Part IThe Story
The Bottle in the Refrigerator
On Wednesday, December 9, 2020, a bottle of Champagne spent the entire day chilling in a refrigerator. It was not opened that morning. The man it belonged to had been on television since 5:30 a.m., cycling through interviews with anchors who wanted to know how a food-delivery company that had lost money for most of its existence could be worth tens of billions of dollars. It was not opened at the opening bell, when DoorDash began trading under the ticker DASH on the New York Stock Exchange, priced at $102 a share. It was not opened at noon, when the interviews ended and Tony Xu, thirty-six years old, did what he did every week: logged into Zoom for his regular business reviews and went through the operation line by line, looking at what had improved, what had broken, and why. CNBC and Fox Business were covering his company's debut at the same hour he was covering the company itself. It was not opened at the closing bell either, by which point DoorDash shares had finished 86 percent above the offering price and the company was worth roughly $72 billion. The offering raised $3.37 billion. Xu owned about 5 percent of the company and controlled about 69 percent of the vote. That afternoon he became a billionaire.
The cork came out later that night.
It is a small thing. You could read it as discipline, or as an anxious man's refusal to let himself enjoy anything, or as a CEO's instinct for a good anecdote. A Sequoia Capital profile written a year later presented it fondly, as the image of a founder who "intended to keep to his weekly routine as best he could." Some readers will find it a little too tidy, a parable of humility from a man whose board had approved, the month before, a "CEO Performance Award" valued at about $413 million. Both readings hold up. Seven weeks before the IPO, on October 23, 2020, DoorDash had sent Xu a confirmatory employment letter at its offices at 303 2nd Street, South Tower, 8th Floor. It set his base salary at $300,000, payable "in accordance with the Company's normal payroll practices."
The bottle matters because of what Xu does with time. The day of his greatest public triumph was spent mostly on the work underneath it, and the reward was deferred until the work was finished. That ordering, labor first and celebration after, did not start in 2020. It goes back to a Chinese restaurant in central Illinois, to a nine-year-old who could barely reach the handle of a lawnmower, to a family that treated a McDonald's Happy Meal as a prize to be earned with a good report card. It also goes back to one night in the spring of 2013, when four people with almost no money stayed up baking cookies for strangers who had not asked for anything.
That night comes last.
By the Numbers
From PaloAltoDelivery.com to the Nasdaq-100
43 minTime to ship the first version of PaloAltoDelivery.com in January 2013
$120,000Y Combinator seed check in summer 2013, for a 7% stake
~40%Share of the bank account spent refunding one night of late orders
86%First-day gain above the $102 IPO price on December 9, 2020
69%Voting interest controlled by Xu at the IPO, through Class B shares
$13.7BDoorDash revenue in 2025, with $935M in net income
40+Countries served after the Wolt and Deliveroo acquisitions
A Name Taken From Television
Even the number at the start of the story changes depending on who tells it and when.
In a 2017 essay for Fortune, Xu wrote that his family arrived in the United States "with only a few hundred dollars in our pockets." In a 2026 conversation at South Park Commons he said it was "250 bucks in the bank." A podcast summary from the same year puts it at $200. His age at arrival also varies. He was four in some accounts and five in others, including his own Fortune essay. The year is consistent: 1989. So is the departure point, Nanjing, and the destination, Champaign, Illinois, a university town of cornfields and engineering buildings where his father had been admitted to the University of Illinois to study aeronautical engineering and applied mathematics. In China his father had been a professor. In Champaign he became a graduate student on a student visa, then a postdoctoral researcher on an H-1B, and some years later a citizen. Along the way he waited tables at an on-campus restaurant to help pay for his studies.
The figures drift because they come from memory. "It's only one that can be told backwards," Xu once said of his own path. When he was mowing lawns or washing dishes, he said, he could not have predicted any of what followed. A biography told backwards rounds things off. A few hundred dollars turns into $250, then into $200, depending on what the teller wants to emphasize.
His mother, Julie Cao, had been a physician in China. The United States did not recognize her medical license. By Xu's account she worked three jobs a day for about twelve years, "all of those jobs," he told Keith Rabois in 2021, "happened to be inside local businesses." One was a Chinese restaurant. The details shift between versions here too. The Los Angeles Times reported that she ran the restaurant to save for medical school. Xu's own 2016 TechCrunch essay calls it "her first business." His Fortune essay describes it as "the first business she worked in." What stays constant is the outcome. She eventually earned an American medical credential and opened her own practice, first acupuncture and medical clinics in Champaign and later a medical clinic in the South Bay. Xu said in 2019 that the clinic took her twelve years to open and that she had run it for about two decades since. In every version she held a credential the new country would not honor, worked her way back to it, and then built her own business around it.
The family used federal assistance programs. "Visiting McDonald's was a luxury," Xu wrote. He worked hard for good grades so he could earn a trip to the Golden Arches, perhaps once every couple of months. "Eating out at a restaurant was a luxury for my family." That sentence would later underlie a company that sells the experience of eating out at home.
Then there is the name. He was born Xu Xun, 徐迅. As a small child who spoke little English, picking it up, he later said, from basketball and television, he chose a new first name from his favorite show. The show was Who's the Boss?, and the character was Tony Micelli, played by Tony Danza. "I walked with my dad to the immigration office and legally changed my name," he told NPR in 2018. A Sequoia profile notes that he made this decision himself, at five, and presents it as an early sign of a boy who "learned early on how to make his own decisions."
The image is hard to set aside: a professor turned graduate student walking his five-year-old son into a federal office so the boy can take the name of a sitcom character, so he will sound like he belongs in a Champaign schoolyard. The show's title is a question about who holds power in a household. Thirty years later, proxy advisers and union-affiliated investors would ask the same question about the company the boy built.
Xu has said that his family moved often and that changing schools taught him what it meant to be an outsider, and how to adapt to a new environment quickly, before it notices you. Stanford's business school framed this in its 2021 interview as a lesson in adaptability. It also sounds like someone learning young to read a room before anyone reads him.
Today another Tony Xu writes a blog from the San Francisco Bay Area. He is a software engineer who posts field notes on AI workflows and A/B testing, in English and Chinese, and in August 2026 he was on paternity leave in China with his family, spending days "deliberately organized around being present rather than maximizing output." He has no connection to DoorDash. He shares the name, which a boy picked off a television screen precisely because it was common, because it would not stand out.
Whatever My Mom Asked
From about age nine until eighteen, Xu worked at his mother's restaurant. He has told this story more than any other, in CNBC interviews, in his IPO-day remarks to Jim Cramer, in podcast after podcast. The most precise version is from the Los Angeles Times in 2018, and it is the least sentimental:
— Tony Xu, to the Los Angeles Times, January 2018It was really whatever my mom asked, so washing dishes, busing tables, sometimes making websites or fixing a payment system or a printer. Nothing glamorous. It's whatever my mom asked me to do to help her execute.
The list matters. A boy washing dishes is a familiar immigrant image. A boy who also fixes the payment system and builds the website is something else: the restaurant's unofficial IT department, solving the back-office problems of a small business before he could drive. Payment systems, websites, printers. Two decades later these would be DoorDash's products, the merchant tools, online storefronts, and order-routing hardware that the company describes, in its corporate boilerplate, as helping "businesses of all kinds grow and innovate."
He also ran his own business at nine, mowing lawns with a friend and charging extra for patterns. "We mowed lawns and charged more for designs, like a checkered lawn," he said in 2015. "I learned you can do more than you think you are capable of doing. I could barely reach the handle of the mower!" He saved money for a Nintendo. These are the usual stories of an entrepreneur's childhood, and you could find them in many founder biographies. The checkered lawn is the specific part. Even at nine he charged for the pattern, not just the cut.
His sharpest early observation about small businesses appeared in his 2016 TechCrunch essay, written when DoorDash had only a few hundred employees and was still raising money with difficulty:
Washing dishes in my mom's restaurant as a kid, I realized that when you run your own business you're always worried about 50 different things at any given time. As a result, the way to support local businesses is not by cutting a few costs. Rather, it comes from creating an entirely new business line and growing their sales, all without adding a 51st hassle for the business-owner to worry about.
The fifty-first hassle is the key idea. Anyone who has run a restaurant knows the owner's mind is a queue of overdue problems: a broken walk-in, a no-show cook, a late linen delivery, a health inspector. Selling that owner software means adding to the queue. Xu's idea, formed while scraping plates, was a product the owner would not have to think about, one that produced revenue the way a dishwasher produces clean plates, out of sight, behind the swinging door.
Whether DoorDash kept that promise is the subject of thousands of pages of litigation. The intention, though, came from the kitchen.
He still washes dishes. In a 2017 Lifehacker questionnaire, asked for his least favorite chore, Xu said it was laundry. "The only reason I have clean clothes is that I trade chores with my wife: I do the dishes (her least favorite thing to do), and she does the laundry." He met Patti at church when they were both Berkeley undergraduates. They married in 2013, the year DoorDash was founded, and in December 2021 they signed the Giving Pledge together. Their foundation is called 10x Better.
The Optimizer's Apprenticeship
Xu thought he would become an academic. He said so plainly at South Park Commons in 2026, and the record supports him. At UC Berkeley he majored in industrial engineering and operations research, the field concerned with getting the most out of limited resources, and graduated with high honors in 2007. Alper Atamturk, who taught him optimization and later chaired the department, recalled at a 2025 ceremony that he gave Xu a rare A+. "It was clear from the start that Tony would go on to do extraordinary things," Atamturk said, the kind of thing professors say about successful alumni, which is not a reason to doubt it.
The academic path was real. "My background is in math applied toward cancer research, which has nothing to do with delivery," Xu told Y Combinator in 2016. "I worked on a very narrow field of cancer research for three years. I published one paper." He liked the problem because it was large and important "and it was very complicated. It was not something I was going to figure out in three months or even a year."
He left research anyway, and his next moves look less like a career than like a sequence of deliberate discomforts. He became a business analyst at McKinsey & Company, then a corporate strategist at eBay and PayPal, where by the accounts of DoorDash's own biographies he "led special projects for the CEO and CFO." He did business development at RedLaser, a barcode-scanning app eBay had acquired. He spent less than two years at each. He explained the pattern to the Los Angeles Times: "I didn't have any business training or background in it, but it was exactly for that reason that I found a place like McKinsey so appealing to work in. I found that it's in places where I am most uncomfortable typically where I can maximize my growth and learning."
Then came Stanford's Graduate School of Business, where he was named an Arjay Miller Scholar, the distinction for the top of the class. During business school he interned at Square, in product. That internship gave him two things that would matter: a stack of the small white card readers Square was then distributing, which plugged into an iPhone's headphone jack, and a connection to the company that would later sell DoorDash its first major acquisition.
Viewed from where it ended, the training is clear. Operations research is the mathematics of routing, scheduling, and allocating scarce resources under constraints. Its classic problems include the vehicle routing problem and the assignment problem, which match agents to tasks at minimum cost. If you set out to train someone to run a three-sided delivery marketplace, you would start with that degree, then give them a consultancy, a payments company, and a point-of-sale startup. It was not planned. "There's no linear path forward," Xu told Lifehacker. "I search for hard problems that offer tangible impact."
When Berkeley's industrial engineering department gave him its first Alumni Achievement Award on April 30, 2025, in the Toll Room of the Cal Alumni House, Xu described his education in his discipline's own terms. He spoke about seeing problems "in work, school, or life" as opportunities to optimize within given constraints. A sophomore named Jake Lee read back a line Xu had once offered as advice to his younger self: "There's no better way to be the expert than just to do the work." The department said nothing about cancer research. The plaza outside the renovated Grimes Engineering Center now carries his foundation's name.
A Booklet of Refused Orders
The founders did not start with a product. They started with a group of people they wanted to serve.
"We didn't start with really with the idea of food delivery," Xu told Rabois. "We started, my co founders and I, with the idea to help every brick and mortar business." The four of them were Xu, Andy Fang, Stanley Tang, and Evan Moore, Stanford students during the 2012–2013 academic year. Their method was immersion, not surveys. They asked shop owners in Palo Alto a simple question: Can we follow you around for a day?
One of those owners ran a macaron shop. Xu has told this part many times, with small variations, to Y Combinator's Garry Tan and to others. Late in the day the store manager showed them a booklet. It held orders she had turned down. All of them were delivery orders.
That was demand nobody was serving, recorded by hand. A bakery in the richest zip code in America was refusing money because it had no driver. The founders imagined the scale: one bakery, then every bakery, then every restaurant, then every store. Xu would later call it the "FedEx of local."
The first product was deliberately minimal. "Whenever you can ship something in 43 minutes to test your idea, I think that's pretty good," Xu told David Senra in March 2026. They bought the domain paloaltodelivery.com because "that alias was available for nine dollars." (Another account puts it at ten.) The site was a static page with eight PDF menus from restaurants the founders already liked. To order, you called a Google Voice number that rang all four founders' cell phones. Whoever answered took the order, placed it with the restaurant, picked it up, and drove it over. Payment went through a Square card reader plugged into an iPhone.
They launched in January 2013. On the first day they got one order.
They skipped class to make deliveries. They paid no salaries. The whole operation ran out of one founder's bank account, and their test of success was crude: if the balance was not falling week over week, they were onto something. Xu later turned this into a short business plan of three questions. Would customers be willing to pay $5 for this service? Would restaurants be willing to pay us a certain percentage? Would drivers be willing to work for this wage? "It was a pretty simple business plan," he told Sequoia, "but those were the answers we needed in order to know whether we had something."
The data surprised them. Deliveries in Palo Alto were faster than deliveries in the much denser San Francisco, because suburbs had parking, single-family homes, and street layouts that made routing easy. The customers were not single young professionals in city apartments, which is what everyone in the Valley assumed delivery was for. They were, as Xu put it, mothers with young children who did not want to wrestle a stroller into a restaurant. In 2013, only about 20,000 to 25,000 of America's roughly one million restaurants offered delivery, almost all of them pizza places in city centers. "85% of restaurants don't deliver," Xu told the Times in 2018, "and because people eat 20 times a week. We thought it'd be a good bet."
They picked restaurants over groceries or retail for an operations researcher's reason. There were about one million restaurants and only about 200,000 grocery stores. More locations meant a denser network, and a dense network was the precondition for a fast logistics system that could later carry anything. Food was the starting category for that network. In 2018, when DoorDash delivered almost nothing else, Xu was already telling reporters, "the goal wasn't necessarily to start and end in restaurants."
Their first big expansion was to east San Jose, not San Francisco or West Los Angeles. People asked why, since the obvious next markets were rich and dense. "We needed to know whether this could be a mainstream product in terms of price point," Xu said, "and moving into a market similar to Mountain View wouldn't have answered that question." Palo Alto had shown that wealthy people would pay for convenience, which nobody doubted. East San Jose would show whether ordinary people would.
In the spring of 2013 the four applied to Y Combinator with about 150 customers and $10,000 in sales. Asked how big the business could get, Xu wrote down "something like $100 million of revenue." DoorDash passed that in 2018. In the first quarter of 2026 it took in about $4 billion, forty times his original estimate in three months. "It's very hard as an entrepreneur when you're working out of your apartment, to know what that size could be one day," he said on Jack Altman's podcast in 2026. "Turns out a lot of people eat food."
Y Combinator invested $120,000 for 7 percent. In June 2013 PaloAltoDelivery.com was incorporated as DoorDash.
There is one more name to account for. When DoorDash filed its S-1 in 2020, Business Insider noticed that one co-founder was not named in it. Evan Moore, whose name appears in every account of the founding, had become, in the magazine's phrase, "the forgotten fourth cofounder." Baidu's encyclopedia entry describes DoorDash as founded by "three Chinese American students from Stanford University," and does not mention him at all. Fang and Tang became billionaires on December 9, 2020, along with Xu. Moore's name stayed in the Wikipedia infobox and in Xu's own Deliveroo letter, which says "we" when it refers to 2013. Founding stories, like immigrant stories, tend to lose details as they are retold.
A Thousand Days of No
The investors said no.
Xu has put the number of rejections at more than one hundred. He has called the period that followed YC something like a thousand days of startup hell. DoorDash "was not the fan favorite at the YC Demo Day in 2013," he said in Sequoia's Crucible Moments series. "There were a lot of skeptics around whether delivery could ever be economical." The skeptics had history on their side. Webvan and Kozmo.com had failed in the first dot-com bust. More recently, on-demand startups had spent heavily on subsidies and then folded.
As late as 2020, Chris Webb, CEO of the restaurant ordering company ChowNow, put the consensus plainly to Fortune: "What I know is, delivery is not profitable. It's been the same way since these companies started." Asked in 2016 what he believed that few people agreed with, Xu gave a one-line answer: "You can make the economics of on-demand delivery work."
Sequoia invested in 2014, with Alfred Lin as the partner. Lin had been Tony Hsieh's longtime operating partner at Zappos, a company built on the idea that customer service was the product, though Sequoia's founder page simply lists him alongside Isaiah Boone. Keith Rabois, who later invested and later called Xu "one of the most phenomenal entrepreneurs I've ever worked with," was among those who believed early. They were a minority.
The company survived by doing the work itself. "One of the best things that could've happened to our company," Xu said later, "was that every single person did deliveries and customer support every single day for the first year." The practice continued at a lower frequency, once a month for every employee, even as headcount grew into the thousands. Xu found his model for this outside Silicon Valley. Asked in 2016 which book had influenced him most, he named The Score Takes Care of Itself, Bill Walsh's memoir of building the San Francisco 49ers into a dynasty through obsessive procedure.
— Tony Xu, Y Combinator Q&A, June 15, 2016There's enough gross margin in a lot of the consumer businesses such that if mistakes were to occur no serious damage would result. If mistakes were to occur at DoorDash, DoorDash would be dead.
What Xu admired in Walsh was how far down the detail went, from "how to run a perfect seven yard post pattern to how to answer the phone properly at the front desk." For a software company whose margins depend on whether a driver waits four minutes or nine at a counter, that level of attention was a matter of survival. Xu later counted about twenty discrete steps in a single delivery, each with its own causes of delay, and described running "five whys" on failures until the real cause was found. By 2026 he said DoorDash ran hundreds of thousands of experiments a week, of which about 95 percent fail. The 5 percent that succeed compound across the whole customer base for the following year.
The money eventually came, in very large amounts. In 2018 DoorDash raised $535 million from SoftBank, Sequoia, GIC, and the Wellcome Trust. In February 2019 it raised $400 million more from Temasek and Dragoneer. Before the IPO it raised more than $2.5 billion in total and accepted heavy dilution to build its war chest. In December 2018 it passed Uber Eats in U.S. food delivery sales. By March 2019 it had passed Grubhub, with 27.6 percent of the market. Its suburban strategy, dismissed for years, turned out to be where most American customers were.
In February 2019, after the Temasek round, Xu told CNBC: "Right now we're choosing to invest in growth and delay profitability for the entire company." Exactly a year later, in February 2020, he visited Fortune's offices and said something that sounded different: "We've had to dial in the unit economics since day one, and make sure the business had a path to profitability long, long, long before all the headlines started popularizing this theme. I don't understand why everyone's talking about [profits] now, because candidly, for us, we had no choice but to focus on a path to profitability from the beginning."
The Fortune writer said she was skeptical and quoted his 2019 statement back. It is possible both statements were true. A company can track contribution margin per order very closely while spending heavily on growth. Still, the shift shows a founder adjusting his account of the past to fit the present, the same thing that happened with the $200 and the $250. When she noted that DoorDash was still unprofitable, Xu answered: "No, but we're working our way there." And: "As a founder, you can't be confused."
DoorDash reported its first full year of positive net income for 2025, $935 million on $13.7 billion in revenue.
A Human and a Math Problem
On Patrick O'Shaughnessy's podcast, Xu described the business as "a human and a math problem." It is the most accurate description of DoorDash he has given, and the most revealing, because the two halves keep coming apart.
The math half worked. The human half kept ending up in court.
The record is long and specific. In 2017 a class action alleged that DoorDash misclassified drivers in California and Massachusetts as independent contractors. The tentative settlement in 2022 totaled $100 million: $61 million split among more than 900,000 drivers, a little over $130 each, and $28 million for the lawyers. Gizmodo placed that $130 next to the $413 million performance award the board had granted Xu the year before.
The tips came next. In July 2019, The New York Times reported, followed by The Verge, Vox, and Gothamist, how DoorDash's pay model worked. Each Dasher had a guaranteed minimum per order. When a customer added a tip, the money went first toward covering that minimum, which meant it partly subsidized DoorDash's guarantee instead of adding to the driver's pay. Drivers received only the portion of the tip that exceeded the guarantee. A customer's gesture of generosity had been turned into a cost offset. DoorDash changed the policy under public pressure. A Brooklyn customer sued. In November 2020, a month before the IPO, the company paid $2.5 million to settle with District of Columbia Attorney General Karl Racine. In February 2025 it paid $16.75 million to New York Attorney General Letitia James for, in her office's words, "cheating delivery workers out of tips." In 2026 California's AB-578 took effect, making it unlawful for any delivery platform to use tips to offset base pay. The practice was now banned in statute.
In July 2021 Dashers went on strike. Their demand was to see the full tip before accepting an order, which DoorDash did not allow. A third-party app called Para had let them see it, and DoorDash had rewritten its code to block Para. The strike's slogan, quoted by Vice, was "NO DASHER = NO DELIVERIES."
There was more. Restaurants were listed on the app without their permission. A St. Louis spot called Lona's Lil Eats sued after finding itself on DoorDash, with customers redirected elsewhere. In August 2021 the City of Chicago sued, and Mayor Lori Lightfoot accused DoorDash and Grubhub of using "unfair and deceptive tactics to take advantage of restaurants and consumers who were struggling to stay afloat during the COVID-19 pandemic." DoorDash denied it. A $1 billion class action claimed iPhone users were charged more than Android users. DoorDash denied that too. In August 2023 Seattle collected $1.6 million over sick time. In February 2024 California imposed a $375,000 penalty for illegally selling personal data. In June 2023 DoorDash introduced an hourly minimum pay option, but it paid only for time on an active delivery, not for time waiting between orders.
That waiting time is where the largest bill came from. On September 22, 2026, New York Mayor Zohran Mamdani stood at City Hall with Sam Levine, commissioner of the city's Department of Consumer and Worker Protection, and announced a $131.5 million settlement. Officials called it the largest municipal worker settlement in American history. More than $115 million would go to more than 260,000 Dashers. The largest single piece, $83 million, covered how DoorDash calculated pay for "online-but-idle time," the hours a worker is logged in and available but not carrying anything. Under the settlement, workers would not need to file claims. The city would find them in DoorDash's own records. DoorDash's statement was unusually blunt: "Simply put, we screwed up."
This is the tension at the center of Xu's career, and it should not be smoothed over. The same instinct that refused to let a broken night of deliveries go uncorrected, which spent its last dollars on refunds nobody requested, ran alongside a pay system that absorbed customers' tips and a logistics model that treated a driver's waiting time as free. An optimizer minimizes the variables it can see. Idle time does not register as a cost if the model does not count it as one.
Xu has felt this charge personally. On January 3, 2026, an anonymous Reddit post went viral. Its author claimed to be a developer at an unnamed delivery company and described weekly meetings where product managers referred to drivers as "human assets" and assigned them a hidden "desperation score." The post ended, "I'm drunk and I'm angry." The tech newsletter Platformer later called it a hoax. By then Xu had already responded on X:
— Tony Xu, on X, January 3, 2026Holy fucking shit is right! This is not DoorDash, and I would fire anyone who promoted or tolerated the kind of culture described in this Reddit post.
He went on: "Dashers are not 'human assets.'" A desperation score "is an abomination." DoorDash had never charged a "driver benefit fee." "We're not perfect by any stretch of the imagination," he wrote, "but we work every day to make our platform better for everyone who comes to it."
The profanity was notable from a man who had spent a decade sounding like a carefully prepared board memo. Years earlier, DoorDash's general counsel, Keith Yandell, had given him a rule he said he relied on daily: Don't let what you say get in the way of what you want to accomplish. "I write down my objectives ahead of time," Xu told Sequoia, "and I try to share them with the other person." The X post broke that rule. It read like someone who had been told for years that his company treated people like data points, and who, faced with a fabricated version of that accusation, reacted instantly and without his usual care.
On April 13, 2026, a Dasher delivered an order to the White House, which DoorDash promoted as a celebration of the federal "No Tax on Tips" policy.
Who's the Boss?
Seven months after the IPO, someone asked the sitcom's question formally.
On June 10, 2021, the proxy advisory firm Glass Lewis recommended that DoorDash shareholders vote against re-electing Tony Xu to the company's board. Its objection was the governance structure: dual-class stock that gave Xu and his co-founders majority voting power, and a staggered board that kept long-term investors from changing direction. "We believe our concerns regarding the Company's governing documents are severe enough to warrant opposing the election" of Xu, the firm wrote. It also called the front-loaded pre-IPO award, the roughly $413 million CEO Performance Award granted in November 2020, something that "warrant[s] scrutiny." Two weeks earlier, CtW Investment Group, which owned less than 5 percent of the stock, had argued that shareholders faced "competitive and regulatory challenges to DoorDash's business model that may not be addressed unless the board is accountable to all its shareholders."
DoorDash's response was simple: investors knew the structure when they bought the shares. Xu would receive none of the award unless the stock rose fivefold. "Building the next phase of DoorDash will be as challenging as creating what we have thus far," the company said.
Xu kept his seat. As of December 2020 he held voting authority over all Class B shares and 69 percent of the vote. The boy who picked his own name grew into a CEO whose position could not be voted away.
He has not looked like a man who enjoys control in the theatrical way some founders do. In 2017 he told Lifehacker he had no desk, "just my MacBook Air, iPhone, and AirPods," so he could "truly feel the office, and be a fly on the wall for many conversations among teams." He spent more than half his time recruiting, "as I find it has the highest output to input ratio." In 2016 he said the skill he most wanted to learn was managing executives, defined as "anyone who has more experience than I do in a particular area. It doesn't have to be someone with twenty years of experience." He described the founder's instinct "to do it yourself and offer a point of view very quickly" as something to unlearn.
By the time of the 2021 Sequoia profile his week had a fixed shape. Friday evenings were date nights with Patti. Mornings were runs, somewhere quiet, in the Marin Headlands or Golden Gate Park, followed by breakfast with his young daughter. He took merchant customer-service calls regularly and did the occasional delivery. In 2026 he told Senra that he reads support tickets every day and looks for the longest ones, sometimes 2,000 words from a Dasher describing exactly how the routing algorithm failed them, and then traces the order himself through internal debugging tools, step by step.
The CEO with majority voting control reading a Dasher's 2,000-word complaint is the clearest picture of how he sees his own role. The control is structural and absolute. The attention goes to the bottom of the operation. Both are real.
The hardest test of that combination came on November 30, 2022, when DoorDash cut about 1,250 corporate employees, roughly 6 percent of the workforce. Xu's memo to staff is worth reading in full as a document of corporate accountability, because it does something such memos rarely do. It assigns the blame specifically and to one person.
— Tony Xu, memo to DoorDash employees, November 30, 2022Most of our investments are paying off, and while we've always been disciplined in how we have managed our business and operational metrics, we were not as rigorous as we should have been in managing our team growth. That's on me.
The memo also apologized that some employees would "wake up to this news as opposed to reading it during more normal hours," and explained that "prior to COVID-19, DoorDash was actually undersized as a company." Departing employees received seventeen weeks of pay, their February 2023 stock vest, and health coverage through March 31. One line connected the memo directly to Champaign in 1989. Under the heading "Immigration support," it said: "We will set the termination date for March 1, 2023, giving those with visa applications (and a desire to stay in the US) as much time as possible to find a new job."
His father had been a student-visa holder who became an H-1B holder and then a citizen. In 2017, a week after the travel ban, Xu wrote in Fortune that "the H-1B program helps us to compete with global companies," and DoorDash publicly opposed the executive order restricting travel from seven Muslim-majority countries. Five years later, laying off 1,250 people, he set their termination date three months out so visa holders would have time. The son of an H-1B holder knew the visa clock.
Keeping the Names
The local delivery business that began with eight PDFs became international through acquisitions, and in each case Xu let the acquired company keep its name.
The first deal looked back to his own résumé. In 2019 DoorDash bought Caviar, a delivery service focused on upscale urban restaurants, for $410 million from Square, where Xu had interned and whose card readers had taken the first payments. On November 9, 2021, it announced the purchase of Wolt, the Finnish company, for more than $8 billion, about €7 billion. Overnight DoorDash went from four countries to more than two dozen. Wolt kept its name and its leadership. In the 2022 layoff memo, Xu wrote that Wolt employees would "be hearing soon from Miki," meaning Wolt's chief executive, not Xu.
In May 2025 DoorDash agreed to buy Deliveroo, the British delivery company, for about £2.9 billion, roughly $3.9 billion, at 180 pence a share. In the same announcement it disclosed a $1.2 billion all-cash purchase of SevenRooms, a restaurant reservations platform. The Deliveroo deal closed on October 2, 2025. Xu's open letter to Deliveroo's restaurants, customers, and riders reads like a promise not to erase anything:
Before I share what's ahead, I want to make one thing very clear: the Deliveroo app and products you know and love aren't going anywhere.
He wrote "High Streets" with capitals, using the British term. He described Deliveroo's DNA as "its customer obsession and relentless pursuit to improve the experience of consumers, merchants, and riders," and said it "remains the same." He signed it "Onward." He also restated the original mission in one sentence: "When we started DoorDash in 2013, our mission was simple: grow and empower local economies."
A man who gave up his own name to fit in now buys companies on the condition that they keep theirs. This could be read as plain brand management, since rebranding a beloved local champion is expensive and risky. But he is consistent about it. The letter's subject is preserving what already works locally, and Xu wrote it.
The network kept growing. DoorDash added grocery in August 2020, alcohol and DoubleDash in 2021, and an exclusive partnership with Ace Hardware's more than 4,000 stores in September 2025. It debuted on the Fortune 500 in 2024 at No. 443. It moved its listing to the Nasdaq in September 2023 and joined the Nasdaq-100 on December 18. In 2025 it introduced Dot, an autonomous delivery robot, which Xu demonstrated to Stripe's John Collison on the Cheeky Pint podcast in April 2026 ("We got Dot coming," Xu said, and Collison said it was "like a TARDIS"). Xu explained that real autonomy means solving the "last two feet," the stretch from curb to doorstep where robots usually fail.
Not every expansion worked. The company that went into Sendai, Japan, in June 2021 announced in February 2026 that it was leaving Japan, along with Qatar, Singapore, and Uzbekistan. The 2018 partnership with Walmart ended in September 2022. In March 2025 a deal with the buy-now-pay-later company Klarna let customers pay for takeout in installments, and the internet mocked it as "Collateralized DoorDash Obligations." CNN asked whether it was a sign of a weakening economy.
The renaming went the other way too. In June 2026, ahead of the World Cup, New Jersey Transit announced that its Meadowlands Line would be called the "DoorDash Meadowlands Rail Line" through June 2027. The company had grown large enough to put its name on public infrastructure.
On Jack Altman's podcast that summer, Xu gave founders advice phrased in the language of his old discipline. "I think a lot of times you kind of, as an entrepreneur, have to take the greedy algorithm," he said. "You have to keep going all the way and seeing if there's more." In optimization, a greedy algorithm takes the best available option at each step without looking ahead. It is fast and often effective, and it is known to get stuck in local optima, where each step looked best but the overall result is not. Xu meant it as encouragement to keep pushing. When Mamdani announced the New York settlement in September, at least one account of the press conference used the phrase against him: the mayor blasting Xu's "greedy algorithm." A term from a Berkeley optimization class had become a political attack line.
"If you're in technology and you are not making improvements, you are actually decaying," Xu told Fortune in 2025. "Until it's over all of a sudden."
Five in the Morning
It was the company's third month, in the spring of 2013. The domain had cost nine dollars, the menus were PDFs, and payments went through Square dongles. The four founders had no salaries and, by Xu's account, less than two weeks of cash.
On one night every order was late, "by at least an hour on every single delivery," he told David Senra thirteen years later. Not most orders. All of them.
Afterward the four sat down and added up what it would cost to refund every customer. Xu has given the figure in slightly different ways over the years, as he does with the arrival money. To Sequoia it was "a double-digit percentage of our bank account." To Senra, "about 40%." To Collison, "over 40%." The decision time is the same in every version.
"Were the customers asking for the refunds?" he said. "No. No one was asking for anything. The night was over."
"Within fifteen seconds, we decided — okay, we've got to make right by the customer. We've got to refund everybody."
They refunded every order. Then, with the company's bank balance cut nearly in half, the four of them stayed up through the rest of the night baking cookies. Xu has never said whose oven it was or what kind of cookies. He has said that before anyone woke up, they drove the cookies around Palo Alto and the neighboring towns and left them at the doors of the people they had failed the night before.
"When you see that you can lose someone's trust on one order," he said, "you realize you've got to earn it again the next day."
The cookies arrived at about five in the morning, while their customers were still asleep.
Part IIThe Playbook
Tony Xu's methods are easy to reduce to slogans like "customer obsession" and "bias for action." DoorDash's own LinkedIn account has done that. The record supports something more specific. Xu is an operations researcher who turned his training into a culture, and an immigrant's son who turned scarcity into a habit he kept long after the scarcity was gone. The principles below come from what he actually did, including where it went wrong, because a playbook that skips the tip-skimming years would be incomplete.
The principles
- 1.Fall in love with a customer before a problem.
- 2.Ship in forty-three minutes.
- 3.Go next where the answer is hardest.
- 4.Choose the category with the most nodes.
- 5.Decompose the job to its twentieth step.
- 6.Make everyone do the delivery.
- 7.Pay the refund nobody requested.
- 8.Run software like a football playbook.
- 9.Stay poor in your habits after you get rich.
- 10.Write "that's on me" in the first paragraph.
- 11.Buy the local champion and keep its name.
- 12.Audit what your objective function cannot see.
D
The Decision Timeline
Key moves from a static page to a multinational, several of them made against prevailing opinion.
2013
PaloAltoDelivery.com launches in January: eight PDF menus, one Google Voice number, one order on day one. Y Combinator invests $120,000 for 7%. Incorporated as DoorDash in June.
2014
Sequoia partners with the company after more than a hundred investor rejections.
2018
Project DASH launches. In December, DoorDash passes Uber Eats in U.S. food delivery sales.
2019
Passes Grubhub by March with 27.6% of the market. Changes its tipping model after public backlash. Buys Caviar from Square for $410M.
2020
Cuts restaurant commissions during the pandemic. IPO on December 9 at $102; closes up 86%.
2021
Glass Lewis recommends against Xu's re-election. Wolt acquired for more than $8B.
2022
1,250 corporate layoffs; Xu writes "That's on me."
2025
Deliveroo ($3.9B) and SevenRooms ($1.2B). Full-year net income of $935M.
2026
$131.5M New York City settlement: "Simply put, we screwed up."
Principle 1
Fall in love with a customer before a problem
DoorDash's origin is usually told as a problem identified and then solved. Xu describes it the other way around. "We really started with a passion for a customer group more so than necessarily a problem that we were trying to solve," he told Keith Rabois. The customer group was brick-and-mortar business owners, the people his mother had worked for and eventually became. The problem, delivery, came only after hundreds of conversations and one manager's booklet of refused orders.
The order matters because it keeps you flexible about the solution. A founder attached to a problem tends to defend a product. A founder attached to a customer can change the product many times without losing direction, because the direction was never the product. That is why Xu could tell the Los Angeles Times in 2018 that restaurants were "a starting point," and why the move into grocery, convenience, hardware, reservations, and robotics reads as one continuous strategy rather than a series of pivots. The customer stayed the same and the product kept expanding.
The method was physical. They did not send surveys. They asked, Can we follow you around for a day? A survey only tells you what people already know they want. Watching someone work for a day shows you what they have stopped noticing.
Tactic: Before writing a line of code, spend a full working day following five potential customers and record what they turn away, not just what they complain about.
Principle 2
Ship in forty-three minutes
PaloAltoDelivery.com was so minimal that it barely counted as a product: a $9 domain, a static page, eight PDFs, a shared phone number, and card readers borrowed from an internship. Xu has said it took forty-three minutes to build. The point was not cleverness. It was to test a hypothesis cheaply. "We thought, actually, one of the reasons why maybe delivery in 2013 hadn't been around yet was just because nobody wanted it." The test was built to show that possibility as fast as possible.
His 2018 advice to Sequoia makes the method explicit: "Especially in the beginning, a company should feel like a project, not a grand vision." Write down the specific risks between you and the first customers who love you, and ignore everything else, including "your long-term product vision" and "how much money you've raised." DoorDash's whole early plan was three questions about price, commission, and wage. The ambition existed, the "FedEx of local," but it stayed out of the way of the testing.
The success metric was equally simple. If the founders' shared bank balance was not going down week over week, the idea was working. An early-stage company does not need a dashboard. It needs one number that cannot be argued with.
Tactic: Reduce your business plan to the three or fewer yes-or-no questions that would kill it, and build only what is needed to answer them this week.
Principle 3
Go next where the answer is hardest
After Palo Alto and Mountain View, the obvious next step was San Francisco or West Los Angeles: dense, wealthy, full of early adopters. DoorDash went to east San Jose. "We needed to know whether this could be a mainstream product in terms of price point," Xu said, "and moving into a market similar to Mountain View wouldn't have answered that question."
Founders usually expand where success is most likely, because each success raises the next round. Xu treated expansion as an experiment, and an experiment that only confirms what you already believe teaches you nothing. Wealthy suburbs had shown that rich people would pay for convenience, which was never in doubt. The open question was whether the product could reach a mass market, and only a less affluent market could answer it.
The same reasoning drove the suburban strategy that Sequoia's Crucible Moments calls one of DoorDash's "wrong moves." Competitors concentrated on downtowns because density seemed to favor delivery. DoorDash's data showed the reverse. Parking, single-family homes, and simple street layouts made suburban deliveries faster, and the customers, mothers with strollers, had a stronger need.
X
The Contrarian Ledger
Where DoorDash went against conventional wisdom, and what it cost or earned.
| Conventional wisdom | DoorDash's move | Result |
|---|---|---|
| Launch delivery in dense city centers. | Start in Palo Alto suburbs; expand to east San Jose. | Validated |
| Build a lead-generation layer and let restaurants deliver. | Build the full stack: merchants, drivers, and routing software. | Validated |
| Protect cash at all costs when runway is under two weeks. | Refund every late order unprompted; bake cookies. | Cultural |
| In a crisis, maximize take rate. | Cut pandemic commissions 50%, a roughly $100M revenue hit. | Costly |
| Rebrand acquisitions to consolidate. | Keep Wolt and Deliveroo names and apps intact. | Ongoing |
| Customer tips go to the worker. | Tips first covered the guaranteed minimum pay. | Reversed |
Tactic: When choosing your second market, pick the one most likely to prove you wrong about the assumption your business depends on most.
Principle 4
Choose the category with the most nodes
Restaurants were not an obvious first category. Margins are thin, food is perishable, and customers are demanding. Xu and his co-founders picked restaurants after a quantitative comparison. There were about one million restaurant locations in America and about 200,000 grocery stores. People eat about twenty times a week. Fifteen percent of restaurants offered delivery, almost all of them pizza. For someone trained in operations research, these numbers pointed to the category with the most possible network nodes and the most frequent demand, which made it the best foundation for a dense logistics network.
Density was the real goal. A delivery network gets faster and cheaper as it gets denser, because drivers wait less and travel shorter distances between pickups. Restaurants were the best way to build density, and density was what would eventually let DoorDash carry groceries, hardware, flowers, and pet food on the same fleet. Xu described the category choice to Rabois as a bet "that restaurants would give the highest frequency of activity," adding: "It took us a few years to figure that out and kind of collect that data."
The lesson applies to any network business. The first market should be chosen for the infrastructure it builds, not for its own glamour or margin.
Tactic: Rank candidate wedge markets by frequency of use multiplied by number of locations, and launch where that product is highest, even if margins are worse.
Principle 5
Decompose the job to its twentieth step
Xu has said DoorDash identified about twenty discrete steps in a single delivery, from order placement to the handoff, and that "it really came from doing the work." When something broke, the team used the classic five-whys method to find the lowest-level cause. That habit grew into an experimentation system that by 2026 ran hundreds of thousands of tests a week, about 95 percent of which failed.
The physical world needs this because, as Xu has said, it "cannot be scraped." No ready-made dataset tells you how long a burrito waits on a counter in Fresno or how many apartment complexes in Phoenix have confusing gate codes. The data comes only from breaking the process into steps small enough to measure and then measuring them over and over. The 5 percent of experiments that work compound across the whole customer base, and that compounding is the advantage competitors cannot easily copy.
This explains why Xu told Fortune that winning was "less about, 'Did we choose this particular strategy or that?'" and more about "consistent execution over time." From outside, strategy is visible. Decomposition is not.
Tactic: Map your core transaction into at least fifteen discrete steps, assign a delay or failure metric to each, and run your next experiment on the worst one.
Principle 6
Make everyone do the delivery
For DoorDash's first year, every employee did deliveries and customer support every day. Later it became once a month, for everyone. Xu himself takes merchant support calls, does occasional deliveries, and reads support tickets daily, prioritizing the longest messages from Dashers. He then personally traces the failed order through internal tools.
This is more than a founder's ritual. It is an information system. Dashboards summarize, and summaries hide the edge cases where physical operations actually fail. A 2,000-word email from a frustrated driver contains information no metric captures: the specific sequence of algorithmic choices that sent someone to the wrong door at the wrong time. Xu has said this practice has directly produced product ideas, including warehousing and autonomous delivery.
It also shapes the culture. A company where the CEO is visibly handling the long tickets cannot easily become a company that discusses drivers as "resource nodes in a video game," the phrase from the hoax post that made Xu so angry. Whether it fully prevented that mindset, given the tip policy, is the subject of Principle 12.
Tactic: Require every employee, executives included, to complete your company's front-line job at least once a month, and have them write up one failure they saw.
Principle 7
Pay the refund nobody requested
DoorDash's most important early decision cost about 40 percent of its cash and was made in about fifteen seconds. Every order one night was at least an hour late. Nobody complained. The founders refunded everyone, baked cookies, and delivered them at five in the morning.
Xu uses the story to make a point about how values actually form in a company. Asked by Stripe's John Collison how DoorDash defined its values, he said: "It's actually my test, really, which is, what are the actions that are naturally occurring? What are the behaviors that are actually occurring inside of an organization? Then how do you actually articulate it." In his view values are not designed and then installed. They are observed after people have acted under pressure and then written down. The cookie night came before any values statement. The statement was a description of it.
The pandemic version was much larger. In 2020 DoorDash cut restaurant commissions by 50 percent, which Crucible Moments puts at a $100 million revenue hit, and ran national ads that pointed customers toward competitors. Spending trust before anyone asks for it is expensive, which is why it works.
Tactic: The next time a failure goes unnoticed by the customer, fix it anyway and compensate them, and record how fast your team decided; that speed is your real value.
Principle 8
Run software like a football playbook
Xu's favorite book is not about technology. Bill Walsh's The Score Takes Care of Itself describes a coach who wrote down procedures for everything, from a seven-yard post pattern to how the receptionist answered the phone. Xu took from it the idea that Silicon Valley mostly lacks: operational excellence as a core competence, not a back-office function.
His reasoning is about margins. Most consumer software companies, he told Y Combinator, have "enough gross margin" that mistakes do no serious damage. "If mistakes were to occur at DoorDash, DoorDash would be dead." When your margin is a few dimes per order, a delay is a loss. That changes how you hire, how you review performance, and how you spend a CEO's time. It is why the business review meetings went ahead on IPO day.
The approach is not exciting, and it is hard to copy because it is not exciting. Competitors with more money, including the Uber Eats division of a company worth several times DoorDash at the time, lost because they could not match the accumulated small improvements.
Tactic: Write a procedure for one front-line interaction your company handles thousands of times a week, down to wording and timing, and review it as seriously as a product launch.
Principle 9
Stay poor in your habits after you get rich
Xu grew up in a household where McDonald's was a once-every-two-months reward. He started DoorDash with no salaries and a bank-balance test. Those years shaped how he treats capital even now. "We had no choice but to focus on a path to profitability from the beginning," he told Fortune in 2020. That was not entirely consistent with his 2019 statement that the company was "choosing to invest in growth and delay profitability," but the underlying discipline was real.
The clearest statement of it is DoorDash's Q2 2023 shareholder letter, co-written with CFO Ravi Inukonda, which set out how the company evaluates new businesses: "a problem, preferably a large one, that we believe we can solve better than others," the availability of "capital, technology, and people," and "a path to a strong long-term return on investment." The letter included a worked example of a hypothetical "Business X" requiring $100 of investment over four years, then generating $10 of free cash flow a year, for a 17 percent internal rate of return. If you have the capital and people, five such businesses beat three. It is unusual for a public company to show its math that openly.
The discipline failed once, and Xu said so: pandemic hiring went beyond what the company could support, and 1,250 people lost their jobs. Even frugal habits weaken when money is abundant. What matters is noticing quickly.
Tactic: For every new initiative, write down the investment, the years until cash breakeven, and the implied return, and publish the framework to your board before the first dollar is spent.
Principle 10
Write "that's on me" in the first paragraph
The November 2022 layoff memo opened with a structure Xu has repeated elsewhere: start with the factors you control and take responsibility for them before mentioning the external environment. "As with all things, I want to start and discuss the factors in our control that led to today's announcement and take accountability for this decision." The macroeconomic explanation came second. "That's on me" came first.
On Harry Stebbings's 20VC in 2021, Xu described the leadership standard as giving all of the credit and taking all of the blame. The 2026 New York settlement statement, "Simply put, we screwed up," followed the same approach at the corporate level. Both are uncommon. Most layoff memos and most settlement statements are written by lawyers to admit nothing.
The memo's specifics showed the accountability was more than rhetoric: seventeen weeks of pay, a stock vest, health coverage, and a March 1 termination date chosen to give visa holders as much time as possible. Accountability without concrete support is just a performance. Xu gave both.
Tactic: In any bad-news communication, name your own decision as the cause in the first paragraph, and include at least one concession that costs the company money and serves the most vulnerable people affected.
Principle 11
Buy the local champion and keep its name
DoorDash's international expansion was built almost entirely through acquisition: Wolt for more than $8 billion in 2021, Deliveroo for $3.9 billion in 2025, plus Caviar, SevenRooms, Bbot, and others. The consistent rule is that local brands survive. Wolt kept its leadership and name. Deliveroo's riders and restaurants were told on the first day that "the Deliveroo app and products you know and love aren't going anywhere."
The reasoning goes back to DoorDash's founding thesis. Local commerce is local. The brand a London customer trusts, the app a Helsinki courier knows, and the term "High Street" itself are assets that a global logo would destroy. Xu's letter claimed the shared mission, "a focus on serving the local communities where we operate," while leaving the local identity intact. The company is global, and the brands each market sees remain local.
The constraint goes both ways. Brands kept separate are harder to integrate, and DoorDash withdrew from Japan and several other markets in 2026. Keeping the name is a bet that local trust is worth more than global efficiency. In delivery, so far, the bet has paid off.
Tactic: When acquiring a company whose value depends on local trust, commit publicly on day one to preserving its name and product, and put the integration behind the brand where customers do not see it.
Principle 12
Audit what your objective function cannot see
This principle is a warning. Xu is very good at optimization. His career is one constrained-optimization problem after another, and he described himself to Berkeley students as someone who sees every problem as an opportunity "to optimize within given constraints." But an optimizer works on the variables it is given, and the DoorDash record shows what happens with the ones it is not given.
!
The Human Half of the Problem
Selected legal and regulatory outcomes, drawn from public reporting.
| Year | Matter | Outcome |
|---|---|---|
| 2019 | Tips applied to guaranteed minimum pay | Policy reversed |
| 2020 | D.C. Attorney General tipping suit | $2.5M |
| 2021 | Dasher strike over tip visibility after Para cutoff | Strike |
| 2022 | Worker misclassification class action | $100M |
| 2023 | Seattle sick-time violations | $1.6M |
| 2024 | California sale of personal data | $375K |
| 2025 | New York tip-pocketing case | $16.75M |
| 2026 | NYC underpayment, incl. online-but-idle time | $131.5M |
Look at the pattern. A tip was treated as a fungible input to a pay formula rather than a gift to a person. Idle time was left out of the hourly guarantee because idle time did not produce a delivery. The largest part of the 2026 settlement, $83 million, concerned exactly that unpriced waiting. These were not obviously the result of malice. Xu's anger at the "desperation score" hoax looked genuine. They were the result of an objective function that counted some costs and not others. A greedy algorithm, the phrase Xu used approvingly in 2026, takes the best step it can see at each moment. It does not notice what is missing from its inputs.
The fix is the same detail-level attention that won the market, applied to the people the model treats as inputs. Read the 2,000-word Dasher email for what the algorithm made the driver absorb, not only for what went wrong with the algorithm.
Tactic: For every pricing or pay formula, list every unit of human time and money it treats as free, and have someone outside the optimizing team sign off on each one.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Tony Xu, Stanford GSB View From The Top, June 2021You're never as good or as bad as people say you are, so I think it's really important to be intellectually honest.
— Tony Xu, to the Los Angeles Times, January 2018As an entrepreneur, every day, every year, there are things that will make you increasingly uncomfortable. If you're willing to bear that uncertainty, I think you'll find a lot of rewards and satisfaction from it. Typically, it's when you experience discomfort that you grow.
— Tony Xu, on David Senra's podcast, March 2026When you see that you can lose someone's trust on one order, you realize you've got to earn it again the next day. There is no such thing as a set it and forget it mentality.
— Tony Xu, to Fortune, as quoted by Berkeley IEOR, 2025If you're in technology and you are not making improvements, you are actually decaying. Until it's over all of a sudden.
— Tony Xu, Stanford GSB View From The Top, June 2021If you're an aspiring entrepreneur, I would say find something you're obsessive about. Because the journey to building anything of meaning, I believe, will take years, maybe decades.
Maxims
- Follow before you build. Spend a day watching the work instead of surveying the worker; the booklet of refused orders will not appear on a form.
- Shrink the vision to a project. Early on, a company should be a few yes-or-no questions, not a manifesto.
- Expand into your doubt. The next market should be the one most likely to prove you wrong.
- Pick the wedge for density. Choose your first category for how many nodes and how much frequency it gives the network, not for its margin.
- Count the twenty steps. A process you have not broken into steps is one you cannot improve.
- Keep everyone on the front line. A company where the CEO reads the longest complaint is harder to corrupt than one that reads only the summary.
- Decide trust in fifteen seconds. Values are what a team does under pressure before anyone writes them down.
- Blame yourself first, in writing. Name your own decision before the macro environment, and back the apology with something that costs you money.
- Keep the local name. Local trust is an asset that a global brand can destroy.
- Price what the model treats as free. Every optimizer has costs it does not count, and the people on the other end of those costs will eventually present the bill.
