FIDJI SIMO: How’s everyone doing? Good?

PATRICK COLLISON: Fidji and I had dinner back a year or two ago, and Fidji was describing her management strategy and how it really hinges on a very radical French candor…

JOHN COLLISON: That we’re in for today.

PATRICK COLLISON: I was going to say. So when we were thinking about who best to moderate this Q&A session, we thought there’d be no better person.

FIDJI SIMO: Well, and I couldn’t refuse an invite to grill the Collison brothers. Thanks for having me, guys, and thanks to the audience for submitting absolutely awesome questions. Really, really excited to dive into it.

But first, just want to say that Instacart has been a partner with Stripe for more than a decade, and honestly, we couldn’t have grown our business as fast as we did if it wasn't for Stripe. I love, as these guys innovate, just to give you a quick anecdote, last year we came to them and we were like, “We absolutely want to launch FSA, HSA payments,” which is a little bit gnarly, lots of government rules packed into that, but it is like $113 billion stored into FSA, HSA and these guys were like, “Yep, we’re up to the challenge.” And we managed to launch on time for the holidays. It is just one of many examples of how working with this company has been awesome.

I guess I'm not doing such a good job grilling you guys, but it's going to come.

I want to start first with one question, not yet from the audience, but we're going to get there, which is, before Stripe in 2007, and most people don't know this, you created a company called Auctomatic which was really supposed to improve the experience for eBay sellers, which is kind of funny because in 2007 I was at eBay trying to improve the experience for eBay sellers.

PATRICK COLLISON: Oh, I never knew that.

FIDJI SIMO: Yes, it looks like we both failed, but anyway… I’m curious if you guys have learned any lessons from that first experience that you then applied to Stripe?

PATRICK COLLISON: There's an obvious lesson where it was in the course of building Auctomatic that we started to get exposed directly ourselves and also for the first-time hearing stories from others about the difficulties with accepting online payments. So it's where we discovered the actual Stripe use case. But the other thing we learned that I think was very influential was… at Auctomatic, everything was so elegantly and carefully built. We wrote all the software in Smalltalk.

FIDJI SIMO: What does that mean?

PATRICK COLLISON: It’s a language.

JOHN COLLISON: It’s an obscure programming language, even to those who appreciate obscure programming languages.

PATRICK COLLISON: And we didn’t use any silly mainstream, relational database. We wrote our own distributed object data store, and Auctomatic had this incredibly fancy technology. We had this thing for saving a VM image so that if someone encountered a problem, we could restart it later and get back into the exact web request, all this stuff. I think Auctomatic got to 50 customers.

FIDJI SIMO: Amazing. Got to start somewhere.

PATRICK COLLISON: We really learned something about the importance of prioritizing actually manifested user value and in most cases, taking the most straightforward path. When we started Stripe, I think we took a very different approach.

FIDJI SIMO: Looks like that worked better the second time around.

PATRICK COLLISON: We learned some lessons.

FIDJI SIMO: There’s something you guys said in a podcast recently, the Dwarkesh Patel podcast, where you talked about beauty. And [I’ve] got to say, I don’t know about you guys, but I rarely hear the word “beauty” when talking about vertical SaaS. And yet, as a European, as a French person, the word “beauty” is literally the meaning of life. It really caught my attention. And I'm curious, can you talk about how you think about beauty and craft in what you do?

PATRICK COLLISON: I think… the best people in the world in any domain treat their craft as a craft. It’s not just something to get through the day or to pay the bills. I think if you want to work with and build teams comprised of the best people in the world, you need to go beyond the ordinary and try to figure out what the best possible solution to whatever problem you’re solving happens to be. That’s just so much more satisfying. In any domain, first you need to choose a domain where you actually care about getting to a really good solution. If you’re building something that you, yourself, feel is trivial, where you, yourself, don’t feel like the problem has that kind of resonant integrity, then it's going to be hard to push yourself to that extent. But if you are in a domain where you think that that really matters, then it’s so much more fulfilling, if you give yourself the space and find like-minded people who want to get to something really fantastic.

On the other side of whatever you’re building, there are actual people. When we go to a city or a building that is in fact beautiful, there’s a kind of… there’s a generosity in that construction that we appreciate. When I’m walking through Paris, I never meet the creators of the buildings or the architect, but it’s a gift that they’ve bestowed on humanity that they didn’t pursue things in some slipshod and slapdash way and created actual Paris. Whatever we’re doing, we all have this incredible fortune to be creating things and things that other people actually use. I think there’s a question, as we put the results of that creation into the world, are we doing something that people find… elevating and… sometimes inspiring or even if it’s not inspiring, maybe just deeply satisfying. You come across some little touch in some little product that shows that the person behind it really cared. I think it actually matters.

JOHN COLLISON: In a way of saying what you’re saying, tell me if this is right. Beauty has a value and a morality of itself. We think there is a business value for Stripe and everything like that to attempt to create things that are beautiful. But even if none of those things were true, it is more enjoyable on a day-to-day basis to work with a team where that is a focus. I was very inspired by Jensen yesterday working on NVIDIA for 30 years, and if you want to be putting many years against a problem and seeing those returns to tenure, then it'll also make duration easier because it’s more satisfying working on something beautiful than something not.

PATRICK COLLISON: Yes, you can work on something… You can work on something you're not super proud of for two or three years, but you can’t work on something that you're not super proud of for 30 years.

FIDJI SIMO: I couldn't agree more, guys. I think there’s a massive responsibility we have as creators to bring out some beauty in the world. Okay, let’s go to audience questions, and we're going to be projecting some of them here. The first one is, “As someone who works with their brother, as I do with my wife, what would you say is your number one advice to keep a family business working?” And I want to mention that this question was submitted anonymously, which makes sense because the person is clearly asking for marital advice. Let's give them some help.

JOHN COLLISON: You have the privilege of working not only with a brother, but also in a separate context with your wife. You’re really playing the game on hard mode.

FIDJI SIMO: How is that going, Patrick? Tell us everything.

PATRICK COLLISON: God, where to start? The advantage of John and I working together is we've… Before starting Stripe, we had decades of conflict-resolution experience and mostly that involved rolling on the floor, casting punches, but you have to start somewhere. And now we’ve really grown. That rarely happens. It actually gets back to… the craft thing on some level, you want to work with people you admire, and I am fortunate enough to have siblings I admire and a wife I admire.

The advice you get here by default… or the prevailing advice I think is wrong. People often say, “You shouldn’t work with your best friends. You shouldn’t work with your brother, your sister, your wife.” It’s among the most fulfilling things. Work with the people closest to you. It’s awesome. We spend most of our daily lives working and building things, creating, and why not do that with the people you’re closest to? It’s awesome.

JOHN COLLISON: I think... I think there’s also an element for us… where we really believe with Stripe there are returns to duration and years on this problem. There’s now a fantastic collection of the 8,000 people at Stripe. It has taken many years to put together that team that now we can go execute on the things we want to.

Similarly, what you saw announced yesterday was now that we have some momentum with Stripe, the customer base, and the team of the people in this room and beyond, we can take the scale of the Stripe ecosystem and make the product better as a result of that. But there’s a way in which we view the first, 10, 15 years of Stripe as laying the foundation, and now you can do the really fun stuff. And I bring that up because most cofounding teams just seem to be unstable. If you look at the cohort retention curves of cofounders, 15 years in, they’re not that stable. And look at any of the large companies—Microsoft, Facebook, or Google, pick your example—and I think just having… doing a company with someone with whom you have a very high trust relationship is pretty valuable for being able to, for us as we look to the next 15 years and all the stuff that we can do to grow the internet economy, I think we get excited about that.

FIDJI SIMO: Makes a lot of sense. Second question—

PATRICK COLLISON: There's nothing like your sibling to give you blunt feedback as well.

JOHN COLLISON: That is important because people grow in stature and prominence… Patrick has a lot of bad ideas and, if you get this “Yes, sir, no, sir, three bags full, sir” thing at companies, you need someone who is willing to tell you this, “Maybe not this year.”

FIDJI SIMO: Patrick, I need to give you a chance to react here. What is the feedback you usually give John? He also has a lot of bad ideas?

PATRICK COLLISON: There are no bad ideas. Some ideas are… their time has not yet come.

FIDJI SIMO: Makes sense. The second question. “What are the primary challenges that Stripe faces in scaling its operations globally and how are these being addressed?” Thanks, Deepit, for asking the question.

JOHN COLLISON: I think the big one is… with a lot of product development, the products are pretty similar between places. Salesforce is one of the most successful SaaS companies ever. It’s the most successful SaaS company. I guess they would certainly say so.

But the Salesforce product is a CRM and you have all this functionality for managing customers and there’ll be small tweaks that are different by country, but fundamentally it's the same product, it’s CRM. Some language localization, things like that. And you spread around the world.

In Stripe’s case, we spent the last 45 minutes talking about the differences by country, but all the payment systems are very different by country. We talked about all the taxation systems, everything that our product touches is specific to the…

PATRICK COLLISON: It’s actually getting more different.

JOHN COLLISON: Absolutely. Yeah. This increasing heterogeneity as the global decoupling we talked about. The products are different.

We also find that culture around payment phenomena is very different. Germans are really big on cash on delivery. They’re a little skeptical of this idea of buying something and then just trusting that the merchant will send it to you. They really want to hold it in their hands and confirm that it’s real and works before they pay for the product. You get all these cultural differences in various places.

I think the challenge for us and also the opportunity for us, but currently it feels more like a challenge than an opportunity, is making Stripe act as a Babelfish translation layer where all of those differences are abstracted for the end customer.

PATRICK COLLISON: And part of the question: how is this being addressed? Heretofore, our strategy has been to try to build as many things as possible, more support for more payment methods, more acquiring countries, more… kind of money movement supported between markets and currencies. And we’re continuing with that.

But the two things we’ve really been focused on are one, building a general purpose. It’s getting a bit into the weeds, but you’re at the weeds payments conference, so right venue. An underlying treasury platform that enables us to build a lot of the core functionality in Stripe once. And then, build ingress and egress rails for different banking networks and systems in different countries, but one common treasury and liquidity management layer.

It’s been a several-year undertaking, it’s now being deployed into production. It doesn’t show up immediately as a user-facing product, but it’ll enable us to move more quickly across many markets in parallel.

Then the second thing, and you heard Will talk about this yesterday, is this idea of enabling other companies to plug themselves into Stripe. When Stripe was much smaller, there wasn’t the same desire or motivation, it wouldn’t have made as much sense for a company like NICEPAY, or, choose your favorite example.

Now because of the scale of the Stripe ecosystem, there are a lot of companies coming to Stripe and saying, “Hey, we would love distribution to your user base. We’d love to work with Instacart, we’d love to do this or that. We’re willing to do the work, can you enable us to do that?”

We’re trying to expose more hooks into Stripe so that those use cases and those integrations can occur organically without Stripe having to be the rate limit, the gatekeeper, or the bottleneck.

Again, that’s now entering production. With Amazon Pay, with Revolut Pay and so forth, there’s an important structural difference there, those integrations are being built by those companies.

We’re happy to do the work. It’s not that we’re trying to shift the work per se. It's more that by enabling that, we can gain far more coverage in parallel. Even though we added more payment methods this year than any prior year, I think that the rate is still going to accelerate as we make that pluggability more broadly available.

FIDJI SIMO: I find that there are people and companies that run toward complexity and realize that if you manage to solve complexity and simplify it down, it creates massive enterprise value and then people run away from it.

I think a big part of your success has been abstracting all that complexity so that your customers don't have to handle it. That makes sense.

Deepit is at it again. He was very inspired and had questions very much upvoted. “Given the focus on the early stage of software-driven innovation, which emerging technologies [are] Stripe investing in and what potential do you see in them?”

PATRICK COLLISON: You just heard about the crypto thing; we’re excited about that. We’re excited about… Obviously we spoke a lot about some of our AI-related functionality yesterday. That’s both deploying AI in the obvious ways. Same with the Sigma Assistant, it’s cool, it’s useful or the Radar Assistant, but it’s obvious. There’s also a lot that we’re exploring, because of the shift in what's possible, there are fundamentally different kinds of things that you can now do that were inconceivable before.

There's also an adversarial aspect to that. A core challenge for a lot of businesses over the coming years is going to be increases in the sophistication of fraud where if tokens and thinking is now free, you can pretty quickly harness an awful lot of deviousness in an automated way. We're investing ahead to try to figure out what post-LLM fraud prevention looks like because I think this can be a really core issue for the online ecosystem. There's also honestly a lot of early-stage technology and early-stage innovation… crypto and AI are hot and cool and you can say you work with them at a party and that's fancy, but we're very inspired by container shipping.

Container shipping is obviously one of the most miraculous inventions of the last century. The globalization of the world that occurred of the ‘70s, ‘80s, ‘90s, 2000s was substantially enabled by the invention of the container.

As we think about early-stage technologies, we don’t want to just be thinking of things that are cool, we want to be thinking about, what’s the containership here and as we think about models to, as John just mentioned, the growing ecosystem, balkanization and complexity, the differences across different markets, new currency controls and all the rest.

As we think about ways to unify across that, we don’t just want to pursue things that you can write cool blog posts about. We want to pursue things that actually move the needle. And sometimes those aren’t particularly glamorous, but they’re really impactful.

FIDJI SIMO: You heard it first: AI is out, containers are in. I’m joking, but it’s actually… You are so right.

PATRICK COLLISON: So far, containers have boosted global trade by a lot more than AI.

FIDJI SIMO: That's actually very on point. “What’s the progress on Stripe Climate and are you considering other initiatives like it?”

PATRICK COLLISON: Really good is a short answer. I think this year, we’re going to contract about $600 million of carbon removal.

JOHN COLLISON: You have to back up and explain Stripe Climate for the uninitiated.

PATRICK COLLISON: You folks know what Stripe Climate is? Put your hand up if you do. Okay, maybe. Fine. Obviously, lots of us are worried about what’s going to happen to the atmosphere and the climate and all the rest. The normal thing that companies do is you buy carbon offsets. But the problem is carbon offsets are overwhelmingly fraudulent. So that’s bad.

When we started to think about what we would like to do at Stripe, but also what we could enable business across the Stripe ecosystem to do, we wanted to do something that was better than that, not fraudulent.

By the way, offsets are usually fraudulent, not because… there’s deliberate fraud involved. It’s intrinsically an unsolvable problem. If somebody’s selling—

JOHN COLLISON: They’re paying someone to say that they are not emitting carbon, that they otherwise wouldn’t have or that they otherwise would have emitted. But you’re paying for an attestation and when you pay people money for an attestation—

PATRICK COLLISON: You’re paying them to not do something.

JOHN COLLISON: Exactly. It’s very hard to prove the counterfactual and the incentives are all topsy-turvy.

PATRICK COLLISON: Exactly. The IPCC reports, the main Intergovernmental Panel on Climate Change, they show that if we’re going to solve this problem, stay below two degrees, we’re going to have to do an awful lot of carbon removal because there’s an overshoot, there’s a ton of CO2 in the atmosphere. Even if we decarbonize the economy, it’s not going to go away.

We need to figure out how we get that out of the atmosphere. In 2018, no company in the world had purchased any carbon removal. It’s this funny stage where we know we’ll need to, at some point, those molecules aren’t going away, we need to pull it out of the atmosphere, but nobody was actually doing it.

So we thought we could lend a bit of acceleration, a bit of credibility to that ecosystem by becoming a commercial buyer of carbon removal. Removal in terms of paying a company to take it out, show you the removed carbon molecules. In that way, it’s much more verifiable.

We launched that to Stripe Climate and thousands of Stripe businesses have now joined us in paying a portion of their revenue to remove CO2 from the atmosphere. Stripe Climate is now by far the largest carbon removal coalition in the world, thanks to you guys.

It's had a hugely catalytic effect on the ecosystem. There were two companies doing it back in 2018. Now there are dozens, potentially even hundreds.

We ended up starting an AMC, an advanced market commitment, called Frontier about two years ago. We have a bunch of larger companies together to commit initially $1 billion dollars, that amount has now grown, to pre-commit that for carbon removal because if you think about starting a company, you want to know there's some demand there. That includes Meta and Alphabet, McKinsey, JPMorgan, a whole host of organizations.

This year we’re going to sign contracts for the removal of $600 million worth of carbon, which nets out to be probably between 20,000 and 40,000. Separately, 20,000 to 40,000 tons of removal will be delivered this year. Actually removed, and 20,000 tons is a lot, it’s like the weight of 300,000 people.

That’s a third of San Francisco. If it fell on your foot, it would hurt.

FIDJI SIMO: Let’s not try.

PATRICK COLLISON: Exactly. Stripe Climate is going really well; we're not really considering anything else at the moment. We’re just focused on making it work. Even though removing 20,000 tons from the atmosphere is good, there’s still a lot of tons there. We don’t get over our skis.

JOHN COLLISON: The other bit that’s relevant here is the place this comes from is… In working on climate, there’s a bit of a schism within the climate community, there are some people who hold the view that’s a bit doomer and humans should feel bad about reproducing and your dishwasher should take a very long time to not wash the dishes and the shower should come out in a trickle. It’s a view that climate change is happening, therefore you need to feel very bad and use less energy. We are of the view that technology will be the solution to this, like it’s been the solution to so many other things within history. There’s a lot of carbon in the atmosphere, and we have to get going on hoovering it out. There are all these geniuses working on a variety of different technical approaches and we should start funding them so that they have to eventually compete in the free market. But we can kick-start that stuff. I think it also reflects for us a view that we should be looking at places where technology can solve some of these problems, which is not a universal solution to all problems but to a lot of them.

FIDJI SIMO: How do you guys decide what to prioritize when you have new initiatives like that, knowing that your mission can cover so much?

PATRICK COLLISON: We plant a lot of seeds and we try to learn iteratively. With Stripe Climate, we didn’t start out to do this giant thing. All we did initially with Stripe Climate was say, “We will purchase a million dollars of carbon removal.” That was the announcement, and it was one person’s 10% project. It was a process of successive iteration learning. When we did that, Stripe customers came to us and you guys said, “Hey, is there any way we could do that with you?” We thought about that and we ended up building Stripe Climate. Then we saw that that really went vertical, that’s when we started to think about the AMC and so on.

We plan to be doing this for a long time. We’ve been doing Stripe for 14 years, I mean nothing compared to Jensen, but for 14 years, we think we'll be doing it for another 14 years plus. We’re not in a hurry, we’re not trying to juice the results for two years’ time. We’re fine planting a seed, checking in two years later and if it seems promising, we’ll do a bit more and after… Stripe Climate has been going for six years now and after six years, it’s pretty cool. But it was organic.

JOHN COLLISON: I think incentives and the culture of how a company works is important. It's been a lot of discussion of Boeing and their tribulations of late and people discussing how this results from the finance people taking over post the McDonnell Douglas merger and an over-focus on shareholder value. When you think about it, there is no universe in which this series of events is good for shareholder value. It's only if you have this incredibly narrow, short-term view that you do this set of actions, it's not an overly shareholder-value focus, it’s an overly failing the marshmallow test, abjectly quarter to quarter, doing very short-term oriented things view.

At Stripe, we really want to steer the company toward the opposite direction. It's part of the value we have as a private company, where we can continue to invest from a product point of view. Also in things like Stripe Climate, if you ask us what is the next quarter earnings per share impact of Stripe Climate, it's hard to point to, but we think it's quite valuable for us to be doing it over the long term and the same with a lot of our product investments. I think it's partly, can you set up the company with the right habits and incentives and ways of working to enable you to do stuff that's valuable. Climate is obviously a slightly oddball example, but there [are] lots of examples within the product suite in particular.

FIDJI SIMO: Makes sense. If you can go solve Boeing's problems in your spare time, John, we'll be grateful. “What plans are there around providing offline payments for mobile apps, but without Stripe hardware and terminals? The use case is for technicians in the home-services field who are often without cell reception for a day or more.” Any plans there?

JOHN COLLISON: We do, with the Stripe readers themselves, provide offline support and that’s awesome because that happens more than people realize, especially these days it feels like, cell phone coverage is getting worse somehow. The 5G spectrum is too clogged or something. It feels like we’re going backwards a little bit in cell coverage. I don’t know the answer, I think they were saying deviceless on device.

FIDJI SIMO: Because all the doomer NIMBYs blocked the cell towers.

JOHN COLLISON: Is that why?

PATRICK COLLISON: I think so.

JOHN COLLISON: Okay. I think we’d have to figure out is that a Tap to pay functionality. I don’t know when that is coming.

PATRICK COLLISON: If you want to do it on the phone itself, you’re probably going to have to go talk to the phone manufacturers.

FIDJI SIMO: I’m going to do two questions at once because they’re related. The first one is, “You've been vocal about ambitious goals beyond financial success. Can you elaborate on your vision for how Stripe can have a positive impact on the world?” Thanks, Casey, for the question. And then someone else asked, “With 1% of global GDP processed in payments volume in 2023, what does executing against the mission of increasing the GDP of the internet look like in the next few years?” I think those two are related, are both about your mission and your impact.

PATRICK COLLISON: The world has only figured out one strategy for increasing prosperity in a broad-based way. And that is capitalism, free markets, open economies, innovation, entrepreneurship.

FIDJI SIMO: We can all clap to that.

PATRICK COLLISON: Pandering to the crowd. Stripe isn’t doing that; you guys are doing that. We hope that Stripe can play some kind of enabling, supporting, behind-the-scenes backstage role in catalyzing it. Sitting in San Francisco, it’s easy to feel that the world is… that we live pretty comfortable lives, even if the showers trickle. Broadly speaking, things are pretty good, but that’s not the experience of most people in most of the world. Median income in the world is like $12,000 a year or something like that. That figure has made impressive progress over the last 20 years. But there’s still a very long way to go.

And John and I grew up in Ireland, if you can’t tell by the accent. Between 1988 when I was born and 2006 when I left Ireland, Ireland was among the fastest-growing economies in the world. When my parents were growing up, Ireland was a pretty destitute place. Ireland was a puzzle. Why was it so poor? It’s right next to the UK; the UK is pretty rich. Why are things going so wrong in Ireland? A profoundly impoverished place. Electricity didn’t arrive until roughly when my parents were born. Both my parents were the first in their families to go to college. Both of their parents were farmers.

And then suddenly, thanks to some enlightened policy changes where we shifted from being an autarkic closed socialist economy to one open to the world, we joined the EU, then Ireland’s economic boom really started. For us growing up around that, you can’t have grown up seeing Ireland in 1988 and 2006 and not appreciate what it is, what economic progress and growing wealth and prosperity can do for a country and a society. You don’t take it for granted. Global GDP is on the order of $100 trillion… but that’s not a ceiling. Our hope is to play the supporting role in hopefully enabling millions of businesses and entrepreneurs create the future that ought to exist.

FIDJI SIMO: When I do product reviews, I always ask my team, “What’s the longest pole?” What do you think is the longest pole or biggest impediment to get to that higher level of prosperity?

PATRICK COLLISON: It's hard to look at the news and feel an unalloyed optimist. There [are] always storm clouds on the horizon here or there. I don’t think one can ever take… open markets and the ability to relocate and to trade for granted. In certain ways, we've gone backwards over the past couple of years. We went backwards during COVID in many respects; China is much more closed off than it was certainly six years ago. Russia is certainly much more inaccessible than it was in the not-too-distant past. Again, we have a recipe that works. I think the question for the world, but this is above Stripe’s pay grade, is, “Can we stick with it?”

FIDJI SIMO: We have an interesting question from Gudmundur. “Do you have a competitor that you learn from?”

JOHN COLLISON: Oh, okay. A lot of people like to talk about Amazon. They famously said for a long time, “We're not customer-focused.” Sorry, “We're not competitor-focused—we're customer-focused. We never look at what competitors are doing.” I think it's total hogwash and not true. Amazon has several competitive intelligence teams; that's the case for all companies. They look at what their competitors are doing because your customers are extremely smart, well-informed people. We'll get to the competitors as well, but all of you are very smart, well-informed people and have choice in what they use. If your competitors exist and have a viable business that’s growing, as many of them do, then clearly there are tons to learn from them. I remember talking to Stan Druckenmiller, an investor, at one point, and he was talking about how he would look at what's going on in stock prices and use that to back propagate on what’s going on in the real economy. And so, he would say, “Trucking stocks are way down. That means trucking volumes are down. That means things are changing in supply chains, I should go investigate that.” It’s the opposite from how you would normally do. You’d normally reason what’s going on in the real economy and then what effect will that have on stock prices. Whereas I like the idea that he was, in a Bayesian sense, looking at what’s going on in stock prices and use that as information signals because equity analysts are smart and reason about the real economy. I think it’s totally the same for us, every single one of our competitors, we have tons to learn from and we look at all of them, their products, and how customers are using them. We talk to customers about how they’re using them because our customers are smart.

If people are telling us information by using a competitor’s product, you’[d] be nuts to not listen to that. It’s a weird thing that happens in Silicon Valley that this… It comes from Steve Jobs and a misinterpretation and a misreading of him, that his people didn’t… “Oh, we just think about what’s the best product.” When he introduced the iPhone on stage, you guys remember that keynote before he showed the iPhone, he showed a chart of the smartphone markets and there were pictures of other smartphones on that chart, but they were obsessing over, “What are people using today? What is wrong with them? What are the opportunities? People don’t like the keyboards.” This weird Silicon Valley. We absolutely learn from all our competitors and they have a lot to teach us.

FIDJI SIMO: That’s very reassuring to hear. What do you...?

PATRICK COLLISON: I’ll just add to that: a mistake companies sometimes make is they only learn from the big competitors.

JOHN COLLISON: That’s a good point.

PATRICK COLLISON: Whereas we learn a ton from the small competitors and even small competitors that aren’t competitors, small companies have a lot to teach us. It’s like, “They’re doing this in this way and that’s pretty interesting.” You have to… And Stripe’s not that big compared to NVIDIA. It’s easy to get some hubris where you look to learn from people of your tier, just clicking around late at night and all sorts of, interesting little companies, you learn a lot.

FIDJI SIMO: I find that people often conflate learning from competitors, which is really good, and doing the same, and copying competitors, which is not necessarily great. If we could separate the two, that would be much better. I’m curious, when you learn from competitors, what do you do that’s uniquely Stripe?

JOHN COLLISON: That’s a good point. What you have to be able to hold in your head at once is one, there’s a huge amount to learn from competitors but secondly, we really believe in Jensen’s idea of the $0 markets. We have pursued many $0 markets over the course of the company. It is telling you information when customers use a competitor and there are some things that they’re happy with and some things that they’re not happy with, you do not go and slavishly copy and replicate that. The art of product is being able to distinguish between them. That is what you are paid for in product. If you can’t distinguish those things, you should not be working in product because that is the entire job.

FIDJI SIMO: Totally agree. We have one more question on AI. I think you guys have touched on it, but I'll ask in case you want to add. “Stripe is the leader in the developer experience. How are you thinking about improving or innovating the experience using generative AI in particular beyond the whole chatbot experience?”

PATRICK COLLISON: I think we touched on this yesterday. It’s clear that advance[s] in ML and AI are going to completely reshape not just the payments landscape, but the financial services landscape. So much of financial services isn’t just about the mechanistic matter of how do you move the money. It’s about predictive tasks around that, with respect to things like fraud or KYC or verification, stuff of that nature, but also business models pricing, it extends the whole way.

With our usage-based billing announcement yesterday, we’re shifting business models in a way that we think will make more sense for businesses that are in turn downstream of AI. I think it’s going to be extremely disruptive. And so, we’ve been... Our view at Stripe is we should assume that every part of the business is going to look quite different in five years. And maybe we’ll be wrong in that and say maybe it’ll take seven years or ten years or something. From the top of the funnel with the checkout flow all the way down to the mechanics of how we orchestrate the ISO message and the request of the card networks and so forth, it’s going to be totally reshaped, and that’s why we’ve been investing so much here.

JOHN COLLISON: But specifically on developer experience... One is, I don’t think we know how AI will… how the best product experiences in developer experience will flow, but that feels like one of the areas that will change the most in the short term because there is nothing as perfect for being improved by LLMs as coding where it is all textual already. And it’s an awful lot to keep in your head. If you go read the Stripe Docs these days, it’s a lot to keep in your RAM, but trivial for an LLM.

PATRICK COLLISON: We only gave it 10 seconds of stage time yesterday, but that GitHub Copilot Stripe bot we built, the little Stripe agent. That's a custom thing we built. So now you can chat directly with Stripe in VS Code.

JOHN COLLISON: The experience of writing code as a developer at Stripe today is already quite AI-powered, not in people trying to use AI to make it happen, but because that’s organically what’s happening. That’s similarly the case when people are building Stripe integrations they use LLMs to help with those integrations. We have not… What's very interesting to me is there’s a big backlog. There’s a big pipeline of AI advances that we have not yet figured out how to prioritize. Even if all AI progress from here stopped, if we shut down the whole thing, if Eliza had his way and the GPT-4 was the last, or 4.5 was the last model ever deployed, there’d be so many product advances from here because we haven’t figured out how to use it in products.

The example I sometimes give is, you know the way when you pull down on any list on iOS to refresh, it rubber bands and you get the loading indicator, that had to be invented and it was invented several years into the iPhone by a guy called Loren Brichter and he realized that when you’re scrolling in a list and you finish scrolling, you want it to refresh. Before then, there was just a manual refresh button. Despite the fact that all the tech exists, already existed for pull to refresh, it took someone in 2011 or 2012 when that UI affordance came in. The right product experience is often non-obvious, and when it arrives, it’s completely obvious and it’s as if it was discovered rather than invented. I think there’ll be a lot of those in how we use AI in a product context, even if we don’t get model advances, which we absolutely will.

FIDJI SIMO: That makes sense. I love this next question “What would you do if you were 19 years old today? Are there still opportunities to build companies at Stripe’s scale and magnitude of impact?”

PATRICK COLLISON: Definitely. Through the Stripe ecosystem, we see candidates for that on a daily basis. The thing that is hard, and John and I were chatting about this this morning, is you have to… The companies that end up working are usually those that don’t sound like good ideas when they’re started. They sound a bit unorthodox, a bit contrarian. The AI companies that are now working, they were started before AI was cool. Instacart was started before the on-demand economy really took off, you’re part of what caused it to take off. When Stripe was started, the term “fintech” barely existed. My advice to the 19-year-olds would be to try to figure out what's a… To think carefully about what's cool and what's high status and then to make sure not to do that.

FIDJI SIMO: Great advice. One last question because everybody's wondering, “Who is the Monumental Labs sculpture downstairs modeled after, Patrick or John?” That's the most important question of this conference, guys.

PATRICK COLLISON: I thought it was actually Will.

FIDJI SIMO: Oh! Will.

JOHN COLLISON: By the way, for context on Monumental, we're quite excited about them because we're talking about this idea that tech is the solution to a lot more issues than we think, climate change being one of them. But another weirdly being statues and…

PATRICK COLLISON: Stonework.

JOHN COLLISON: Exactly, stonework and beauty. If you see a grand building from the 1800s, there's all this elaborate stonework on it and then, the reason that the buildings built in the ‘60s and ‘70s are by and large so awful is partly because we did not have the labor markets shifted. There were higher-value uses of people’s time, so we did not have the people who are capable of doing such things. But now, they use the—

PATRICK COLLISON: A literal case of “We don’t have the technology.”

JOHN COLLISON: Exactly. It’s the meme from Twitter. Now they’re using both computer-controlled milling but also GenAI to generate the 3D models to make it much lower-cost than it was, to be able to do this elaborate stonework. That was like a bit oblique to be making that point with the statue downstairs. But that was the point. There are Stripe users that we’re extremely excited about because it’s yet another example of a surprising domain where new technologies can really help. I think they went rogue and made it look like Patrick personally, but that was not part of the design brief.

FIDJI SIMO: We won’t settle the debate here apparently. Thank you so much, you guys. It was really insightful.

PATRICK COLLISON: Fidji, thank you.