Evan Spiegel turned down $3 billion from Mark Zuckerberg at 23 years old. He was still living at his dad's house. Facebook had just paid $1 billion for Instagram and $19 billion for WhatsApp. Two years earlier, Snapchat raised $485,000 at a $4.25 million valuation. It was already doing 75 million monthly users. Spiegel and his co-founder Bobby Murphy had each cashed out $10 million in an earlier round, enough that walking away wouldn't break them. They met Zuckerberg at Sheryl Sandberg's condo in Santa Monica. Took it to the board. Said no. Eleven years later, Snapchat is still independent. Spiegel says he's never doubted it for a single day.

On July 8, 1932, the Dow Jones hit the lowest point of the entire Great Depression. Down 89 percent from its 1929 high. That same day, the richest man in American history turned 93 and recorded a birthday message for the country. He started in 1859 with 2,000 dollars, half his own savings, half borrowed from his father. His peak fortune, built three decades later, was around 900 million dollars. Economists put today's equivalent anywhere from 300 billion to over 600 billion dollars, depending on the method. By the time he recorded that birthday message, he had already given away 540 million dollars of it. He didn't mention the market. He didn't predict a bottom. He said this: in the 93 years of my life, depressions have come and gone. Prosperity has always returned, and will again. He wasn't guessing. He'd already lived through the panics of 1873, 1893, 1907, and 1929. The Dow never closed lower than it did that day. It took 25 years to fully recover, but it never saw that bottom again.

Pablo Escobar's cartel spent 2,500 dollars a month on rubber bands. Not product. Just rubber bands, to hold the cash together. At his peak his fortune was estimated between 25 and 30 billion dollars. He made the Forbes list of the world's billionaires seven years running. His own family says they wrote off 10 percent of the cash every year, eaten by rats or ruined by water, buried in fields and warehouses because there was nowhere left to put it. In 2020, 27 years after his death, one of his nephews found 18 million dollars still hidden inside the wall of one of his old properties. Nobody had come back for it. Most people spend their life chasing more money. Escobar had the opposite problem. More cash than the world had places to hide it.

Most billionaires buy a yacht. But She built a 344 kilometer railway instead. It's part of a 10 billion dollar project that includes the mine on one end and the port on the other. A single loaded train on that line weighs about 35,000 tonnes. She didn't lease track on someone else's network to move the ore. She built and owns the whole chain herself: mine, railway, port. In 2016 Gina Rinehart, went further into the outback. Hancock Prospecting took a 67 percent stake in S. Kidman and Co, Australia's largest private landholder, in a deal worth 365 million Australian dollars. The pastoral leases that came with it cover an area roughly the size of South Korea. Most fortunes get measured in dollars. Hers gets measured in tonnes of rock and square kilometers of desert, moved on rails she owns outright.

The instruction was one sentence: check my Uber driver status and text Alex an update. PhoneUse opened the Uber app first. No driver found yet, no ETA. Then it switched to Messages, pulled up the thread with Alex, read back far enough to see Alex already knew they were meeting near a Starbucks, and wrote its own line: Uber is still finding a driver, so there isn't a confirmed arrival time yet. Nobody wrote that sentence. Nobody told the agent what Alex already knew. Two apps, neither one with an API, and a phone being driven the way a person drives it, reading one screen to decide what to type on another. Computer use agents live inside a browser tab. This one was holding a real phone, switching apps mid task, carrying context between a ride and a conversation nobody fed it directly.

George Soros put $10 billion against the British pound in September 1992, most of it borrowed. He made more than a billion dollars in a single week. The Bank of England raised interest rates from 10% to 12% in one morning. Then promised 15% by the afternoon. The pound kept falling anyway. Stan Druckenmiller ran the trade day to day and called it inevitable. Soros pushed him past the full Quantum Fund, to one and a half times it. The downside risk on that leveraged bet was about two and a half percent. One and a half times that comes out to roughly four percent. A four percent loss wouldn't have hurt the fund at all. The mechanism was simple. Borrow pounds, convert them into other currencies, wait. When the pound fell, Soros repaid the loan in now-cheaper pounds and kept the difference. Britain pulled the pound out of the European exchange rate mechanism that day. British taxpayers lost an estimated six billion pounds defending a price that couldn't hold. Ten thousand dollars invested with Soros back in 1969 was worth more than ten million dollars by the time this trade happened. Everyone remembers the billion dollar week. Almost nobody remembers the actual risk was four percent.

The image editing leaderboard just got a new name sitting next to OpenAI and Google: Krea. Krea 3 Ultra landed at #4 on Artificial Analysis's image editing leaderboard. 1151 Elo. Google's Nano Banana 2.1 scored 1152. One point apart. Same score as Microsoft's MAI-Image-2.6. What happened: - Krea 3 Ultra: 1151 Elo, #4 overall - Only OpenAI and Google rank above it - Krea 3 Fast: 1125 Elo, ahead of Nano Banana 2 - Votes are blind raters don't know which model made which edit Why it matters: This isn't a marketing claim. It's a blind benchmark where the model has to win on output, not on who built it. A team a fraction of Google's size just proved it can go head to head with the biggest labs in the world, and the judges had no idea who they were voting for. The gap between "big lab" and "best model" is closing faster than most people are pricing in.

Leona Helmsley left $12 million to her dog. She left two of her own grandchildren nothing. In the video below, archival footage shows the woman New York tabloids called the Queen of Mean, around the same time a former housekeeper testified in court that Helmsley once said: "We don't pay taxes. Only the little people pay taxes." Helmsley ran a hotel and real estate empire with her husband Harry. In 1989 she was convicted of federal tax evasion and served 19 months in prison. When she died in 2007, her will left $12 million in trust for her white Maltese, Trouble. Her brother received $10 million. Two of her grandchildren got $5 million each, on one condition: they had to visit their father's grave at least once a year to collect it. Two other grandchildren got nothing. The will gave no reason, only said they knew why. A judge later cut the dog's trust from $12 million to $2 million. Trouble needed round-the-clock security after the news broke. People sent death threats to a dog. Keep this part. The woman who told her own staff only little people pay taxes made sure almost nobody in her family collected without conditions. Except the dog. Was Trouble the only one in that family she actually trusted, or was the will just her last word on everyone who'd disappointed her? Save the video to watch Leona Helmsley's trial footage in full, and follow for more stories about what billionaires do with the people closest to them.

$5,000 in savings. Zero experience in fashion, retail, or manufacturing. Twenty years later, she sold a majority stake in the company she built from that $5,000 in a deal that valued it at $1.2 billion. She was selling fax machines door to door during the day. At night she was cutting the feet off her own pantyhose to test an idea. She wrote her own patent application because she couldn't afford a lawyer. Every hosiery mill in North Carolina turned her down. One finally said yes only because his daughters talked him into it. She never took outside funding. She kept the whole company, the whole time, until she decided herself to sell a piece of it. Her name is Sara Blakely. At 41 she became the youngest self-made woman on the Forbes billionaire list. If you're sitting on an idea and waiting for permission or funding to start testing it, save this. She tested hers with scissors and a pair of pantyhose.

An agency dropped her and told her she'd never book a job again. She'd been modeling long past the age the industry considered acceptable. They wouldn't even put her new photos on their own website. She sat down in their lobby and refused to leave until they released her from the contract. They let her go. As far as they were concerned, she had nothing left to offer. Three months later, with a small agency, she had a billboard in Times Square, a billboard at Madison Square Garden, a magazine cover, and her first editorial shoot. People started telling her they'd been trying to book her for years. She never got those calls. Her old agency had quietly been turning down work in her name. Her son is Elon Musk. If someone's already decided you're finished, save this. The agency that wrote her off was wrong, and the numbers proved it within a season.

She had $5,000 in savings and zero experience in fashion, retail, or manufacturing. She was selling fax machines door to door during the day. At night she was cutting the feet off her own pantyhose to test an idea. She wrote her own patent application because she couldn't afford a lawyer. Every hosiery mill in North Carolina turned her down. One finally said yes only because his daughters talked him into it. She never took outside funding. She kept the whole company. Her name is Sara Blakely. The company is Spanx. At 41 she became the youngest self-made woman on the Forbes billionaire list. If you're sitting on an idea and waiting for permission or funding to start testing it, save this. She tested hers with scissors and a pair of pantyhose.

She left an abusive marriage with three kids and no savings. Today, one of them is the most famous person on earth. Back then, she explained her bruises the way a lot of women in her country did. She said she'd slipped in the shower. She slept on a pull-out in the kitchen so her kids could have the one bedroom. The hospital paid her $300 a month for her internship. They charged her $77 of it just to let her and her kids eat and stay there. She didn't quit. She ran a modeling school, saw nutrition clients from 8am to 7:30pm, then taught two night classes a week just to get a free office she couldn't otherwise afford. At 69, she became the oldest spokesmodel in CoverGirl's history. Her son is Elon Musk. Nobody talks about the twenty years before the rocket launches. Would you have kept going at the rate she did, with no end date in sight?

Lehman Brothers held $639 billion in assets the day it filed for bankruptcy. Every counterparty on the other side of that balance sheet thought they knew their exposure. None of them had priced in what happens when ten banks mark down the same asset in the same week. The model below isolates three variables: exposure to a single counterparty, the haircut applied once that counterparty defaults, and how many other institutions get forced to take the same haircut at the same time. It starts with a $10 billion exposure and an $8 billion capital buffer. At a 10% haircut, the loss is $1.3 billion. Bruised, still standing at 84% of buffer. At 25%, contagion pulls the loss to $4.1 billion. Buffer down to 49%. At 40%, something breaks. The direct writedown alone would have been survivable. The contagion multiplier wipes the entire $8 billion buffer in the same week. Past that point the bank isn't undercapitalized. It's insolvent on paper before anyone has time to raise new capital. Watch the two lines in Fig. 1 cross right at that 40% mark, loss climbing while buffer falls, meeting at zero. This is the part people miss. A bank doesn't fail because one counterparty defaults. It fails because that default gets repriced across every institution connected to it, all at once, before anyone can react. Lehman wasn't undercapitalized by the numbers on its own balance sheet. It was undercapitalized by the numbers once everyone else's balance sheet got marked down too. The Mathematics of Systemic Risk Interbank Exposure, Haircuts and Cascading Default save this

Michael Saylor owned 52% of an $11 billion company at 34. Three weeks after bragging about it on national TV, the stock dropped 62% in a single day. The company had overstated its revenue. Regulators forced a restatement. His paper fortune, somewhere around $5 to 6 billion, got cut in half before lunch. He didn't rename the wreckage and disappear. He ran the same software company for the next 20 years while it stayed a name nobody brought up anymore. Then in August 2020 he moved $250 million of the company's cash into Bitcoin. That position is up more than 2,600% since. He renamed the company Strategy. It now holds over 800,000 Bitcoin, more than any company on the planet. He stepped back from CEO to executive chairman and let someone else run the software business. The bet that made him famous nearly erased him. The one that made him rich again had nothing to do with software.

Jamie Dimon spent 15 years at Citigroup. Then Sandy Weill fired him. A year later, Dimon called Weill and asked him to lunch. Most people would spend years trying to prove their old boss wrong. Dimon sat across from his and listed the mistakes he'd made. Weill agreed with him. The person who fired you knows your weak spots better than anyone. Dimon just asked.

Jamie Dimon just ran the math on AI and jobs, out loud: two million truckers, $150,000 down to $25,000. He ran the numbers himself. Automate trucking, and a driver making $150,000 a year is looking at a next job that pays $25,000. His words: "you'll have civil unrest." Then the interviewer asked him directly: will JPMorgan have fewer employees in five years? His answer: "My guess is it'll be fewer employees. Yeah." The same man already has 150,000 JPMorgan employees using an internal LLM every week, and 500 live AI use cases running across the bank. It's the CEO of the biggest bank in America, already deploying the thing, telling you what it does to headcount before it happens to his own company.

Amazon is worth $2.67 trillion today. It started with $245,573 money Jeff Bezos's parents handed him in 1995 after he told them straight up there was a 70% chance they'd lose every dollar. They wired it anyway. He'd already quit a senior VP job at a hedge fund in New York to do this. Not because the math worked. Because of one question: which decision will I regret more at 80? In his own words: "I didn't think I'd regret trying and failing, and I suspected I would always be haunted by a decision to not try at all." The framework was never about e-commerce. It's about running the regret calculation before you run the fear calculation. His parents ran the same calculation on him and bet on him anyway.

Bill Ackman turned $27 million into $2.6 billion in three weeks. He made most of it while telling CNBC, on the verge of tears, that hell was coming for the US economy. The timing looked worse than it actually was. His hedge was already worth $2.75 billion six days before that call happened, so the interview didn't make the money, the credit markets already had. Here's the mechanism. In early March he bought credit protection on about $65 billion of corporate bonds for roughly $27 million a month, cheap because almost nobody was pricing in a pandemic yet. Three weeks later, as credit spreads blew out, he closed the position for $2.6 billion. Total cost: one month of premiums. He didn't sit on the cash. On the same call where he said Hilton was going to zero, he said he'd been buying it all the way down. Hilton, restaurant brands, Starbucks, the same stocks that had made his fund look reckless months earlier. Pershing Square had lost $9 billion in assets between 2015 and 2019 on bad calls like Valeant. This was the trade that reversed it: the fund gained 7.9% in March 2020, the month the S&P fell 17%. He wasn't right because he sounded scared on television. He was already right, six days before anyone heard him say it.