Nobody in the Valley thinks about ice rinks.
That line sat with me longer than the rest of a conversation between Patrick Collison and Amjad Masad. Collison runs Stripe. Masad runs Replit. Someone asked, more or less, where Collison would look if he were starting over. He did not name an AI lab. He named the software nobody fashionable wants to touch.
I have heard versions of this advice for years. Vertical SaaS. Boring industries. Schlep blindness. What felt new was the timing. The distance from an idea to the first dollar has roughly halved. On Stripe’s numbers, the time to reach a million, ten million, even a hundred million in recurring revenue is about half what it was in the last SaaS boom. Twenty percent of new startups now charge a customer inside thirty days, up from eight percent in 2020. New business creation on the platform nearly doubled in a year, a bigger jump than the pandemic spike.
When building the thing was the scarce skill, the programmer won. When a working product can be stood up quickly, the scarce skill is knowing why the old thing is still tolerated. The domain expert gets the edge. A teacher who understood schools built MagicSchool on Replit and rode it toward a company later described as worth half a billion dollars. The insight was the teacher’s. The tools closed the gap.
Collison’s map has three drawers.
The first is the good version of a cludgy 1990s incumbent. For seven or eight years, when people asked him whether he was excited about drones or crypto or AI, his stock answer was vertical SaaS. The joke was the point. In a lot of industries the dominant tool is still a desktop product with a bad mobile app, no real collaboration, and nothing that resembles the software the rest of us use. It stays because the buyers are not startups and the problem is boring to the people who build software fastest.
Toast did this to the restaurant systems operators hated and could not leave. ServiceTitan did it for HVAC, plumbing, and electrical, and its founders had grown up around their fathers’ service companies. Veeva did it for pharma. Procore for construction. Clio for law firms. The pattern is visible once it has already paid.
The open version is less famous. Auto dealers still live inside CDK and Reynolds & Reynolds. Body shops split their week between an estimating tollbooth and a whiteboard. Independent pharmacies run on PioneerRx and QS/1. Small trucking companies dispatch from a 1990s transportation system. Wholesale produce houses still take orders off voicemail into a machine older than the web. Credit unions under a few billion in assets have a phone app wrapped around a core from the 1980s. Ask an owner what they hate most about their software, then ask what they could never turn off. The second answer is the product.
The second drawer is stranger. Not a better app. A whole sector software skipped.
Masad’s example was a founder in Europe, working alone, who built software for ice-skating rinks and was on the way to a multi-million-dollar business. Scheduling ice time, youth hockey, resurfacing, birthday parties, waivers. A real operating system. Rink software exists now. Bond Sports and ROLLER sell it. That does not retire the example. It shows how long a category can sit there before anyone in the usual places looks.
You find the same thing anywhere the answer to “what software do you use?” is Excel and a guy named Dave. Marinas, with slip reservations and haul-outs in a harbor master’s notebook. Septic pumpers and well drillers, route businesses with regulatory paperwork and almost no shared system. Livestock auctions. Tow companies, where impound inventory and lien sales are a legal workflow run from a fax line. None of these is a small life. Most of them are just invisible from a laptop in San Francisco.
The third drawer is the one Collison flagged as riskier. What is already popular with young people, and still low status.
Status is a filter. It keeps sophisticated competition out. By the time an activity is respectable, the window has usually closed. Sneaker resale was a forum hobby until StockX and GOAT. Streaming games was kids in bedrooms until Twitch. Daily fantasy was a dorm-room spreadsheet until DraftKings. Pickleball and run clubs were this pattern a few years ago. They are lifestyle brands now.
The still-embarrassed list is the useful one. Discord servers that are actually the business, for cards or exams or a local car meet. Virtual fashion and brainrot games on Roblox, which supposedly paid creators about $1.7 billion in the twelve months through June 2026, and which most serious operators still file under children’s distraction. AI companions, enormous in usage and persistently awkward to admit. The test is simple. If a serious person laughs when you name the buyer, you are early. If they ask for the deck, you are late.
I keep coming back to who this favors. Not the person with the cleverest idea about artificial intelligence. The person who already knows the rink, the classroom, the route, the trade. For a long time that knowledge was stranded. You could see the broken workflow and have no reasonable way to build the replacement. That gap has narrowed. It has not disappeared. Taste, trust, and the willingness to sit with a boring buyer still take time. But the old excuse, that only a certain kind of technical person could turn an observation into income, is weaker than it was.
Collison’s advice is easy to nod at and hard to follow. The valuable surface is whatever looks too small, too provincial, or too uncool for the people who can now build software fastest. And software building with AI tools has become almost supersonic.
Ice rinks are not the opportunity. They are the reminder. The opportunity is the thing you already understand that everyone else has decided not to see.