Part IThe Story
The Background Color of the Top Frame
Sometime in 2004, in a house in Ottawa that did not belong to him, a young German programmer tried to change a background color and could not.
He had paid for a custom design for his snowboard shop. It was a clean layout built on cascading style sheets, the "new fangled 'web standards'" that the more fastidious corners of the internet were starting to care about, and he tried to pour it into Yahoo Stores, then one of the respectable ways for a small merchant to sell things online. Yahoo would barely let him alter the background color of the top frame. Six years later, answering questions for Signal v. Noise, the blog of the software company 37signals, he still could not write the sentence without three exclamation points in parentheses: "(!!!)". He had already tried Miva. He had tried osCommerce. "Truth be told," he wrote, "all those systems made my skin crawl because of how bad they were."
The circumstances made the frustration worse. Tobias Lütke was in his early twenties and had come to Canada for a woman. Employers had offered him programming jobs, and then he discovered that the visitor's visa he had entered on did not permit him to take any of them. So he kept the job he already had: remote work for a startup back in Germany, writing financial back-end software in Java, "the driest of the dry kind of stuff." He disliked remote work and found it isolating. He believed he had burned out on programming itself. The snowboard shop was meant to be the way out of code, not further into it. It would be a retail business with real margins in the right niche, using his technical skill as leverage, a setup that "would eventually let me earn my money, without having to do any programming."
Then, "in a moment of utter disgust," an instant message arrived from a friend about "this Ruby on Rails thing that David just released."
He downloaded it and fell in love with it. Within a day or two he had worked out that the problem was never programming. It was "programming in uninspired programming languages." Two months later the shop, Snowdevil, was running on store software he had written himself. The 2004 season was good enough to turn a profit, and somewhere in that winter the itch moved from the snowboards to the thing that sold them.
The later retellings each furnish the founding with a different room. In one it is an apartment, in another a garage, in a third an Ottawa coffee shop, in a fourth the home of his girlfriend's parents. The legend has been sanded and lacquered by twenty years of podcast hosts and keynote bios. What stays constant across the versions are the small, unglamorous facts: a frame, a color, three exclamation points, and a man named David.
Twenty-two years on, the frame had grown. In 2025 the company that came out of that disgust, Shopify, reported $11.56 billion in revenue and $1.23 billion in net income. In 2024 it processed $292.3 billion in transactions for more than five million customers, among them Tesla, LVMH, Nestlé, PepsiCo, Red Bull, Whole Foods Market and Hyatt. In January 2026 it co-announced, with Google, an open standard for artificial-intelligence agents to find products and complete purchases on their own. Its founder still writes code and still ships small tools to GitHub at hours that are hard to account for. In July 2026 he also told the internet that retirees living on pensions should lose the vote.
A biographer could connect the exclamation points to the pensioners with one straight line, and the line would be wrong. The path between them runs through Koblenz and a ponytailed craftsman at Siemens, through a decade of living in someone else's house, through an unwanted promotion, a pandemic bet and its unwinding, and finally onto a racetrack in Florida. It loops more than once.
By the Numbers
The Frame, Twenty-Two Years Later
$11.56BShopify full-year revenue, 2025, up 30% year over year
$292.3BTransactions processed on the platform in 2024, 57% in the U.S.
5M+Customers using Shopify's commerce platform
2 monthsTime Lütke spent building Snowdevil's store software in Rails
40%Voting power held by Lütke under the 2022 founder share
11,600 → 7,600Shopify headcount, 2022 to 2025
$1,000,001His 2019 Team Trees donation, one dollar more than Elon Musk's
A Miserable Student With a Schneider CPC
He was born in Koblenz, in the Rhineland, in 1980 or 1981 depending on which registry you trust. The German national biography says 1981 and Wikipedia says 1980, and neither seems to think the question matters much. What matters is the machine. When he was six (or seven, by another account) his parents gave him a Schneider CPC, the German-branded version of the Amstrad Colour Personal Computer. He played Pac-Man and Space Invaders on it obsessively. Games were scarce, so he started taking apart the ones he had. By eleven or twelve he was rewriting their code, learning from computer magazines whose exercises he typed in before he knew that what he was doing had a name. "I got so fascinated with my computer," he told an interviewer decades later, in a grey hoodie and his trademark flat cap, "it was more interesting than anything else and it was certainly more interesting than school."
A German columnist later reported that the boy played fifteen or sixteen hours a day. His mother eventually admitted, "We were very worried about him back then." His parents sent their introverted son to a therapist because he seemed to be hiding entirely behind the screen. He had a reading and spelling disability. He would tell Der Spiegel in 2017, in a headline that does not hedge, "Ich war ein miserabler Schüler": I was a miserable student. After the tenth grade he left the Görres-Gymnasium with the lesser certificate, the Realschulabschluss. In Germany that is a quiet verdict on a child's future.
The child was not lacking in commercial instinct, only in interest in what the classroom was selling. "I was also a kid that always ended up figuring out what everyone else was into," he told Rick Rubin, "so I would buy a lot of, I don't know, Magic Eye books or something and then sell them to my classmates." The arbitrage served the hardware. As a teenager he donated blood for money at a German hospital, since his type, O positive, was in demand, and used the proceeds to buy his first Nvidia graphics card. Years later he told Jensen Huang that he had bled for his product. Huang, Lütke recalled, looked at him and said, "you'd be surprised how often I hear stories like this."
At sixteen he entered the apprenticeship system that Germany runs alongside its universities. Four days a week he worked at Siemens and one day he attended trade school, the Carl-Benz-School in Koblenz. There he found the first of the men who would rhyme with him. His supervisor, his Meister, was named Jürgen. In Lütke's telling Jürgen weighed around two hundred pounds, had long gray hair tied in a ponytail, and rode a BMW motorcycle to work while everyone around him wore suits. Jürgen refused the dress code, ignored the company's standard programming language, and called out bad ideas in public. He was, Lütke said, "so competent that no one could impose any kind of dress code on him." Lütke later wrote an essay about the arrangement, "The Apprentice Programmer," and he still talks about it in the vocabulary of manual trades: he learned programming "like essentially a blue collar craftsmanship."
There was a second, more distant master. In the mid-1990s John Carmack, the programmer behind Doom and Quake, published his working notes in plain-text
.plan files. They were dated, mundane and relentless. In February 2021, rereading them, Lütke tweeted that a series from "(I think?) August 1996" had "ended up making a huge impression on me as a kid." He added: "Not only is it a master class in systems design it's also so... driven?"That question mark carries real admiration with no condescension in it, and it is worth keeping in mind. Lütke did not learn that building was a moral act from a seminar. He learned it from the logs of a man shipping code at three in the morning, and from a ponytailed craftsman whom no one could make wear a tie.
The House on Someone Else's Money
The cliché almost makes the next part embarrassing. He met Fiona McKean, a Canadian, on a snowboarding trip to Whistler, or "a ski trip to B.C.," as he told students at Waterloo. They lived in Germany for a year. When she went home to finish her studies, he followed her. He reasoned the way a teenager reasons with a map. "I heard about Silicon Valley and California," he told Rubin. "I knew where it was on a map at least, I had never been there, but Canada was closer to Silicon Valley than Germany. So I'm like, that seems like the right direction. Let's go." He arrived in 2002 or 2003, and the accounts disagree. He imagined he would end up in San Francisco. He has said that, growing up, he wanted to move to North America, and that once he got there he decided Canada was "the U.S. for Europeans." He once considered deliberately changing his accent. He did not, and the Financial Post still describes it as "noticeable."
Fiona McKean would become a diplomat. Her parents, Bruce and Dale McKean, were retired federal public servants in Ottawa, a city whose whole economy is organized around government service. For roughly a decade the young couple lived in the McKeans' house at low cost while Lütke built the thing. He drew no salary. Bruce McKean put in early seed capital. In the leanest stretch, according to one account, friends and family covered payroll. The couple were still in the McKean house when they were expecting their first child.
The Ottawa Citizen described those years, with gentle sarcasm, as "the archetypical life of a millennial couple — or at least its urban legend version." When the paper visited again, around 2014, the family had moved into a new house of its own on Ivy Crescent in New Edinburgh. It was 2,500 square feet, built of brick and cedar by the architect Andrew Reeves on a lot 32 feet wide, where a partially gutted cottage had stood empty for six years. Building it involved a small civic fight. The city had changed its parking rules, so the couple had to press it to reopen a rear public laneway. Some neighbours were unhappy, because they had quietly appropriated parts of the public right-of-way for themselves.
It is a small story and a revealing one. The founder wanted access through a lane that belonged to everyone and that a few had fenced off for their own use. He did not get it by asking nicely. He got it by insisting the rules be applied.
He did work long hours in those years, and he did not romanticize them. "There was a time for 16-hour days early in the process," he told a journalist in 2012. "People get a bit crazy about this. But in fields where you use a lot of creativity to solve challenges, the best people might have five concentrated hours of work in a day … If I can get five concentrated hours out of someone, that's all I can ask, myself included."
When 37signals approached Shopify in 2010 to be profiled in a series called "Profitable & Proud," about companies that had never taken outside funding, the interview ran and readers objected. Shopify had an angel investor. The editors pulled it from the series with an apology. The interview stayed online under a new heading, a self-sufficient company caught with one foot on borrowed ground. In fairness, the borrowed ground had always been there. The house had been borrowed too.
The Software Was the Product
In the canonical telling, a merchant saw Snowdevil's store and asked to license the software, then another merchant did, and the founders noticed that the store was worth more than the snowboards. Shopify launched in June 2006. As The Shopify Story, excerpted in the Ottawa Citizen, puts it, it all started with snowboard sales.
The details wobble when you press on them. One retelling in the Toronto Star, by venture investors writing about the art of the pivot, sets the decisive moment on "a grey Ottawa morning in 2008." A Toronto entrepreneur calls to ask whether he can buy the software behind Lütke's "flailing" shop, and then, in a line built for a slide deck, "Everyone wanted the platform. Nobody wanted the snowboards." That account has Snowdevil dying in the financial crisis. Another has the snowboard business fading simply because snowboards do not sell in summer. Shopify's own Wikipedia entry has the platform launched in 2006, two years before that grey morning. I do not know which version is true. Probably all of them are slightly true, in the way founding stories usually are once enough people have told them for long enough.
The undisputed record is more interesting than the myth. Lütke joined the core team of Ruby on Rails in 2004, the year of the instant message. Rails was David Heinemeier Hansson's framework, and Lütke was among the earliest outsiders trusted to shape it. He wrote and released open-source libraries that became infrastructure for strangers. ActiveMerchant handled payments. Liquid, a templating language created at Shopify and in use since 2006, let merchants and designers change their storefronts without being able to break them. DelayedJob, a database-backed job queue "Extracted from Shopify," still sits on his GitHub profile with about 2,200 stars. Typo was a weblog engine.
Liquid in particular reads in hindsight like a direct answer to the Yahoo Stores humiliation. It is a language that lets anyone change the background color of the top frame, and many other things, without being handed the keys to the engine room. He built for others the freedom he had been denied, with exactly the guardrails he had needed.
Rails also gave him what he called the "flywheel push." He told Waterloo students in 2015 that Ruby was "so difficult but so much more correct." He tends to talk about software the way other people talk about ethics.
For the first three months Shopify was free to users while Lütke finished the code. Then it started charging. By 2010 he could report about 6,000 active merchants in more than 60 countries, who had together sold over $100 million in products in the previous year. The company was named Ottawa's fastest-growing company that year. "We have been profitable since 2008," he wrote. The sentence that probably mattered most came earlier in the same interview, said almost in passing: "Shopify is our idea of how an online store should be run."
In June 2009 the company opened an API and an App Store so that outside developers could build on Shopify and sell what they built to its merchants. One of those developers, Klaviyo, launched there in 2012 as an email-marketing tool. A decade later Shopify invested $100 million in it, and it eventually became a public company of its own. The platform was being built to feed businesses it would never own.
Lütke has his own aphorism for the pivot, delivered in 2021: "Almost every good decision starts as a bad one first." He has another, harder one about where products come from. He told an interviewer that he had found the existing store software so bad that he decided "to do something about it just to spite them."
An MBA in a Couple of Weeks
In 2008 Shopify had about ten employees and roughly $60,000 a month in revenue. Then its chief executive left.
That was Scott Lake, the friend who had co-founded Snowdevil and then Shopify with Lütke and Daniel Weinand, the man who handled the business while Lütke handled the code. Lake departed after the founding years. Asked later by the Globe and Mail what role Lütke played, he answered in seven words: "He is the brain trust. Tobi is Shopify."
Lütke did not believe that at the time. By his own account he had paid no attention to the business side of the company. His first instinct was to hire a replacement, so he flew to Silicon Valley to look for a CEO and to talk to venture capitalists. The meetings went badly in a useful way. The investors used terms he did not know and asked about his business in ways he could not answer. At night he went back to his room and looked up what he had been asked, so that he would be ready for the next meeting. "When I took over as CEO," he said later, "I had to essentially get an MBA in a couple of weeks." He came home without a CEO.
On the same trip, investors told him something he could not decode. "It must be so great to be in Canada because you're next to Waterloo," they said. "I didn't know what Waterloo was," he admitted to an audience at the University of Waterloo years later. "I had to look it up." By 2025, Shopify was hiring roughly 40 percent of its interns from that university.
It took other people to talk him into the job. John Phillips, an early angel investor in Toronto who had paid for designers in the company's first stretch, persuaded him to stay on as CEO. Peter Fenton of Benchmark helped him put together his first investor deck, though Shopify never took Benchmark's money. Lütke settled on a governing idea that sounds obvious now and was not obvious to him then: "the person who runs the product should run the business."
Then he did what an apprentice does and read the manual. Andy Grove's High Output Management became the text he returns to most. "Basically, at the end of the day," he told Tim Ferriss, "making business is an engineering exercise, which is brilliant for me because that actually made the whole thing about becoming CEO significantly less scary to me." The next book corrected the first. Robert Cialdini's Influence, he said, "was just the most mind-bending book you can imagine, because it essentially taught you all the ways humans are flawed and influenceable, and how, yes, computers are predictable, but once you make things for people you need to go into storytelling. Which was news to me, frankly."
— Tobias Lütke, How I Built This with Guy Raz, 2019You know, wonderful things about computers is when you tell them what to do, they'll keep doing it and they'll do it forever until you stop. Turns out humans are not like that.
He built a private apparatus for self-correction because no one else would provide one. "Introspection is all you've got," he has said. "You're not going to get a lot of feedback as CEO." He describes his days as if recorded: "I perceive my days as if there is a camera behind me. I can replay my day and learn from it." He hired a coach and pushed his executives to do the same. He wrote rules that sound like code comments: "Only the people who use a process are allowed to come up with a process." He asked for a list of accomplishments and objectives every Monday. He framed new initiatives as thirty-day trials, so that failing was cheap.
He also learned some of it from StarCraft, which is less silly than it sounds. On a gaming podcast he spent an hour on how the real-time strategy game had taught him about attention. You have only so much of it, and you lose either by staring at a single battle while your production stalls or by watching everything at a distance while a fight you should have won slips away. Some threats are carriers and most are interceptors, he explained, and if you destroy the carrier the interceptors die with it. He once offered an internship to a student on the strength of his ranking as a professional StarCraft player. In February 2021 Shopify started its own esports organization, Shopify Rebellion, and signed the former world champion Byun Hyun-woo and the Canadian player Sasha "Scarlett" Hostyn.
"Ignorance is the single biggest driver of entrepreneurship," he said once. "It was two years before it ever occurred to me that this might not work."
Asked by Guy Raz how he was doing as CEO, he answered at first with something close to a confession: "I'm not good. I mean, my team was very, very patient with me. There was a lot to learn. It's a very subtle job."
Arming the Rebels
The money arrived in the measured, increasing amounts that institutions use. In December 2010 Shopify raised a $7 million Series A from Bessemer Venture Partners, FirstMark Capital and Felicis Ventures at a $20 million pre-money valuation. At that point merchants were moving about $132 million a year through the platform. In October 2011 it raised $15 million more. In August 2013 it launched Shopify Payments with Stripe, so that merchants no longer needed a third-party gateway, along with a point-of-sale system for physical stores. In December 2013 it raised $100 million.
That same autumn, in September 2013, BlackBerry, then still the great Canadian technology story, announced it would cut 4,500 jobs worldwide after shedding more than 7,000 in the two previous years. Shopify had 300 employees, and Lütke expected to double that within a year. He anticipated hiring many of the people BlackBerry let go. When IBM held layoffs in Ottawa, Shopify set up a recruiting booth outside the IBM building on the same day. It was a small, cold, efficient piece of opportunism, and talent moved from one era to the next. The Financial Post later summarized the result: BlackBerry became "a shadow of its former self," and Shopify replaced it as Canada's tech darling.
In February 2014 Shopify launched Shopify Plus for large merchants. Revenue that year was $105 million. In November the Globe and Mail named Lütke its CEO of the Year under a headline that also described him: "Our Canadian CEO of the year you've probably never heard of." He was thirty-three. "It took about 10 years to become an overnight success," he said. That spring the Heartbleed security bug had been detected minutes before it became public, and Shopify scrambled to patch its merchants' stores. "That was a crazy, crazy couple of hours," he remembered.
— Tobias Lütke, The Globe and Mail, 2014I always loved retail. I love the ideas behind it. I think small-business retail is one of the areas where capitalism works so wonderfully well.
On May 21, 2015, Shopify listed on the New York and Toronto stock exchanges, raising more than $131 million at US$17 a share. It ended the first day well above that, having briefly popped about 65 percent. A wire photograph shows Lütke on the floor of the NYSE with Fiona and their three children, Fraser, Sam and Tristan. Two hours after trading began, the Globe asked how it felt to be worth $100 million more than when he had gone to bed.
"It is not something that motivates me, so I don't particularly care about money," he said. "I care about working on interesting problems." Then, with a self-awareness that is very much his: "I know that sounds like something like a PR agency would put together, but … this is the way I think about it in reality." Then came the economics: "Given the success rate, if you want to get wealthy, entrepreneurship is a horrible way of doing it. There are significantly easier ways of doing it." And the physics, which is how he usually thinks about anything: "In my worldview, time is energy that you can invest in things, and money is energy that you can invest."
He described the roadshow as a refusal to perform. "We're just telling people what kind of company we built and invite them to invest and we're not selling, we're just explaining." The company, he said, "was never built to win business plan competitions."
Four months after the IPO, Amazon closed its Webstore product for merchants and named Shopify the preferred provider for migrating them. The largest store in the world had handed its small-store customers to the company that built small stores.
Lütke had already chosen his metaphor. Shopify would be the rebel army against Amazon's web empire, and he would be "arming the rebels." In July 2016, on stage in Montreal, Kara Swisher called Amazon a "Borg" and asked whether any retailer could compete with it. "I don't think you can," he said, a startling answer from a man who sells to retailers. His point was that they did not have to. Amazon could be one channel inside a Shopify dashboard among many. What Shopify owned was underneath every channel: "The back office of an online store is surprisingly complex, it's not really a technology play." He closed with a stray bit of data that delighted the room. The best-selling item on Shopify at that moment was Pokémon-inspired jewelry.
The rebellion attracted enemies. In October 2017 the short seller Andrew Left of Citron Research published a report claiming that Shopify overstated its merchant count and calling it a "get-rich-quick" scheme in breach of FTC rules. Left had made his name accusing companies, loudly and on the record, and on the day of the report Shopify's stock fell more than 11 percent. Lütke called him a "short-selling troll." Left came back in April 2019 and predicted the shares would fall by half within twelve months. In January 2020 he told his investors he had closed the position. No FTC investigation followed.
In 2017 Fortune, putting Lütke on its 40 Under 40 list, wrote that the "low-key German-born programmer has fought to give Shopify's 500,000 merchants all of the attention." Ottawa was coming to the same conclusion. On November 5, 2018, the Governor General gave him the Meritorious Service Cross for his work in building the Canadian technology industry. The citizen who had once been forbidden to work in the country was now being decorated by it.
Five or Even Ten Years
In the first months of the pandemic, with physical shops closed by order, Shopify's market value briefly passed the Royal Bank of Canada's and it became the most valuable listed company in the country. That had happened to Canadian technology companies before, and the precedents were Nortel and Research in Motion, which is not reassuring company. "We can't help on the front line, we're not doctors and nurses," Lütke said in his accent, over video, from home. "But there is a second wave coming, which is the economic crisis, which is going to derive from the humanitarian crisis. We can't solve it, but we can hopefully help." In May 2020 Shopify announced that most of its workforce would go permanently remote. The office culture built in Ottawa was given up almost overnight.
The windfalls were large. Shopify had partnered with the payments company Affirm, and when Affirm went public on January 13, 2021, Shopify's 8 percent stake was suddenly worth $2 billion. In August 2021 Shopify crossed $2,000 a share on the Toronto exchange for the first time, and Lütke's stake reached US$13.3 billion, which probably made him the richest person in Canada. His holdings consisted of just under 7.5 million Class B shares carrying ten votes apiece, about 80,500 ordinary Class A shares, and roughly 580,000 options. Since the start of 2020 he had sold about US$490 million of stock. In Germany a columnist observed that Shopify was now worth more than any company on the DAX, built by a Koblenz dropout almost nobody at home had heard of.
The Financial Post noted that he drew a salary of one dollar. The Toronto Star noted that the Globe's ranking of the highest-paid CEOs had him at the top one year, with $205 million in total compensation. Both are true, as is common at that altitude.
Inside the company, the strain was visible. In early 2021 about half of Shopify's C-level executives left. Some former employees told Business Insider about tension over how the company had handled questions of race and social justice. Lütke answered in a memo to staff. Shopify was a team, not a family, he wrote, and it would focus on its business rather than on fixing social ills. People outside the company who admired him for it praised the honesty. People inside who disagreed did not.
Then came the forecast. Lütke believed the pandemic had permanently pulled the future forward and that e-commerce's share of retail would "leap ahead by five or even 10 years." He staffed for that belief and doubled headcount from early 2020 to more than 10,000. He bought for it too. On May 5, 2022, Shopify agreed to pay US$2.1 billion in cash and stock for Deliverr, a San Francisco fulfillment startup, on top of the warehouse-robotics maker 6 River Systems, which it had bought in 2019 for about $450 million. A software company was building warehouses, at least by proxy.
In June 2022 shareholders approved a special "founder share" for Lütke, giving him 40 percent of the company's total voting power, and confirmed him as both chief executive and chair. The same meeting approved a ten-for-one stock split. Control was at its height just as the forecast began to fail.
On July 26, 2022, Lütke told staff in a memo that about 10 percent of the company, roughly 1,000 people, would be gone by the end of the day, mostly in sales, accounting and recruiting. He was cutting "over-specialized and duplicate roles" and "some groups that were convenient to have but too far removed from building products." Then he wrote the sentence that people remember.
"Ultimately, placing this bet was my call to make and I got this wrong. As a consequence, we have to say goodbye to some of you today and I'm deeply sorry for that."
The stock fell 13.6 percent that day and closed at $40.69 in Toronto. Its split-adjusted record high, set the previous November, had been $222.87. That year the Ottawa Business Journal tracked Lütke falling from second to eleventh on Canada's rich list. Later he told an interviewer, "They were some of our hardest days."
The unwinding was quick. In May 2023 Shopify cut about 20 percent of its workforce and sold its logistics operations. Deliverr went to Flexport, in exchange for a 13 percent stake in Flexport, and 6 River Systems went to Ocado Group. TechCrunch reported that the layoffs had been mishandled in a way that made employees "feel like NPCs," non-player characters, a gaming insult that lands with particular force at a company that sponsors a StarCraft team. Laid-off employees filed a $130 million class action over severance. A company that had bought warehouses to compete with Amazon was again a company that made software, now with a minority stake in someone else's warehouses.
— Tobias Lütke, as quoted by Startup CFOCompanies are manifestations of their founders. Shopify is good at all the things I am good at and bad at all the things I am bad at.
A Principled Process
He has never lacked convictions. The problem is that he holds them in public, and the internet keeps the receipts.
The first major test came in February 2017, when a #DeleteShopify campaign demanded that the company drop the online store of Breitbart News. Lütke refused. His argument was procedural rather than ideological: removing a merchant over its views would mean enforcing a moral code, and he asked who would get to write that code. The position fit the founding. He had built Shopify on his own disgust at being told by Yahoo how much he was allowed to change. Now he had a platform of his own and declined to be the one setting limits.
The platform drew similar fights for years. In October 2018 The Logic reported that several groups designated as hate groups by the Southern Poverty Law Center were running stores on Shopify. In 2022 Media Matters and then the CBC criticized the company for hosting the store of the account Libs of TikTok. Lütke replied to an activist on Twitter on November 26: "Shopify has a published acceptable use policy and a principled process to apply it. Pressure groups on all sides try to influence it sometimes and CBC needs to see through that not amplify bad faith narrative." In November 2024 Bloomberg reported that a Shopify store was selling antisemitic merchandise, and that the company had dropped a ban on "hateful content" from its acceptable use policy that July. Business Insider reported that Shopify employees had found a noose emoji uploaded to the company Slack, and that the CEO had then shut down internal discussion. Corporate Knights ran a feature under a headline that told the story by itself: "How Shopify went from Canada's tech darling to anti-DEI."
In February 2025 Kanye West bought a Super Bowl advertisement for his Yeezy store. The store's only featured item was a white T-shirt printed with a swastika. Shopify's support staff were initially told to say "no comment" if merchants asked. Former Shopify executives criticized the company in public. About twenty-four hours later the store came down, and Shopify said the reason was the risk of fraud, not the swastika. Of every moment in the record, this one most exposes the gap between "a principled process" and the principle behind it.
That same month the United States imposed 25 percent tariffs on Canadian goods, and Ottawa retaliated. Lütke called the retaliation the wrong choice. "Hitting back will not lead to anything good. America will shrug it off," he said, and then gave Washington's case in its own terms: "Trump believes that Canada has not held its side of the bargain, and he set terms to prove that we still work together: get the borders under control and crack down on fentanyl dens." His chief operating officer, Kaz Nejatian, went further and said Canada "has turned a blind eye to being used as a training ground for foreign countries, gangs and terrorist groups." Two years earlier, in June 2023, Lütke had refused a Canada Revenue Agency request for six years of records on 121,000 Canadian stores and called it "low-key overreach." His relationship with the state that had once denied him a work permit was suspicious, transactional and occasionally openly contemptuous.
Then, in the last weekend of July 2026, a former TD Bank executive named Eric Thor posted a proposal on X. Votes would be weighted "proportional to the level of income tax you pay." People who paid no income tax would get no vote, and people paying $500,000 or more a year would get five. Thor was a banker floating a thought experiment of the kind bankers rarely sign. Lütke answered: "Good system." He said it would "reward productive people with leverage." Then he went further and addressed pensioners.
"Now you are a dependent and that means no voting, just like dependents under age. Enjoy the deal, let people with a stake in the future decide."
The Canadian Civil Liberties Association pointed out that Section 3 of the Charter guarantees every adult citizen the right to vote. Dennis Pilon, a York University political scientist, called the idea "the most classic anti-democratic set of ideas, that have typically been held by the most powerful." Corporate governance experts pointed to the founder share, under which a man holding roughly 6 percent of Shopify's economic value controls 40 percent of its votes. A system of weighted votes was not a theory to him. He already lived under one.
Back in the early years, in a house in Ottawa, a programmer with no work permit drew no salary for close to a decade. The house belonged to Bruce and Dale McKean, retired federal public servants. Bruce wrote one of the first checks.
Reflexive
On April 7, 2025, Lütke posted a screenshot to X with a short explanation: "I heard this internal memo of mine is being leaked right now, so here it is." The subject line was its thesis. "Reflexive AI usage is now a baseline expectation at Shopify."
The memo was nearly 1,300 words long and went to about 8,100 employees. Most of it read like the apprentice addressing the workshop. "Using AI well is a skill that needs to be carefully learned by… using it a lot." Employees should use the tools "as a thought partner, deep researcher, critic, tutor, or pair programmer." It brought back an old image from an earlier leadership memo, Shopify as a Red Queen race, "you have to keep running just to stay still," and made it sharper: "Stagnation is slow-motion failure. If you're not climbing, you're sliding." It set its own arithmetic: "In a company growing 20–40% year over year, you must improve by at least that every year just to re-qualify. This goes for me as well as everyone else." AI use would be added to performance and peer reviews.
The line the internet seized on was about hiring. "Before asking for more headcount and resources, teams must demonstrate why they cannot get what they want done using AI. What would this area look like if autonomous AI agents were already part of the team?" When someone asked whether candidates could use AI in Shopify interviews, he answered in five words: "has been for a long time."
Critics were plentiful and some were useful. A Forrester analyst noted that "you can't prove a negative." A communications consultancy argued that the memo "simmers with" frustration and lectures readers who, he implied, had already missed a hint. Commentators suggested that a company whose share price was falling would have struggled to justify new hires anyway. The headcount chart supports the cynics as much as the visionaries: 11,600 employees in 2022, 8,300 in 2023, 8,100 at the end of 2024, and 7,600 in 2025. More cuts followed into 2026, including in the partnerships team. Revenue moved the other way: $8.88 billion in 2024, then $11.56 billion in 2025, a 30 percent increase, with $3.67 billion in the fourth quarter alone and a $2 billion buyback announced alongside it.
There were still stumbles. On December 1, 2025, Cyber Monday, some merchants could not log into their admin portals or ring up sales on Shopify's point-of-sale terminals for most of the day, and the stock closed down 5.9 percent. Infrastructure is invisible until it fails, and then it is the only thing anyone notices.
In March 2025 Shopify moved its U.S. listing from the New York Stock Exchange to Nasdaq, and in May it joined the Nasdaq-100. In January 2026 Shopify and Google announced the Universal Commerce Protocol, an open standard intended to let AI agents discover products and complete purchases with any merchant. In September Meta joined Shopify's agentic storefront channels. The aim was to be the back office for buyers that are not people, which is the Swisher answer again, scaled up: whoever controls the storefront, Shopify keeps the plumbing.
The man running all of this still codes in public and does not hide it. Ben Horowitz asked him on stage why a company worth close to $100 billion had a CEO who kept writing software. His GitHub profile lists a local command-line search engine for notes and documents called
qmd, with about 30,000 stars. It also lists try, described as "fresh directories for every vibe," a Rust project called walgit, a disk-usage visualizer built for Omarchy, a Linux setup assembled by David Heinemeier Hansson, and a macOS menu-bar app for managing audio-device priorities. They are utilities a craftsman makes for his own bench.He told Tyler Cowen that he expects to be "terribly embarrassed" when he looks back at the companies of the 2020s. He has said elsewhere that he works "under the assumption that we have no idea how to build companies yet, and that 50 years from now people will look back at the companies of today." The confidence and the doubt sit side by side, and he does not seem to notice the contradiction. He told Cowen that Canada suffers from a "go-for-bronze mentality," and that much of Shopify's success came from "just me and my co-founder basically allowing everyone to go for world class." He cited Own the Podium, the national program behind Canada's strongest Winter Olympics, held on home snow in 2010, and that is the only Canadian institution he praises without reservation. He has also talked about skating to work in Ottawa, a man who can make even the winter commute sound like a competition.
Fourth in Class
In 2024 he started racing prototypes and won the amateur class of the HSR Prototype Challenge presented by IMSA, with three wins and five podiums in six races. On December 5, 2024, Era Motorsport announced its lineup for the following month's Rolex 24 at Daytona: Paul-Loup Chatin, Ryan Dalziel, David Heinemeier Hansson and Tobias Lütke, in the LMP2 class.
Heinemeier Hansson is the David from the instant message. He released Ruby on Rails in 2004, the framework that saved Lütke from his burnout and gave Shopify its "flywheel push." His company published the 2010 interview and then took it back out of "Profitable & Proud" because of the angel. He has long been one of the loudest voices in software for small, self-funded, opinionated companies. Like Lütke, he has recommended William Irvine's guide to Stoic philosophy, which he called "an easy way to learn about [Stoicism] and recognize why this resonates." Lütke calls Marcus Aurelius's Meditations "My most read book." These are two programmers who read the Stoics, now sharing a race car.
Endurance racing gives a CEO very little to manage. Drivers take turns in a single car through the night. No one gets to rewrite the rules mid-race, and no one gets to change the length of the lap. A Ben Horowitz joke from years earlier comes back: of all the CEOs he knew, Lütke was "probably the least snowboard kickback, like, let's hit the powder dude." He has proved less suited to the ski lodge than to the pit lane.
The No. 18 Era Motorsport car was an Oreca 07 with a 4.2-litre Gibson V8. At Daytona in January 2025 it completed 764 laps, finishing twelfth overall and fourth in class. Over the 2025 IMSA season Lütke finished fifteenth in the LMP2 standings with 1,647 points. His best result came at Petit Le Mans in Georgia, where the car finished third in class. It was bronze.
He kept going. In 2026 he moved to TDS Racing and took a podium at Elkhart Lake. In June, at the 24 Hours of Le Mans, sharing the car with Mathias Beche and Kévin Estre, he completed 355 laps and finished twenty-seventh overall, thirteenth in LMP2 and fifth in the Pro-Am subclass. He drives an Oreca on weekends and, by one account, prefers his bicycle to his Tesla.
The 2025 entry list for the LMP2 class includes a line for car No. 18 that reads, in the flat format these documents use:
18. David Heinemeier Hansson / Tobias Lütke / Kakunoshin Ohta
Two programmers on one line, separated by a forward slash. David first.
Part IIThe Playbook
Lütke's playbook is not a list of virtues. It is a set of habits formed by an apprenticeship, an immigrant's lack of standing, and a long period of dependence, then tested by success and by error. Some of what follows is admirable and some is cautionary. Several principles are both, which is typical of founders who hold their convictions this tightly. Each is drawn from the record in Part I. None of them is "work hard," since everyone already knows that one.
The Lütke Playbook
- 1.Let disgust write the spec.
- 2.Sell the tool you forged for yourself.
- 3.Learn the business like a codebase.
- 4.Own the back office, not the storefront.
- 5.Let the ecosystem grow fat on you.
- 6.Fine the stock price, don't fight it.
- 7.Grant permission to go for gold.
- 8.When you bet the company, sign the bet.
- 9.Subtraction is the founder's job.
- 10.Make the new tool a baseline, not a pilot.
- 11.Hire the grind, not the credential.
- 12.Control is a tool; you will be judged by where you point it.
Principle 1
Let disgust write the spec
Shopify was not born from a market map. It came out of a specific, documentable humiliation: being unable to change the background color of a frame in Yahoo Stores. Lütke has talked about creation as "a conversion of energy," in which passion, anger or dissatisfaction gets channeled into making something, and he said he built his own software "just to spite them." The early product's virtues map closely onto that grievance. Liquid gave merchants and designers the freedom to customize that Yahoo had denied, with the safety that makes freedom bearable.
Disgust makes a good spec because it is precise. A vague ambition such as "better e-commerce" produces vague software. "I could not change the top frame" produces a templating language. When a founder can name the exact moment a tool failed them, they also know the acceptance test the replacement must pass.
Tactic: Write down the three most specific moments your current tools humiliated you, and treat each one as a test case your product must pass before you ship.
Principle 2
Sell the tool you forged for yourself
The snowboards were not the business. They were the test environment. Lütke built the Snowdevil store in two months because he needed it, and other merchants wanted it for the same reason. His later formulation is founder-market fit: build for a problem you have, which "allows you to do high bandwidth customer development with yourself giving you an unfair advantage."
The non-obvious part is the willingness to give up the original identity. He had moved to Canada intending to be a retailer who no longer programmed. He ended up a programmer whose customers were retailers. The pivot meant admitting that his escape route had turned into his destination. "Almost every good decision starts as a bad one first," he said.
Tactic: Inventory the internal tools your company has built to run itself, and ask of each one whether a stranger would pay for it.
Principle 3
Learn the business like a codebase
Lütke did not choose to be CEO. Scott Lake left in 2008, the company had ten people and $60,000 a month in revenue, and Lütke's first move was to look for a replacement. He failed to find one, so he did what programmers do with an unfamiliar system: he read the source. He studied each investor meeting at night and turned to Andy Grove, whose book told him that "making business is an engineering exercise."
The method still matters more than the books. He built feedback loops where none existed: the imagined camera behind him, Monday lists, coaches for himself and his executives, and thirty-day trials. "Introspection is all you've got," he said. An engineer debugging a system that offers no error messages has to instrument it himself.
⏱
From Snowboard Shop to Platform
The decisions that built, stretched and narrowed Shopify.
2004
Lütke builds Snowdevil's store in Rails in two months; joins Rails core team.
2006
Shopify launches in June; Liquid templating language released.
2008
Scott Lake departs; Lütke becomes CEO; company turns profitable.
2009
API and App Store open on June 2.
2010
$7M Series A at $20M pre-money; $132M annualized transactions.
2013
Shopify Payments with Stripe; POS; $100M Series C.
2015
IPO on May 21 at US$17; Amazon picks Shopify for Webstore migration.
2016
Shopify Capital launches; over $5.1B advanced since.
2022
Founder share grants 40% voting; Deliverr bought for $2.1B; 10% layoffs.
2023
20% layoffs; logistics sold to Flexport for a 13% stake.
2025
"Reflexive AI usage" memo; revenue reaches $11.56B.
2026
Universal Commerce Protocol launched with Google.
Tactic: For your next unfamiliar role, keep a private log of every question you couldn't answer that day and close each one before the next meeting.
Principle 4
Own the back office, not the storefront
When Kara Swisher asked whether retailers could beat Amazon, Lütke said no, and then explained why that did not matter. The storefront could be Amazon, Facebook, Twitter, TikTok or, from 2026, an AI agent. "The back office of an online store is surprisingly complex, it's not really a technology play." Shopify's lasting position comes from absorbing the messy, formerly human processes of commerce: inventory, payments, taxes and orders. Customer interfaces change, and the plumbing stays put.
The proof is in the record. Amazon gave its Webstore merchants to Shopify in 2015. The Universal Commerce Protocol, co-developed with Google in 2026, extends the same logic to buyers who are software. Whoever controls the front door, the transaction still runs through Shopify's pipes.
⚖
Conventional Wisdom vs. the Lütke Approach
Where Shopify's choices diverged from the default playbook.
| Conventional wisdom | Lütke's approach | Mode |
|---|---|---|
| Fight the dominant marketplace for the customer. | Make the marketplace a channel inside your dashboard. | Scale |
| Hire a seasoned CEO when the founder is a programmer. | "The person who runs the product should run the business." | Scan |
| Move to Silicon Valley to raise and recruit. | Stay in Ottawa; recruit outside BlackBerry and IBM layoffs. | Scale |
| Tax your app developers to maximize take rate. | 0% commission on developers' first $1M. | Try |
| Pilot new technology in a small team. | Make AI use a baseline expectation in performance reviews. | Try |
| Moderate controversial merchants case by case. | Published policy, "principled process," minimal removals. | Risk |
Tactic: Identify the most tedious, error-prone process your customers run manually, and build there instead of on the surface where competitors crowd.
Principle 5
Let the ecosystem grow fat on you
The App Store opened on June 2, 2009, three years after launch, when most companies would still have been guarding every feature. By 2020, partners earned more than $230 million on the platform, and a typical merchant was running six apps. Klaviyo began there in 2012 and grew large enough for Shopify to invest $100 million in it in 2022. In June 2021 Shopify cut its commission on developers' first million dollars of annual revenue to zero.
The logic is not charity. Every app that solves a merchant's problem is a feature Shopify did not have to build, and a reason for that merchant to stay. The same reasoning applies to partners: the Affirm stake was worth $2 billion on IPO day, and the Flexport stake was the consolation prize for the logistics retreat. Lütke has been more willing than most founders to let other companies get rich around him, on the bet that the platform gains more from that.
Tactic: List the features on your roadmap that an outside developer could build better, then open the API and step aside.
Principle 6
Fine the stock price, don't fight it
After the IPO, Lütke worried that employees would watch the ticker instead of the work. His fix was small and a little absurd. Anyone caught checking the stock price during work hours had to buy Timbits for their team the next day. He admitted that he once had to pay up himself. The philosophy behind the joke is serious: "we are all here to work on something called the fair market value of a company, which is unknowable," while "the current stock price … is trying to guess at the fair market value but is completely unrelated."
The rule worked because it was cheap, social and enforceable, a cultural norm with a mild cost attached, not a speech. It sits alongside his candor with investors: Shopify had "a gas pedal," it could go off it and "be wildly profitable," and the trust it had built with the market let it keep accelerating.
Tactic: Pick the single distraction most corroding your team's attention and attach a small, public, slightly silly penalty to it.
Principle 7
Grant permission to go for gold
Lütke's sharpest cultural diagnosis is of his adopted country: "a go-for-bronze mentality." His remedy is mostly permission. "I think a lot of our success has been due to just me and my co-founder basically allowing everyone to go for world class," he told Tyler Cowen. "Ratcheting up ambition for a project is something that one has to do in a company in Canada." His model is Own the Podium, which led to Canada's best Winter Olympics.
Many leaders assume ambition has to be hired. Lütke's experience suggests it often has to be authorized. People in cautious cultures are frequently waiting for someone to say the large version of the plan out loud. The irony that he himself took bronze at Petit Le Mans does not weaken the principle. It just means he raced in a class where he was up against professionals.
Tactic: On your next project kickoff, explicitly name the world-class version of the outcome and ask the team what it would take, before anyone proposes the safe version.
Principle 8
When you bet the company, sign the bet
The pandemic forecast that e-commerce would "leap ahead by five or even 10 years" was wrong. Headcount doubled, Deliverr cost $2.1 billion, and the correction cost about 1,000 jobs in July 2022 and another 20 percent of the workforce in May 2023. What set Lütke apart was not that he avoided the mistake. Almost everyone in technology made some version of it. What set him apart was the sentence: "Ultimately, placing this bet was my call to make and I got this wrong."
This is the apprentice's ethic carried into the boardroom, the CEO as the person who "never get[s] to blame anyone ever again." Taking ownership does not repair the damage. The class action and the "NPC" headlines show how much harm remains even when the apology is sincere. Its value is that it closes off excuses, so the company can move straight to the correction instead of arguing over who was at fault.
♟
Bets and Unwinds
Major strategic wagers and how they resolved.
| Bet | Size | Outcome |
|---|---|---|
| Rails as the foundation (2004) | Two months of work | Scale Still core architecture |
| Open App Store (2009) | Platform control | Scale 10,000+ apps by 2024 |
| Affirm partnership (2020) | 8% stake | Scale $2B at IPO |
| 6 River Systems (2019) | ~$450M | Risk Sold to Ocado, 2023 |
| Deliverr (2022) | $2.1B | Risk Sold to Flexport for 13% stake |
| Pandemic headcount (2020–22) | 10,000+ staff | Risk Cut 10%, then 20% |
| Agentic commerce / UCP (2026) | Open standard | Try Google, Meta aboard |
Tactic: Before your next large bet, draft the memo you'd send if it fails, with your own name in the first sentence of the blame.
Principle 9
Subtraction is the founder's job
"The best thing founders can do is subtraction," Lütke told Shane Parrish. "It's much, much, much easier to add things than it is to remove things." The record shows how hard that is in practice. Shopify added a film studio, a Los Angeles flagship with a "genius bar," esports teams, warehouse robots and a fulfillment network. Then, under pressure, it removed most of the physical layer in a single year and went back to software.
The lesson is that subtraction rarely happens at the right time unless someone with authority insists on it. Additions arrive with a champion attached. Removals arrive with only a cost. In Lütke's case it took a 70 percent fall in the stock to force the cut. A founder who practices subtraction routinely, before a crisis requires it, avoids having to do it all at once and in public.
Tactic: Once a quarter, require each team to propose one product, process or meeting to kill, and kill at least one.
Principle 10
Make the new tool a baseline, not a pilot
The April 2025 memo did not invite experimentation. It said "Reflexive AI usage is now a baseline expectation." The mechanisms backed up the rhetoric: AI questions in performance and peer reviews, AI integration in monthly business reviews, and teams required to show why AI could not do the work before asking for more headcount. "Using AI well is a skill that needs to be carefully learned by… using it a lot."
The critics had a point. You cannot prove a negative, and a memo that lectures can deepen the anxiety it is meant to calm. The structural insight still holds. Optional pilots let organizations defer change indefinitely, while baseline expectations make change the default. Lütke has done this before. Rails was a bet on a tool most people thought was obscure, and he chose it because it was "so much more correct."
Tactic: For any capability you believe is essential, put it in the performance review rather than the training catalog.
Principle 11
Hire the grind, not the credential
Lütke left school after tenth grade, struggled with dyslexia, and learned his trade from a ponytailed Meister at Siemens. Unsurprisingly, he hires for demonstrated mastery over pedigree. He offered an internship to a professional StarCraft player because getting that good at that game requires grit under adversarial pressure. He set up a recruiting booth outside IBM on its layoff day. He hires about 40 percent of Shopify's interns from Waterloo's co-op program, the closest Canadian equivalent to the German apprenticeship model that formed him.
The deeper idea is that competence counts as character. Jürgen could ignore the dress code because his team delivered. Lütke's admiration for Carmack's
.plan files, "so... driven?", is admiration for visible, dated, accumulated work. A credential says what a person was once certified to do. A track record of grinding shows what they actually do.Tactic: In your next hiring round, add one question about something the candidate became excellent at outside of school, and weigh the answer seriously.
Principle 12
Control is a tool; you will be judged by where you point it
The founder share gives Lütke 40 percent of Shopify's votes on roughly 6 percent of its economic value. That kind of control made the 2022 and 2023 corrections, and the AI pivot, possible without a proxy fight. It also means every public statement he makes reflects on the company in a way an ordinary CEO's would not. The Breitbart defense, the Libs of TikTok tweet, the Yeezy takedown justified by fraud risk, and the July 2026 endorsement of weighted voting have all been read against that structure.
This principle is a warning more than an instruction. Concentrated control lets you act fast and absorb the consequences of being wrong. It also removes the protections that would normally soften how your views land. When the governance experts lined up the founder share next to his "Good system," they were not making an abstract point. A leader who argues that leverage should go to "productive people" will have his own leverage examined first.
Tactic: If you hold outsized control, write down the public positions you'd be uncomfortable seeing printed next to your cap table, and decide in advance whether they're worth it.
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
— Tobias Lütke, memo to Shopify staff, July 26, 2022Ultimately, placing this bet was my call to make and I got this wrong.
— Tobias Lütke, to The Globe and Mail on IPO day, May 21, 2015Given the success rate, if you want to get wealthy, entrepreneurship is a horrible way of doing it. There are significantly easier ways of doing it.
— Tobias Lütke, The Knowledge ProjectThe best thing founders can do is subtraction. It's much, much, much easier to add things than it is to remove things.
— Tobias Lütke, internal memo posted April 7, 2025Reflexive AI usage is now a baseline expectation at Shopify.
— Tobias Lütke, The Globe and Mail, 2014It took about 10 years to become an overnight success.
Maxims
- Disgust is a specification. The exact moment a tool humiliates you defines the test its replacement must pass.
- The shop can be scaffolding. Sometimes the business you start exists only to reveal the business you should run.
- Read the source. Treat an unfamiliar job like an unfamiliar codebase, one unanswered question at a time.
- Own the plumbing. Storefronts change hands while the back office stays put.
- Let others get rich on you. A platform is measured by the businesses it feeds, not by the features it hoards.
- Fine the distraction. A box of Timbits enforces focus better than a speech about it.
- Authorize ambition. In cautious cultures, people are often waiting for permission to go for gold.
- Sign the bet. Take ownership of the mistake first and the correction can begin sooner.
- Subtract before you're forced to. Cuts made early are cheaper than cuts made in public.
- Leverage invites scrutiny. Whoever holds the extra votes will have his own case examined first.
