that showed up. Kessel had been hired to protect the bookstore. He ended up building the thing that competed with it.

In 2014, Whitney Wolfe Herd was 25 years old, fresh off a sexual harassment lawsuit against Tinder, a company she had helped build from the beginning. She wanted to quit tech entirely. Instead she built Bumble in Austin with $10 million from a Badoo investor, a 20% stake, and one rule nobody had tried before: on heterosexual matches, women had to message first. If they didn’t within 24 hours, the match disappeared. It was a product decision built directly from personal experience. In 2021 she took Bumble public at a $13 billion valuation, becoming the youngest female founder to ever ring the Nasdaq bell. Today Bumble was acquired.

In November 1982, a headhunter called John Sculley’s office at Pepsi. Sculley was 43, the youngest president in Pepsi’s history, and widely expected to become its next chairman. He had spent his entire career there. He had built the Pepsi Challenge, the blind taste test campaign that had turned a perpetual second place brand into a genuine threat to Coca-Cola. He turned down headhunters constantly. He took the call anyway. The client was a computer company in California called Apple. Their founder wanted to meet him. Sculley flew out, mostly to get the headhunter off his back. He arrived at Apple’s headquarters and noticed he was the only person in the building wearing a suit. The employees dressed less formally than the maintenance staff at Pepsi. The 27-year-old who greeted him had jet black hair and dark, intense eyes. He walked Sculley through what he was building and why it mattered. Sculley was impressed. Jobs was fascinated by how Pepsi marketed itself. At the end of the meeting, Sculley said he wasn’t interested. Jobs flew to New York a few weeks later and showed up at his office. Then he started calling every three or four days. He flew back again. Sculley kept saying no. This went on for five months. Every weekend. Phone calls. Flights. Lunches. Jobs showed Sculley parts of the company nobody outside saw. He made the case over and over in different ways. In March 1983, on a Sunday in New York, Sculley told Jobs for the last time: he had thought about it, and he wasn’t coming to Apple. Jobs paused. Then he stepped forward until he was about eighteen inches away. “Do you want to spend the rest of your life selling sugared water, or do you want a chance to change the world?” Sculley said it hit him like a fist in the stomach. He couldn’t sleep that night. The line wouldn’t leave him alone. He called Jobs the next morning and said yes. Sculley left a $500,000 salary, a near-certain path to the chairmanship, and a company he had devoted his life to. Apple paid him $1 million a year to come. For eighteen months it worked. Jobs focused on product. Sculley focused on the business. Apple was growing. Then they started clashing. Jobs wanted to cut the price of the Mac to drive volume. Sculley thought the margins mattered more. The board sided with Sculley. In May 1985, Sculley stripped Jobs of his role running the Macintosh division. Jobs tried to organize a coup while Sculley was traveling. Sculley found out, called an emergency board meeting, and forced a vote. The board sided with Sculley again. Jobs resigned from Apple in September 1985. The man who had spent five months flying across the country, calling every three or four days, standing eighteen inches away to deliver the most famous recruitment pitch in business history... that man had just been removed from his own company by the person he recruited. Jobs started NeXT the following week. Twelve years later, Apple bought NeXT and brought him back. Sculley was long gone by then.

In 1971, Steve Wozniak’s mother mailed him a magazine article. It was from Esquire. The headline: “Secrets of the Little Blue Box.” Fourteen pages about a group of underground hackers called phone phreaks who had figured out how to hack AT&T’s entire phone network using nothing but audio tones. The most famous of them was a man named John Draper, Captain Crunch, who had discovered that a toy whistle found in a Cap’n Crunch cereal box produced exactly the 2600-hertz tone AT&T used to control its long-distance lines. One whistle. Free calls anywhere on earth. Wozniak read the article and called his friend Steve Jobs. Jobs was 16. Wozniak was 21. They both assumed it was fiction. The technical detail was too precise, too specific. Nobody could actually do this. So they drove to the Stanford Linear Accelerator library and looked up the frequencies themselves in an AT&T technical manual. It was real. Wozniak built the device. Jobs sold it for $150 each, mostly to college students and people in dorms. They made $6,000 total... their first product, their first partnership, their first taste of what it felt like to go up against a corporation and win. Then Wozniak called the Vatican from a payphone, pretended to be Henry Kissinger, and asked to speak to the Pope. Just to see if it worked. There was one night a police officer spotted them in a parking lot with the device and asked what it was. Wozniak’s hands were shaking. Jobs slid the blue box into his pocket and kept talking like nothing had happened. Jobs said later: “If it hadn’t been for the blue boxes, there would have been no Apple. Woz and I learned that we could take on large companies and beat them.” The company that eventually put a phone in every pocket on earth started with two teenagers and a toy whistle from a cereal box.

Grace Hopper spent years being told that computers could only do arithmetic. Her argument was that programmers shouldn’t have to write in the machine’s notation. Let them write something closer to English and have the machine translate it. The field thought this was naive, or a waste of cycles, or both. She built the compiler anyway. Every interface since has followed the same arc. It starts in the machine’s language because that’s all the machine can take, and it ends in ours once the machine gets good enough to meet us. Voice AI is still in the first half. You direct a performance by typing [whispers] or wrapping it in SSML, then guessing which tag produces the delivery you heard in your head. Fish Audio’s Drama 3 takes the note instead. “Say it like you’ve given up on life.” It’ll switch voices mid-sentence, voice a whole scene from one script, and change a single word without regenerating the take around it. Markup was never how anyone describes a performance. It was just the only thing the model could read.

I think about this clip a lot...

When electricity reached factories in the 1890s, owners did the obvious thing. Pulled out the steam engine, dropped it in a big electric motor, and left everything else exactly as it was. But the floor was built around steam. One central power source: overhead shafts, belts running down to every machine. The layout existed because of how far power could travel before friction ate it. Productivity barely moved for three decades. The gains only came when factories were rebuilt, with a motor inside each machine and a floor laid out around the work instead of the driveshaft. Computer use is in the belt-and-shaft phase. The models got dramatically better and they’re still driving a machine built for human eyes and hands. Screenshot, interpret, click, screenshot again to check. Hundreds of loops per job. @PineAIAssitant rebuilt the machine instead. A computer built for AI, reading web pages as structure, with the intelligence to use it built in. Co-founder Dylan Wang: “We spent years making the model smarter, now we’re making the computer worthy of the model.” Swapping the engine was never the hard part. Nobody wanted to rebuild the floor.

ASML spent more than twenty years building a machine nobody else would attempt. Molten tin droplets fired across a vacuum chamber, 50,000 times a second, each one hit by a laser and vaporized into plasma hotter than the surface of the sun. That’s just how the light gets made. Every advanced chip on earth now passes through their machines, and there is no second supplier. Twenty years of hard engineering isn’t something a competitor can decide to catch up on. Sabi just raised $50M from Khosla, Accel, Initialized and Kevin Weil. The part worth noticing is that they build the hard things themselves... their own non-contact EEG chip, fabbed at TSMC, and their own model trained on neural data they collected. A cap that reads your thoughts was never going to run on somebody else’s hardware. That’s why nobody had one.

This story still blows my mind... At 1am on a Sunday, Elon called an all-hands meeting at the SpaceX factory in Boca Chica. One question: why isn't this factory running 24/7? His engineers said they needed more people to run shifts. 48 hours later, SpaceX had hired 252 workers and doubled the factory workforce. Most companies and CEOs would have scheduled a planning meeting to discuss the staffing gap and take months to hire those people. But Elon just did that within 2 days.

Peter Thiel on why zero growth is not a neutral outcome: Thiel pushes back on the precautionary impulse to slow AI. Even if extinction risk probabilities are 5-10%, that doesn't necessarily mean you should always go slower. "If you do nothing, if there is no progress on any dimension, I don't think our societies work at all." He defines the middle class as people who expect their children to do better than themselves. The intergenerational progressive compact in Western society was built on the assumption of progress—economic, material, health, social. "When that breaks down, I think the whole society starts to derange in very, very strange ways." People implicitly assume the alternative to AI is a zero growth world that will somehow be a peaceful social democratic society. "I don't think that's true. The alternative is also far from neutral." You can never just turn the clock back. The question is not AI versus the status quo. It's AI versus societal breakdown. Thiel: the cure for polarization and extremism is not less progress. It's more.

Mark Zuckerberg on his singular competitive advantage over Sam Altman and Elon Musk: Because the company was profitable early on and has always been under his complete control, he could afford his natural instinct to tune out external criticism and focus instead on what he calls "this pretty deep need I have for internal cohesion." Zuckerberg explains: "A lot of people probably care about public perception more than I do, but I probably care about team cohesion more than most other people do, which means that it weighs on me more. So at times when it's not strong, it probably stresses me out more." He continues: "My main tool in the world is sitting down, building a team, and methodically improving something over a long period of time. It's not a very sexy process." This echoes what Steve Jobs said about Bill Gates—that Gates's advantage was his willingness to grind methodically over long periods. "But also fundamentally, because of that architecture where I don't need to raise money, I don't have to convince a lot of other people along the way." He contrasts this with Sam Altman and Elon Musk, whose companies were both capital intensive and unprofitable early, and thus had to be memed into life. Zuckerberg reflects: "Now, the question is, how much did that situation turn them into what they are, versus how much did it just select for them? I think there's probably both some feedback loop where doing it built their confidence and built their skills at being public figures, but they also had some fundamental skill at being able to do it." "They probably also like it more. I think some people have that deep need to be liked externally by people they don't know, and I don't have that." "Maybe that's a disadvantage, because I think to some degree, you want that. It probably actually does matter what people who don't know you think. But I think to some degree, doing it is so hard that you only really go out of your way to do it at an excellent level if there's some part of your psychology that needs it." "I don't need that, and they may."

Mark Zuckerberg's lifelong obsession with Augustus and the Roman Empire: The game 12-year-old Mark built after the virus crash was set in the Roman Empire. The player would conquer the world one territory at a time. Perhaps that was a sign. This through line continues throughout his life: In high school, he excelled at Latin. Nearly a decade later, he still recalled the Latin verse for Jupiter's decree in the Aeneid: "For these I set no bounds in space or time. I have given empire without end." At Harvard, one of the first websites Mark built was a class study tool for the art history course called Rome of Augustus. At early crisis Facebook meetings to fend off threats from Google, Mark would end his speeches by declaring: "Carthage must be destroyed." All hands meetings would end with him shouting "domination." Sean Parker, Facebook's early president, warned Mark that shouting domination could one day be cited in an antitrust suit. Parker later observed that Zuck had this imperial tendency in him. In recent years, Zuck has periodically mused in public about the really harsh approach Augustus employed before establishing 200 years of world peace. "That didn't come for free," Mark said. "He had to do certain things." At the company's flagship annual conference in 2024, Mark wore a shirt with the words: "either a Caesar or nothing." Mark's wife Priscilla Chan joked that on their honeymoon in Rome, his constant pursuit of Augustian sites and statues made the emperor their third wheel. The names of their three children are Maxima, August, and Aurelia.

Jimmy Iovine grew up in Red Hook, Brooklyn. His father was a longshoreman. Nobody expected him to end up anywhere near a recording studio. He got in through one relationship, Roy Cicala, who ran the Record Plant in New York. Cicala took a chance on him as a janitor. Iovine absorbed everything he could and within a few years was engineering sessions for John Lennon, then Bruce Springsteen, then Tom Petty. Each one wasn't a contact. It was a friendship. When Dr. Dre mentioned the earbuds that came with the iPod sounded terrible, Iovine didn't just nod. He said: "Forget sneakers. Let's make speakers." That conversation became Beats by Dre. Apple bought it for $3.2 billion. "Dre and I have never had an argument. It's because of that relationship that we are both as successful as we are." Iovine didn't collect people. He chose a few and went all the way in. "Build friendships with a sniper rifle, not a shotgun." His career is what that looks like over 50 years.

Michael Bloomberg on getting fired at 39: After 15 years of 12-hour days and six-day weeks at Salomon Brothers, managing partner John Gutfreund told Bloomberg his life at the firm was finished. "So there I was, 39 years old and essentially hearing, here's $10 million and your history," Bloomberg recalls. "I was terminated from the only full-time job I'd ever known and from the high-pressure life that I loved." Most of the 63 partners were asked to stay on as employees of the new company. Bloomberg and a handful of others were pushed out. "Was I sad on the drive home? You bet, but as usual, I was much too macho to show it and I did have $10 million as compensation for my hurt feelings." He explains his mindset at that turning point: "If they had said, we had another job for you, I'd have done it in a second. I was willing to do anything that they wanted. It was a great organization and I would have been happy to stay. I would have never left voluntarily." Bloomberg refused to dwell on what he'd lost: "Afterward, I didn't sit around wondering what was happening at the old firm. I didn't go back and visit. I never look over my shoulder. Once finished, gone. Life continues." Steve Jobs said it best: sometimes life hits you in the head with a brick, but don't lose faith. Bloomberg didn't. He took that $10 million severance and jumped right back into the arena... this time as an entrepreneur. Fifteen years later, he had a billion-dollar business.

Eight weeks after returning to Apple in 1997, Steve Jobs walked into a product review meeting. Apple was 90 days from bankruptcy. They had 15 product categories, dozens of Mac variants, a digital camera, a TV set, printers, speakers. The company was trying to be everything to everyone and hemorrhaging a billion dollars a year doing it. Jobs asked one question: "Which ones do I tell my friends to buy?" Nobody had an answer. He cut 70% of the product roadmap on the spot. Then he said something that stuck: "I got out of the meeting with people who just had their projects cancelled, and they were three feet off the ground with excitement because they finally understood where the heck we were going." That's the part nobody talks about. The people whose work got killed weren't demoralized. They were relieved. Clarity did that. One year later Apple made $309 million in profit.

Jeff Bezos on what founders need to understand about brands:

Why Sam Altman dedicated his life to AI: "I always thought that AI would be the most amazing kind of craziest thing. I never thought I'd actually get to work on it, but I always, I always like loved it." He came to college to study it. He worked in the AI lab the summer between his freshman and sophomore year. But nothing was working. A professor memorably told him: "The one thing we know doesn't work is deep learning. We tried that for a long time. You know, it's the kind of most guaranteed way to have a bad career." Sam was an impressionable freshman. He assumed the professor was right and pursued other things. It was clear by 2005 that AI was not working. He happened to get into startups accidentally, but then fell in love with it. The startup investing detour was super helpful. Going from investor to founder is unusual... the opposite of the normal path. "As an investor, you kind of watch all the crux moments. So you don't get the kind of like operating practice that you do just day in and day out running a company. But you've seen a lot of the like big crux moments, a lot, kind of all day long you see those." When facing a decision at OpenAI, Sam can think back: this is what happened when this company had a similar situation. This founder made this mistake. This founder got it really right. As a kid, Sam was very optimistic. The more something seemed impossible, the more intrigued he was. "The idea that we could invent a technology that would let us do everything else, that would just empower people in this way that no other single technology could, that always seemed like innately incredibly appealing to me." AI is the strongest version of that vision he can imagine.