• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Patrick Collison thinks the lean startup playbook is losing its edge, a recap says

    A user's recap of the Stripe CEO's Startup School 2026 interview says he favors starting from a position nobody else is trying to occupy, rather than finding a small niche and iterating outward.

    GR
    2 Sources, ,

    TLDR

    A September 7, 2026 recap of Patrick Collison's Startup School interview describes his view that the lean startup playbook is losing its edge as more people have the tools to pursue niches. His alternative, according to the recap, is to start from a position nobody else is trying to occupy. The recap also says new business formation on Stripe was running just under 2x year over year—the largest relative jump the company had recorded. On AI, it describes his argument that knowing something yourself is much faster than asking a model. It says he still writes everything himself and has yet to read an AI-generated essay he found compelling.

    Combined views

    24.1K

    2 Sources, first seen 23d ago

    Combined views

    24.1K

    2 Sources, first seen 23d ago

    119 likes
    23d ago
    first seen 23d ago
    119 likes
    11 comments
    220 saves
    11 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    11 comments
    220 saves
    11 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    2 Sources

    @gokulrThe Crevice is Crowded @patrickc (Patrick Collison), Co-Founder & CEO, @stripe , interviewed by @harjtaggar (Harj Taggar) (@ycombinator, Startup School 2026) Summary: Collison thinks the lean startup playbook is losing its edge. Finding a small crevice and iterating outward worked when the internet was smaller and fewer people had the tools to work a niche. His alternative is to decorrelate aggressively and start from a position nobody else is trying to occupy, which is what most of the standout companies of the last 10 years did. The optimism is grounded in Stripe's own numbers: new business formation on Stripe is running just under 2x year over year, the largest relative jump the company has recorded. 1. Cognitive L1 Cache. Knowing something yourself is still much faster than asking a model for it. Collison borrows Jeff Dean's latency numbers that every programmer should know, where an L1 cache lookup and a network round trip differ by orders of magnitude. Knowledge has the same hierarchy. You can run far more round trips inside your own head than you can by typing a prompt or dictating one, and Stripe and the AI labs keep bidding up raw cognitive ability in their hiring. 2. The Writing Holdout. Collison still writes everything himself and has yet to read an LLM essay he found compelling. The utility function for good writing is very hard to define, which makes the domain hard to train against with RL. Models can prove serious mathematics and still miss what makes an essay land. Every tool now offers him pre-written replies, in Gmail and now WhatsApp, and he says he has sent zero of them in his life. 3. Twice Through The Trapdoor. Dropping out of college is reversible, and Collison has the unusual distinction of having done it twice. He left after his freshman semester to start a company with Taggar, went back to MIT for another year, then left again to start Stripe. As far as he can tell, nobody has ever cared (parents worry it will damage a reputation for life). If you enjoy college there is no harm in finishing, and if you do not, he puts the cost of leaving at de minimis. 4. Take The Under. Humanity has always had an affinity for millenarian models where society is about to be permanently transformed. Collison points to The Winged Gospel and the conviction after the invention of aviation that nothing would ever be the same. Aviation mattered enormously and still did not produce the sociological rewrite its excitable proponents predicted. He would take the under on this being the last couple of years to create a company. 5. Squirrels In A Trench Coat. Stripe worked because the problem was concrete and viscerally felt. Founders hallucinate customer problems all the time, imagining pain nobody would actually pay to remove. Payments passed the test loudly: everyone disliked the incumbent options, and the paperwork was antiquated and had to be filled out at a bank in person. It also looked absurd, 2 kids starting a financial services business before the word fintech existed, and Collison describes watching prospective partners quietly feel for the button to call security. 6. Production Users At Month Two. Stripe took almost 2 years from first commit to public launch and got away with it because real customers were using the product the whole time. First lines of code went down in autumn 2009. The first live production user, Ross Boucher at 280 North, arrived in January 2010, when all Stripe could do was charge a card. Boucher asked how to see his charges, so they built a dashboard, then how to refund a payment, so they built refunds, then whether he would ever receive his money, so they built payouts. 7. The Crevice Is Crowded. Buying Google Ads to find an underserved crevice and expanding outward from it is getting aggressively tilled. The internet is a much bigger place than when that doctrine was written, and the tooling to work a niche is available to everyone at once. Collison's alternative is to decorrelate aggressively and pick a divergent starting point nobody else is trying to occupy. Look at the standout companies of the last 10 years, the labs and Anduril among them, and most of them are anti-lean-startup. 8. What If You Succeed. Founders spend their planning on the possibility of failure and almost none on the converse. Collison's question is what happens when it works: you raise the money, you get the customers, you hire the people, and now you own this for 10 years, 17 years, maybe 30. Larry Ellison has been at Oracle for what will soon be half a century. Ask before the money lands whether you will enjoy the thing you are building, because success is the outcome you are signing up for. 9. Schlep Blindness Inverted. Payroll and money movement are the unglamorous work nobody starts a company to do, and in aggregate Stripe has been the opposite of a schlep. Every business Collison sees is an applied theory of how some corner of the world works. A new company amounts to a contrarian thesis on a counterfactual. He has never met a Stripe customer and thought that was boring. 10. The Google Question. The fear that a big incumbent will simply do your thing has a checkered track record. 20 years ago every founder was asked what happens if Google does this, and Google had the talent, the capital and the servers to do most of it. Human organizations are complicated, and prosecuting a hundred priorities at once produces interference between them that outsiders never model. Collison separates two risks: the labs deciding to expand into your space, and model capability improving enough to obviate the task itself. The second one is the one to forecast. 11. Formation Rate Doubling. New business creation on Stripe is running just under 2x year over year, the largest relative jump the company has ever recorded. The COVID surge in early 2020 inflected that growth rate to somewhere around 50%, so this is a different order of change. The quality numbers moved too. The median business is doing better than a year ago, the probability of any given business reaching $1M, $5M or $10M of revenue is improving, and time to revenue for companies incorporated through Atlas is falling. 12. Status Quo Risk Flipped. Enterprises now buy from startups because standing still has become the dangerous option. The old objection was that a startup is unvalidated and might not exist in 2 years, so the CIO would not take the meeting. Companies today are spring-loaded to try things, driven by a real terror of being left behind with antiquated ways of operating. YC companies are signing meaningful contracts inside the batch, which is why Collison doubts the story about AI concentrating the economy into a few hands and expects many thousands of winners.

    2 Sources

    @gokulrThe Crevice is Crowded @patrickc (Patrick Collison), Co-Founder & CEO, @stripe , interviewed by @harjtaggar (Harj Taggar) (@ycombinator, Startup School 2026) Summary: Collison thinks the lean startup playbook is losing its edge. Finding a small crevice and iterating outward worked when the internet was smaller and fewer people had the tools to work a niche. His alternative is to decorrelate aggressively and start from a position nobody else is trying to occupy, which is what most of the standout companies of the last 10 years did. The optimism is grounded in Stripe's own numbers: new business formation on Stripe is running just under 2x year over year, the largest relative jump the company has recorded. 1. Cognitive L1 Cache. Knowing something yourself is still much faster than asking a model for it. Collison borrows Jeff Dean's latency numbers that every programmer should know, where an L1 cache lookup and a network round trip differ by orders of magnitude. Knowledge has the same hierarchy. You can run far more round trips inside your own head than you can by typing a prompt or dictating one, and Stripe and the AI labs keep bidding up raw cognitive ability in their hiring. 2. The Writing Holdout. Collison still writes everything himself and has yet to read an LLM essay he found compelling. The utility function for good writing is very hard to define, which makes the domain hard to train against with RL. Models can prove serious mathematics and still miss what makes an essay land. Every tool now offers him pre-written replies, in Gmail and now WhatsApp, and he says he has sent zero of them in his life. 3. Twice Through The Trapdoor. Dropping out of college is reversible, and Collison has the unusual distinction of having done it twice. He left after his freshman semester to start a company with Taggar, went back to MIT for another year, then left again to start Stripe. As far as he can tell, nobody has ever cared (parents worry it will damage a reputation for life). If you enjoy college there is no harm in finishing, and if you do not, he puts the cost of leaving at de minimis. 4. Take The Under. Humanity has always had an affinity for millenarian models where society is about to be permanently transformed. Collison points to The Winged Gospel and the conviction after the invention of aviation that nothing would ever be the same. Aviation mattered enormously and still did not produce the sociological rewrite its excitable proponents predicted. He would take the under on this being the last couple of years to create a company. 5. Squirrels In A Trench Coat. Stripe worked because the problem was concrete and viscerally felt. Founders hallucinate customer problems all the time, imagining pain nobody would actually pay to remove. Payments passed the test loudly: everyone disliked the incumbent options, and the paperwork was antiquated and had to be filled out at a bank in person. It also looked absurd, 2 kids starting a financial services business before the word fintech existed, and Collison describes watching prospective partners quietly feel for the button to call security. 6. Production Users At Month Two. Stripe took almost 2 years from first commit to public launch and got away with it because real customers were using the product the whole time. First lines of code went down in autumn 2009. The first live production user, Ross Boucher at 280 North, arrived in January 2010, when all Stripe could do was charge a card. Boucher asked how to see his charges, so they built a dashboard, then how to refund a payment, so they built refunds, then whether he would ever receive his money, so they built payouts. 7. The Crevice Is Crowded. Buying Google Ads to find an underserved crevice and expanding outward from it is getting aggressively tilled. The internet is a much bigger place than when that doctrine was written, and the tooling to work a niche is available to everyone at once. Collison's alternative is to decorrelate aggressively and pick a divergent starting point nobody else is trying to occupy. Look at the standout companies of the last 10 years, the labs and Anduril among them, and most of them are anti-lean-startup. 8. What If You Succeed. Founders spend their planning on the possibility of failure and almost none on the converse. Collison's question is what happens when it works: you raise the money, you get the customers, you hire the people, and now you own this for 10 years, 17 years, maybe 30. Larry Ellison has been at Oracle for what will soon be half a century. Ask before the money lands whether you will enjoy the thing you are building, because success is the outcome you are signing up for. 9. Schlep Blindness Inverted. Payroll and money movement are the unglamorous work nobody starts a company to do, and in aggregate Stripe has been the opposite of a schlep. Every business Collison sees is an applied theory of how some corner of the world works. A new company amounts to a contrarian thesis on a counterfactual. He has never met a Stripe customer and thought that was boring. 10. The Google Question. The fear that a big incumbent will simply do your thing has a checkered track record. 20 years ago every founder was asked what happens if Google does this, and Google had the talent, the capital and the servers to do most of it. Human organizations are complicated, and prosecuting a hundred priorities at once produces interference between them that outsiders never model. Collison separates two risks: the labs deciding to expand into your space, and model capability improving enough to obviate the task itself. The second one is the one to forecast. 11. Formation Rate Doubling. New business creation on Stripe is running just under 2x year over year, the largest relative jump the company has ever recorded. The COVID surge in early 2020 inflected that growth rate to somewhere around 50%, so this is a different order of change. The quality numbers moved too. The median business is doing better than a year ago, the probability of any given business reaching $1M, $5M or $10M of revenue is improving, and time to revenue for companies incorporated through Atlas is falling. 12. Status Quo Risk Flipped. Enterprises now buy from startups because standing still has become the dangerous option. The old objection was that a startup is unvalidated and might not exist in 2 years, so the CIO would not take the meeting. Companies today are spring-loaded to try things, driven by a real terror of being left behind with antiquated ways of operating. YC companies are signing meaningful contracts inside the batch, which is why Collison doubts the story about AI concentrating the economy into a few hands and expects many thousands of winners.