Hacks

In the early days, book distributors required a minimum order of 10 books. We didn't have the sales volume yet to afford buying 10 books every time a customer ordered one, and we couldn't hold inventory.

We found a loophole: the system only required you to order 10 books, not receive them. We would order the one book our customer wanted, plus nine copies of an obscure book about lichens that we knew was permanently out of stock. The distributor would ship the one book, backorder the other nine, and we bypassed the minimum quota.

Before Facebook, I took an art history class at Harvard called "Rome of Augustus." I hadn't studied at all, and the final exam required us to identify hundreds of historical art pieces.

Two days before the exam, I built a simple website displaying all the art pieces and sent it to the class mailing list, claiming it was a "study tool." My classmates immediately flooded the site, filling in the historical significance of every single piece. I essentially crowd-sourced my final exam studying to the rest of the class and passed.

To launch our CRM, we needed to punch above our weight against the industry giant, Siebel Systems. When Siebel held their massive user conference in San Francisco, we hired actors to pose as protesters outside the venue.

They carried signs demanding "The End of Software" and handed out our marketing materials. We even hired a fake news crew to "cover" the protest. The police eventually showed up to manage the crowd, which drew actual local news coverage. We hijacked our competitor's million-dollar event for the cost of a few actors.

When we first launched, getting users was a slow, agonizing process. I used to go into the Apple Store in Palo Alto, casually walk down the aisles of display computers, and change the homepage on every single Mac and iPad to Pinterest.com.

Then I'd stand in the back and watch people interact with it. It was unscalable, manual retail hacking, but it got us our earliest active users.

We didn't know if people actually wanted food delivery in Palo Alto, and we didn't have the time or money to integrate with restaurant POS systems. We literally went into local restaurants, took physical paper menus, uploaded PDFs of them to a basic website, and put my personal cell phone number at the top.

When a customer called to order, I would drive to the restaurant, buy the food with my own credit card as a normal customer, and deliver it. We hacked the restaurant partnerships by just pretending they already existed.

When I transferred to the University of Pennsylvania, my roommate Adeo Ressi and I needed a way to pay for our housing and tuition. We rented a massive, cheap, unfurnished 10-bedroom fraternity house.

On the weekends, we would buy cheap kegs, cover the windows with black trash bags, and turn the house into a massive, unlicensed nightclub. We charged a $5 cover and had up to 500 people show up in a single night. One night of partying paid for our entire month's rent.

When we built our very first commercial relational database, we knew that enterprise buyers are deeply suspicious of brand new, untested software. Nobody wants to risk their company's data on a "Version 1.0."

So, we simply named our very first release "Oracle Version 2." It completely bypassed the psychological hurdle of enterprise procurement. Customers assumed Version 1 had already been tested and refined by somebody else.

IBM needed an operating system for their new personal computer, but we didn't actually have one to sell them. Instead of building one from scratch, we found a local programmer named Tim Paterson who had built something called QDOS (Quick and Dirty Operating System). We bought the complete rights to QDOS for just $50,000.

Then, we turned around and licensed it to IBM. But the real hack was the contract: we refused to sell it to IBM exclusively, retaining the right to license it to other computer manufacturers, which effectively made us the controllers of the entire PC revolution.

When I started the personal styling service in my apartment, I didn't have wholesale vendor accounts or clothing inventory. I would go to local boutiques and retail stores, buy a bunch of clothes on my personal credit card at full retail price, and ship them to my early test customers.

Whatever they didn't want, I would physically carry back to the retail stores and process as a standard customer return. I hacked a multi-million dollar supply chain using standard store return policies.

I applied to Y Combinator as a solo founder, but Paul Graham explicitly told me that YC strongly prefers co-founders. I was given an ultimatum: find a co-founder in a couple of weeks or lose the spot.

I went to MIT, hung out in the computer science lounges, and started chatting with a guy named Arash Ferdowsi. After a two-hour conversation, he agreed to drop out of MIT and marry our professional lives together just so we could hack the YC admissions requirement.

To prove the hypothesis that people would buy shoes on the internet without trying them on first, I walked down to the local Footwear Etc. store in San Francisco. I asked the manager if I could take photos of his shoes. I posted those photos on a rudimentary website.

When someone bought a pair, I walked back down to the store, bought the shoes at full price, and went to the post office to ship them. I validated a billion-dollar market without holding a single shoebox of inventory.

In 2009, our New York listings weren't getting booked. We realized the problem was that hosts were uploading terrible, low-lit photos using flip phones. We didn't try to build a better photo-uploading tool; we hacked the trust system directly.

We rented a $5,000 camera, flew to New York, and went door-to-door to our hosts' apartments, pretending we were "professional photographers sent by the company." We took beautiful photos, the listings started converting instantly, and revenue doubled in a month.

I was in my twenties, trying to get to the British Virgin Islands to see a girl, but my flight out of Puerto Rico was cancelled due to a lack of passengers. I was stranded. I called a local charter company and found a private plane for $2,000. I didn't have the money.

So, I grabbed a small chalkboard, wrote "Virgin Airlines: $39 to the BVI," and walked around the terminal to all the other stranded passengers. I sold every seat, paid for the charter, and flew for free. It was my first airline.

We had literally zero marketing budget. So I took our $500 limit and printed thousands of high-quality stickers of our alien mascot, Snoo. I traveled around the country and put them everywhere — on streetlamps, bathroom stalls, subways, and signs. I placed them near the offices of tech journalists.

People started thinking Reddit was this massive, ubiquitous underground movement, and the press started covering us just to figure out what the alien was.

I built the prototype for Instacart in a month, but I missed the Y Combinator application deadline by two months. I managed to get a partner meeting through a warm intro, but they rejected me. I needed to show them the product actually worked.

So, I opened my own app, ordered a six-pack of Maca root beer, and routed the delivery directly to YC partner Garry Tan at the YC headquarters. When it arrived, Garry called me and said, "Alright, it works. You're in."

When I was 16, I got a summer job selling newspaper subscriptions to the Houston Post. Cold calling was terrible. I realized the people most likely to buy subscriptions were people undergoing major life events — specifically, newlyweds moving into new homes.

Instead of knocking on random doors, I hired a friend to go to the county courthouse and copy down the names and addresses from recently issued marriage licenses. I sent personalized letters to those specific addresses and made $18,000 that summer.

When we launched, we needed women to be the ones initiating the platform. I bypassed digital ads entirely and physically went to the University of Texas at Austin. I bought hundreds of yellow roses and handed them out to girls on campus, telling them to download Bumble.

Then, I walked into the fraternities and told the guys, "All the girls on campus just downloaded this app, you need to get on it right now." It seeded our entire initial user base.

For years, I told the press that the idea for Netflix came to me because I was charged a $40 late fee by Blockbuster for returning Apollo 13 a few days late. It was a perfectly crafted origin story that made us look like the scrappy consumer champions fighting the evil corporate giant.

In reality, the idea was born out of a desire to create a direct-to-consumer e-commerce model like Amazon, but "I hated late fees" hacked the PR narrative perfectly.

When we were trying to secure our first major line of credit to manufacture the Apple II, we invited investors to our "office," which was still just a chaotic garage. To look successful, we deliberately parked my expensive Porsche and Wozniak's sports car around the corner out of sight.

We didn't want the bankers to see the cars and think we were reckless kids flush with cash. We presented ourselves as starving, hyper-focused engineers, and got the loan.

The health insurance brokerage industry was old and slow. Brokers would sell a company health insurance and take a 5% commission every year, doing almost nothing. We bypassed their entire sales system. We built amazing HR software and gave it to companies completely for free.

The only catch was that they had to make us their "broker of record." We used free software as a Trojan horse to legally steal millions of dollars in recurring insurance commissions.

We realized that traditional advertising was too expensive and didn't create the network effects we needed for a payments platform. So we hacked the economic system of user acquisition.

We simply deposited $10 directly into the account of anyone who signed up, and gave them another $10 for every friend they referred. We literally bought our first million users. It cost us millions in venture capital, but it established the monopoly we needed to survive.

Selling enterprise software is a nightmare because you have to convince a stubborn Chief Information Officer (CIO) to buy it. We decided to bypass the corporate bureaucracy entirely. We offered free, personal Box accounts directly to lower-level employees.

They started using it to share files because it was better than the company intranet. Once thousands of employees at a company were actively using Box, the CIO had no choice but to buy the enterprise license to regain security control.

When I was building the early VR prototypes in my garage, custom lightweight screens didn't exist. The only screens small enough and high-res enough were inside modern smartphones, but phone manufacturers wouldn't sell just the screens to a kid.

So I tracked down the Asian supply chains that manufactured replacement screens for broken cell phones. I bought them out of surplus stock, disassembled them, and hot-glued them into ski goggles.

We needed to get college students to use our carpooling service, but we had no budget for campus ambassadors. Instead of setting up a boring booth, I bought a massive, furry, neon pink mustache suit. I would literally run around college campuses in the suit, handing out flyers.

It was so ridiculous and disruptive that campus security would chase me, which only drew more of a crowd and made the students think we were a rebel brand.

To get the initial critical mass of attractive users on the app in Los Angeles, my co-founder and I threw a massive, exclusive birthday party at a high-end mansion. We invited the most social, well-connected students at USC.

But we instituted a strict door policy: to get past the bouncer, you couldn't buy a ticket — you had to show that you had the Tinder app downloaded on your phone. We got 400 highly active local users in a single night.

In the early Stanford days, we needed massive amounts of computing power but couldn't afford expensive server racks. Instead of buying industrial metal housing, we went to a toy store and bought massive bins of Lego bricks. We built our server racks out of Legos.

It was completely modular — if we needed to add a new hard drive, we just snapped more Legos onto the top. It bypassed thousands of dollars in hardware costs.

I wanted to translate the entire internet, but hiring professional translators would cost billions. I realized millions of people wanted to learn a new language for free.

So, I built a system where the "exercises" users were completing to learn Spanish or French were actually sentences pulled directly from untranslated websites. By learning a language, our users were secretly doing free, crowd-sourced translation work for us, which we could then sell to CNN and Buzzfeed to fund the company.

When we were negotiating the sale of Viaweb to Yahoo in 1998, the Yahoo executives were lowballing us in a conference room. I knew we had to show strength, but I'm not a naturally aggressive negotiator.

So, I literally just stood up, packed my bag, and started walking toward the door without saying a word. The physical act of walking away panicked them. They immediately caved and met our price of $49 million before I reached the hallway.

I needed to fund the first prototypes of my wrist-strapped camera, but I had no investors. I moved into my Volkswagen van and drove up and down the California coast.

I bought cheap shell beads from a supplier in Bali for $1.90 each, strung them onto belts, and sold them out of the back of my van to surfers for $20 a piece. I hacked my initial seed round using a 1,000% markup on beach jewelry.

When we announced zero-fee stock trading, we didn't launch an app; we just launched a waitlist. But we hacked the psychology of queuing. When you signed up, we showed you your exact number in line (e.g., "You are #14,352"). Below that, we put a button that said, "Invite a friend to move up in line."

People became obsessed with their physical "status" in the queue. They spammed their social networks just to drop their number, getting us 1 million signups before writing a line of trading code.

When we first launched "Picaboo," adults thought disappearing photos were pointless. We realized our target market was high schoolers, but we couldn't run ads to them. So we literally went to a local mall in Los Angeles and handed out flyers to teenagers.

But we pitched it specifically as a tool to pass notes in class without teachers being able to confiscate the evidence, and to share answers on tests. Framing it as a tool for school rebellion sparked our first viral loop.

In our first month, we had barely any traction. The problem was that professionals didn't want to invite their entire address book because it felt like spam.

To hack this social friction, I looked at the system of "advisors" in Silicon Valley. I personally invited high-status individuals — successful VCs and founders — and told them the platform was an exclusive network. Once the elite were on it, middle-management eagerly imported their entire contact lists just to get a chance to connect with them.

We were a tiny startup competing against established software like Quicken. We had no money for Google Ads. I noticed that personal finance bloggers had highly engaged audiences but made almost no money. I hacked their monetization.

I sponsored their blogs directly for a few hundred dollars a month. In exchange, they put a massive "Mint" badge on their sites. It made our tiny company look like a ubiquitous, heavily-funded financial institution overnight.

When we launched our 24/7 lifecasting camera, nobody cared. We needed press. We found out Yahoo was hosting a massive, exclusive media event in San Francisco. We didn't have press passes.

So, I just walked right into the venue wearing my ridiculous head-mounted camera rig, acting like I belonged there as part of the "new media." The physical absurdity of my setup confused the security guards enough that they let me in, and I spent the night interviewing the actual journalists.

I wanted to start a co-working space, but landlords wouldn't rent an entire floor to a guy with no money. I found a landlord in Brooklyn with a partially empty building.

I negotiated a deal where I didn't pay rent upfront; instead, I promised to physically subdivide the space myself, rent out the desks to freelancers, and split the profits with him. I hacked the commercial real estate system by essentially acting as his unpaid property manager in exchange for equity in the lease.

When we applied to Y Combinator, I was still a high school student in Ireland. The final interviews were in person in California. I had my high school graduation exams — the Leaving Cert, which dictates your entire college future — during the exact same week.

I hacked the logistics by flying from Dublin to San Francisco, doing the 10-minute YC interview with Paul Graham, and immediately boarding a plane back to Ireland, studying on the flight so I could sit for my final exams the next morning.

I wanted to get into Y Combinator, but I was a solo founder and didn't know anyone in the network. I needed references. I searched the internet for every YC alumni I could find and cold-emailed them.

But instead of asking for a favor, I subject-lined the emails: "YC S12 Batch - Meeting." I acted as if I was already practically accepted and just seeking advice. The presumed status got them to open the email, meet with me, and eventually write the recommendations that actually got me in.

I built a peer-to-peer lending platform, but we faced a chicken-and-egg problem: lenders wouldn't put money in because there were no borrowers, and borrowers wouldn't request loans because there was no money.

To prime the pump, I quietly maxed out my own personal credit cards and used the cash to personally fund the very first batch of loans on the platform. I used my own debt to create the illusion of a functioning financial marketplace.

I was an indie musician who needed to launch a crowdfunding platform, but getting press for a "new tech tool" is impossible. I knew the internet loves spectacle.

So, I spent every dollar I had building a giant, physical set piece out of Millennium Falcon toys and programmed industrial robots to play instruments alongside me. I released a viral music video. Only at the very end did I mention, "By the way, I built a site called Patreon." I hacked tech PR by burying it inside an entertainment stunt.

I finally convinced Neiman Marcus to stock my new undergarments, but they put them in the back corner of the hosiery department where nobody walked. I knew they would drop me if I didn't get sales immediately.

So, I went to the local hardware store, bought a bunch of display racks, walked into the Neiman Marcus stores dressed nicely so employees assumed I was corporate, and physically moved my products to the front checkout registers myself. Sales skyrocketed.

It started because Jim McKelvey lost a $2,000 sale. He's a glass blower in St. Louis; a customer wanted to buy a faucet, only had Amex, and Jim didn't have a way to take it. He called me — we'd known each other since I was a teenager.

The first prototype we built was a 1.6-millimeter headphone jack soldered to a guitar pick that read magnetic stripes through the iPhone's microphone input. Visa wouldn't certify it. Apple wouldn't approve a payment app from two unknowns. The card networks said nothing this small could be a real point of sale.

So we just carried iPhones into coffee shops, food trucks, and farmers' markets and demonstrated that the phone the merchant already had could replace the cash register they couldn't afford. By the time the networks paid attention, we had real volume going through the readers.

Janus and I had built Kazaa, a peer-to-peer music-sharing network that the music industry had spent two years suing out of existence. The architecture, though — a self-organising, server-less overlay that could route data through millions of nodes without any central infrastructure — was a thing of beauty.

When we started Skype in 2003 we didn't build a new networking stack from scratch. We took the Kazaa engine and used it as the foundation for voice. Calls were routed peer-to-peer over the same self-healing mesh that had once routed Metallica MP3s.

Skype scaled to a hundred million users on three engineers because the hard work — the network — had already been built and battle-tested before the company existed.

In 1985 we launched PostScript: a page-description language that let any computer print anything to any printer at any resolution. The trojan horse was that we gave it away. We licensed PostScript to Apple for the LaserWriter at near zero, and to every other printer manufacturer the same way.

We made our money on the typefaces — Adobe Type 1 fonts — and on the only software that could really exploit PostScript at the time, Adobe Illustrator. Once PostScript was the printing standard of the world, anyone who wanted to design or publish anything had to come to us.

We did the same play in 1993 with the PDF: free Acrobat Reader to seed the format, paid Acrobat tools to anyone who actually wanted to make PDFs. Both formats are now infrastructure.

By 2007 the streaming engine worked beautifully. The catalogue did not exist. Universal, Warner, Sony, and EMI controlled almost every song on Earth, and each one had to be persuaded individually.

I refused to launch in any market until all four had signed. I sat in their offices in London, New York, and Stockholm. I rewrote the same pitch fifty different ways. The deal we eventually struck wasn't just licensing fees: each of the four took a piece of equity in Spotify itself.

We launched in October 2008. The first day all four labels' libraries went live, the catalogue was about three hundred times the size of iTunes.

In 1996 our investor Tim Draper sat us down and said: "put a P.S. at the bottom of every outgoing email." We argued about it for weeks. We didn't want to come across as desperate. He insisted.

The line we eventually shipped was: "P.S. I love you. Get your free email at Hotmail." The link was hot. Every email a Hotmail user sent was an advertisement for Hotmail to whoever they were emailing.

In 18 months we went from zero to twelve million users. Microsoft bought us for $400 million in late 1997 — at a time when our marketing budget was still under $50,000.

When we sold WhatsApp to Facebook in 2014, the cheque was for nineteen billion dollars. The thing we cared about most wasn't the cheque — it was a clause in the deal.

The clause was a written guarantee that WhatsApp would never carry advertising and never mine its users' messages for ad targeting. Mark Zuckerberg signed it. The general counsel of Facebook signed it. Every executive in the room signed it. We had it bound into the merger agreement.

Jan had grown up in Soviet Ukraine where the postman opened your mail. He would not build a company that did the same thing under a friendlier name. Years later, when the clause started bending, both of us walked.

Our underwriters at Morgan Stanley wanted to price Netscape at $14 a share. The night before the IPO, on August 8, 1995, demand was so heavy they raised the offer price to $28. The next morning the stock opened at $71 and closed the day at $58.

The hack wasn't the price. The hack was going public at all. Netscape had not made a single dollar of profit. We were giving the browser away for free, with no obvious revenue model beyond enterprise server software that nobody was buying yet. Every IPO before us had at least claimed a path to profitability.

By proving that growth — measured in users, not earnings — could be priced as the unit of value, we set the template for every internet company that came after us.

For years, every email a free Mailchimp user sent through our system carried a small line of text in the footer that said "powered by Mailchimp," linked back to a sign-up page.

Multiply that by a few billion emails a month. The free users weren't customers — they were a distribution channel. They paid us in backlinks every time their reader scrolled to the bottom.

We never spent a dollar on advertising. We never raised venture capital. We never built a sales team. The bottom-of-the-email link was the entire engine.

In 2006 my cousin Nadia in New Orleans was struggling with maths. I tutored her remotely from Boston using Yahoo Doodle. As more of her cousins joined, my voice was giving out. I started recording the explanations on a $20 graphics tablet I'd bought at Best Buy and posting them as unlisted YouTube videos. I shot them at night in a cupboard at home so I wouldn't wake my daughter.

The cupboard turned out to matter. The audio was bad — but the video had no face, just a voice and a digital chalkboard, and it turned out that anonymity was a feature. Students who would never raise their hand in class would replay the same minute fourteen times.

In 2010 Bill Gates said in an interview that he was using my videos with his own kids. I quit my hedge-fund job that month and made Khan Academy a non-profit.

In August 2011 Ryan and I launched a single web page. The page had a JavaScript console embedded on the right side and instructions on the left side, telling the user to type things into the console. There was no login, no payments, no database, no signup. There was nothing on the back end at all.

We posted it to Hacker News with a one-line description. Within seventy-two hours we had 200,000 users — more than every other coding-education site combined. Y Combinator emailed us; we were in the next batch within the week.

The product was a JavaScript console with a tutorial wrapper. The hack was that nobody had thought to put the lesson and the runtime on the same page.

In November 2010 I needed to know whether anyone wanted a tool that scheduled tweets. Instead of building it, I built a single landing page with two screens.

Screen one had a "Sign up" button and a tagline. The button led to screen two: a fake pricing page with three plans. Click any plan, and you'd hit a third screen — "Hello! You've caught us before we're ready" — that collected your email and ended the journey.

The first screen validated demand. The second screen validated willingness to pay at a specific price point. I had the answer to both questions in a week, with about 120 signups, before I had written a single line of code. Only then did I start building the actual product.

For most of the 2000s our company was a tiny Amsterdam outfit nobody had heard of. Most of our spend was on Google AdWords. We outbid Marriott, Hilton, IHG, and every chain on the planet on their own brand keywords — "Marriott London," "Hilton Paris," "Holiday Inn New York" — and on every conceivable city pair.

The arithmetic worked because our conversion rate was three to five times the chains' direct sites. We could afford to pay more for clicks because we made more from each click. Year after year of compounding bid wars, the ad pool got more expensive for everyone, and we kept being the highest bidder.

By the time the chains realised what was happening they had ceded the entire paid-search front of the funnel. Today Booking.com is one of Google's largest single advertisers on Earth.

From day one we built Costco around a hard rule: nothing in the warehouse, ever, would be marked up more than 14% over wholesale. Most retailers run 25% to 50%. We chose 14% because it was below the threshold at which most consumers would bother to comparison-shop.

The other half of the rule was that we would not advertise. Ever. No TV, no radio, no print, no billboards. The only marketing was the membership card itself.

Both rules forced us to make our money on the membership fee, not the merchandise. The membership renewals — about 90% globally — became the entire profit. We could pay our cashiers nearly twice what Walmart did because we weren't subsidising margins on the items they were ringing up.

In 2002 I sent an internal memo to every engineering team at Amazon. Five rules. One: all teams will henceforth expose their data and functionality through service interfaces. Two: teams must communicate with each other through these interfaces. Three: there will be no other form of inter-process communication allowed. Four: it doesn't matter what technology they use. Five: all service interfaces, without exception, must be designed from the ground up to be externalisable.

I ended the memo with a sixth point: anyone who doesn't do this will be fired.

The point of the memo was internal modularity. The accidental side effect was that, four years later, every internal Amazon service was already designed to be externally rentable. We launched Amazon Web Services in 2006.

In 2015 we founded OpenAI as a 501(c)(3) non-profit on the conviction that AGI was too important to be developed by a profit-seeking company. By 2019 we'd hit a wall: training the models we believed in required tens of billions of dollars, and no non-profit could raise that kind of capital.

We solved it with a structural hack. We created OpenAI LP — a 'capped-profit' subsidiary. Investors could put in capital and earn returns, but the returns were capped at 100x the initial investment. After that ceiling, every dollar would flow back to the non-profit board.

Microsoft's $1 billion in 2019 was the first investment under this structure. Then $10 billion. Then a multi-tens-of-billions extension. The ceiling exists on paper. The world that built around it does not really know what to do with it.

In late 2022 we tried to convince Microsoft to ship the AI features we wanted built directly into VSCode. They were polite about it; nothing happened. We were a four-person company trying to land changes in Microsoft's flagship developer product. It was never going to work.

So we forked it. We took the open-source VSCodium build of VSCode, rebranded it Cursor, and rebuilt every layer of the editor that needed to be AI-aware. The fork was technically permitted — VSCode's core is MIT-licensed — but no one our size had been audacious enough to do it.

By 2024 Cursor was the fastest-growing developer tool in history. The hack was that the world's most popular IDE was open-source, and we were willing to ship a separate version of it to ourselves.

Newsletter creators were told for a decade that paid subscriptions don't work — that ad-supported media was the only model. We needed to prove otherwise, and fast.

So in 2020 we created Substack Pro. We offered established journalists six-figure cash advances against future subscription revenue, with the catch that we kept their subscribers if they didn't earn out. Matt Yglesias took $250,000. Andrew Sullivan took half a million. Each defection from a legacy newsroom was its own press cycle.

Most of the writers earned out within a year, but it didn't matter — by then the floor had collapsed under traditional newsroom contracts, and every editor in New York knew Substack was paying.

In 2017 we were two Brazilian founders trying to get a corporate credit card for our YC startup. Every issuer required a personal guarantee — a US credit score, a social security number, a personal liability for anything the company spent. We had none of those. We were 21 years old and had been in the country for six months.

So we built the card we needed. Brex underwrote credit based on the company's bank balance, not the founder's personal credit history. We legally couldn't have done it as a bank, so we partnered with a small Utah bank and issued the card under their licence.

Within a year every YC company we knew was using Brex. The hack was that our entire underwriting model came down to reading a single API field — the company's cash position — that no incumbent bank had bothered to look at.

In 2016 we open-sourced Next.js — a React framework — under the MIT licence. It cost us nothing to give away, and the only revenue model the framework would ever have was the hosting infrastructure built around it.

The math was: every developer who picked up Next.js had a problem at some point that wanted to be solved by a managed deploy platform. Vercel was that platform. We let the framework spread for free, and when it crossed ten million weekly downloads, we owned the infrastructure question of where any team that adopted it was going to host.

Today Next.js powers TikTok, OpenAI, Notion, GitHub, Hulu, and Patreon. A non-trivial slice of those teams runs on Vercel.

In 2018 Sketch was the dominant design tool — Mac-only, $99 a licence, with a perpetual update fee. Universities trained designers on Sketch because it was already industry standard.

We did one thing: we made Figma free, forever, for students and educators. Free as in $0, with no upgrade prompts, no licence checks, no fine print. Universities switched curriculums within two years.

The next generation of designers showed up in industry already fluent in Figma, demanding their employers support it. Sketch had to compete on price. Adobe had to launch a free Figma clone (XD). We had spent essentially nothing on the play, because students don't have money to spend anyway.