ANNOUNCER: Please welcome Stripe corporate advisor, Claire Hughes Johnson.  

CLAIRE HUGHES JOHNSON: Good morning. I am delighted to be the host of this morning’s AMA with Patrick and John. And thank you to everybody who submitted questions through the vibe-coded app. I hope it worked. I hope you filed a bug if it didn’t work. And I do want to let you know that Patrick and John have not seen the questions that I’m going to ask ahead of time. 

So this will be full, unfiltered Collison. And I will do my best to control the answers if they get out of control. But with that, let me welcome Patrick and John to the stage. 

Hello. Fancy seeing you here. Do you feel like you’re being haunted by the COO ghost of scaling past? 

JOHN COLLISON: I thought we’d escaped this. [laughter] 

CLAIRE HUGHES JOHNSON: Oh no, she’s back. 

PATRICK COLLISON: We’re about to get our performance review. 

CLAIRE HUGHES JOHNSON: Yeah. I’m not afraid to give feedback to founders. I think that’s known. So we’re going to kick off. 

PATRICK COLLISON: Feedback is a gift. And with Claire, one receives many gifts. 

CLAIRE HUGHES JOHNSON: Many gifts of feedback, many gifts. So we actually wanted to start off with a question that you both like to chat about certainly on the socials, which is housing. And this question got a lot of traction in the app. So you see it here, those of you on the screen, housing supply issues in this city being one that’s one of the worst in terms of very high housing prices. And we’re going to talk about if we’re making progress. 

However, I’ve already exercised some decision making here and decided that, although you both like to talk about this topic, we might want to bring in someone who’s spending actual job time on it. And I’m very excited to welcome a surprise guest to our AMA, the mayor of San Francisco, Dan Lurie. Welcome. Thank you for joining us. 

[Applause]

DAN LURIE: That was quite the reveal. [laughs]  

CLAIRE HUGHES JOHNSON: This is a reveal. And you even brought a chair. 

DAN LURIE: Yes. 

CLAIRE HUGHES JOHNSON: Excellent. Excellent operational planning. So, Mayor Lurie, not everyone here is from San Francisco. So I just want to give some context, which is you just recently came into office January of this year. And prior to being the mayor, you were a leader in philanthropy and in other organizations, a very active citizen of San Francisco but new to the job. And this is housing—we’re going to start with this, and then we’re going to come to some other questions. And I have a feeling these two may have some for you as well. 

But we do really care about this question because we want to make sure it’s affordable to be in San Francisco for the companies that want to be here and the employees of those companies. So what are your thoughts on how we can make progress on that? 

DAN LURIE: Well first, welcome to the greatest city in the world, to all of you. I do say this—I say this: When San Francisco’s—when we’re at our best, we’re the greatest city in the world. And we have not been there over the last few years. We all know this. But we are much closer to being back than most realize. And it’s just an honor to have you all here. It’s an honor to have Stripe hosting their conference. And by the way, these two told me backstage that we’re going to make some news right now that they have committed to coming back through 2028 for this conference. So thank you all for the vote of confidence in San Francisco. [applause] 

You know, one of our big challenges is housing. And I remember seeing you at a meeting with the governor a number of years ago talking about this. So I know it’s a passion of yours, and it’s a passion of mine. We have to build more housing here in San Francisco and statewide. We have a housing shortage, no question. And so we’ve put out a family zoning plan that is going to help us build housing for the next 30, 40 years. Give people certainty. 

PATRICK COLLISON: And this is like a quite ambitious, significant plan. 

DAN LURIE: Yes, it’s zoned for— 

PATRICK COLLISON: Some are calling it radical. 

DAN LURIE: Well, it’s measured in that we don’t expect— 

PATRICK COLLISON: We think it’s wonderful. 

DAN LURIE: Good. It’s 36,000 units for zoning. Will that get built over the next few years? Probably not. But it gives families certainty, knowing what they can do with their property. And they haven’t had that. And also, we need to do better on our permitting here in San Francisco. On average, seven years, $1.1 million to get one unit of housing built. It’s insane. We have to shorten the timeframe because in construction, we all know, time costs money. So we got to shorten that down. 

So we have to create certainty in permitting, certainty in zoning, and then we need the macroeconomic forces in terms of interest rates. We need those falling so that you can start building again. But this plan that is going to go through a lot of conversations over the next three months including one tonight that I’m going to that will be interesting where—you know, people are scared of change. I get that. But we’re going to maintain San Francisco’s character while making sure that it’s affordable and able to draw employers back to San Francisco with the knowledge that housing can be built here. 

CLAIRE HUGHES JOHNSON: Great. Excellent. Sounds very—like you got a tactical plan there. So let’s talk about the fact that, as we mentioned, you have not been a career politician, never been an elected official. And I’ve heard you’ve already got property crime down 45%. So congratulations. 

JOHN COLLISON: In particular, like a lot of the discourse— 

CLAIRE HUGHES JOHNSON: Yes, [applause] we like data-driven results. 

JOHN COLLISON: And in particular, a lot of the discourse around San Francisco over the last few years has been, you know, it just can’t be fixed because you’ve got the structure with the mayor and the board of supervisors and the unions and everything— 

PATRICK COLLISON: Non-profits, like the whole industrial complex. 

JOHN COLLISON: Exactly. It’s just an unfixable problem. And so there you go. And you seem to be embarrassing that type of—

PATRICK COLLISON: You walk around San Fran— 

CLAIRE HUGHES JOHNSON: But now you’re in it. Now, you’re in it. 

PATRICK COLLISON: —you walk around San Francisco, and the difference is palpable. So what are the leadership tips? 

DAN LURIE: Well, I think one is we do have a strong mayor system in San Francisco, and that got a little confused. The Board of Supervisors here, our city council, they have power. It’s grown over the last 20 years, but the mayor still has a lot more to say about it. And so I’m going to lean into that. I’m going to own that. But we’re also having a different approach, which is we’re trying to work with people. 

We’re trying to cooperate with our city council. It’s a shocking thing. When you walk down the hall, which I do, and say hi to them, they’re like, "What are you doing in my office?" They haven’t seen a mayor walk down to their office before. Department heads—we have 58 departments, just probably too many. We have a $16 billion budget. 

But many of our large agency heads—let’s say the top 20—they did not have weekly department-head meetings. We have weekly department-head meetings. Every Tuesday morning, 9:00 a.m., we get around the table. And we say, "Okay. What’s going on? How can we help each other?" They weren’t doing that. And it wasn’t because they were bad people. They just weren’t being led in a way that— 

CLAIRE HUGHES JOHNSON: Yeah, not a lot of collaboration. 

DAN LURIE: Not a lot of collaboration. So it’s a new day in San Francisco. I’m trying to lead by example by being out on the street talking to people that are suffering, that are struggling with addiction. I was on 16th and Mission this morning before coming over here and talking to a young man, who is from Belize, and has been on the street for three years. And I asked him how he was doing. He goes, "I’m in the throes of something right now." And he, like, literally had tears in his eyes. 

Like the people that are struggling on our streets—they don’t want to be there. We need to stand up more mental health beds, more drug treatment beds. And we have to stay focused. And so I was telling you two backstage, my focus in San Francisco is that there needs to be accountability for bad behavior. And public safety is my number-one priority. Nothing else matters. We can’t have eight thousand of you come in and not feel safe. And I hope you’ve felt safe down here. Our officers are doing an incredible job. [applause] Thank you to SFPD. 

So public safety—number one, we’re short 500 officers. We are attacking that recruitment problem and that retention problem. Second issue, the behavioral health crisis on our streets. We know that we have that issue. We’re going to stand up more beds. And third is the permitting, making sure that life is easier for small business and big business and that we get housing built. 

And my message to everyone around the world is: San Francisco is open for business. We want you here. We want you coming back here. And I think we stay laser focused on that. We bring people together. We don’t demean people and throw darts at people every time that they disagree. Let’s work together. Let’s get the job done. Let’s get results for the people of San Francisco. 

CLAIRE HUGHES JOHNSON: Thank you. Great. 

PATRICK COLLISON: I think it’s a very mysterious thing how there are three cities that are tied for being the best in the world: San Francisco, South San Francisco and Dublin. But that’s— [laughter] I don’t know how that happened, but that’s the reality. 

DAN LURIE: I do like Dublin. Dublin’s a fun time. 

PATRICK COLLISON: I find these—the idea of thinking about a city as a system and a city as a platform for commerce, for residents, for a community to be extremely interesting. And we had Alain Berteau come and speak at Stripe back a couple months ago. And he’s a very interesting thinker on some of these issues. 

But if we kind of pursue this idea of, you know, city as platform, you know, Stripe as a platform. And we just spent the last two days covering what’s coming in the Stripe platform over the coming year and what we hope to enable in— 

JOHN COLLISON: The roadmap of specific improvements. 

PATRICK COLLISON: Exactly. So for San Francisco as platform, what’s on the roadmap? What’s coming? 

DAN LURIE: Well, I think, if we can get those three things right, I—right now, we don’t have enough people coming downtown. We don’t have enough people thinking of this area that you’re in as a 24/7 community. And we’re going to do that. So we’re looking at bringing universities here, more housing downtown. Like, this has got to be a live, work, play neighborhood. 

But to be quite honest, I think of our job and my job as creating the conditions so that San Francisco can achieve liftoff again. We are the home of innovators like yourselves. And what you have built in this company is simply incredible. And we want more of that. Like, I want OpenAI and Anthropic and Stripe. I want you back here, by the way, in San Francisco. 

But I want to create the conditions so that every entrepreneur, innovator, artist, cultural leader thinks, “I’ve got to be in San Francisco.” And by the way, we’ve been there before. And we’re going to be there again very soon. And so what’s next is, we are a platform because the brainpower is here. The beauty is here. I mean, you know, it’s not Dublin, I get. But we have it all. And when we’re at our best, I’m telling you, no one can beat us. And with AI coming on strong, everyone—people in Texas are saying, "I’ve got to move back to San Francisco, I guess." 

CLAIRE HUGHES JOHNSON: [laughs] 

PATRICK COLLISON: It has been funny to see various people who declared in very— 

JOHN COLLISON: Definitive terms. 

PATRICK COLLISON: —exactly,that San Francisco was fully over are now—they don’t tend to proclaim their return to San Francisco with quite as much fervor, but you see them returning. 

DAN LURIE: No. And we’re getting some good kudos on that front. And let me be clear: we have a long way to go. It didn’t just happen overnight that we got into where we’re at right now, but we are going to get there more quickly than most expect. But we are going to be relentless every single day like I know you are at your company. You can’t—this is hard work, and we’re up for it. My administration’s up for it. And we want to partner with businesses like yours, small and big. And we want to invite the world back to San Francisco. 

CLAIRE HUGHES JOHNSON: Well, Mayor Lurie, we have to let you go— 

DAN LURIE: Thank you very much. 

CLAIRE HUGHES JOHNSON: —back to your day job. 

DAN LURIE: Thank you. 

CLAIRE HUGHES JOHNSON: But I heard having clear priorities, creating an environment for success, encouraging collaboration not just in your direct team but in the whole organization, and holding people accountable is what I heard. 

JOHN COLLISON: You love all this stuff. 

CLAIRE HUGHES JOHNSON: I love this. You all know I love this stuff. 

DAN LURIE: It’s an honor to be with you all. 

CLAIRE HUGHES JOHNSON: Leadership matters. 

DAN LURIE: Thank you very much. 

CLAIRE HUGHES JOHNSON: Thank you, Mayor Lurie. 

JOHN COLLISON: Thank you for your leadership. [applause] 

CLAIRE HUGHES JOHNSON: Thank you. I was going to say I’m surprised he didn’t take his chair. But look. All right. [laughter] We’re good. We’re good. Sorry. We’re going to get back to the AMA, and we’re going to start with our first question. 

JOHN COLLISON: That was cool. 

CLAIRE HUGHES JOHNSON: That was. Yeah. I mean, I know you wanted to keep going, but we’re going to come back and talk about some Stripe stuff. So we’ve got some stablecoin questions we’re going to lead off with. So let’s go to the first question, which is: what is the biggest hurdle in your views to greater adoption of stablecoin? 

I was in a breakout yesterday where every single panelist quoted you as saying it was the room temperature semiconductor that we all needed. But we need more room temperature semiconductors. What’s the hurdle? You want to go first? 

JOHN COLLISON: I think there’s a few things. One, everything is still a little janky today. I don’t know if you guys have noticed that. But we set up— 

CLAIRE HUGHES JOHNSON: The technical term. 

JOHN COLLISON: —to drive more adoption and awareness at Stripe, we have a little barista coffee station where you can get a coffee, and you pay for it. And previously we just accepted credit cards with Stripe. And we added the option to pay with stablecoins. And let’s just say the experience is not quite as quick as the card experience. And it’s like, well, which chain do you want to use? And do you have the right—and it’s bouncing out to— 

PATRICK COLLISON: You know, if you order a coffee that takes six minutes to make, that’s an appropriate thing to order when trying to pay with crypto today. [laughter] 

JOHN COLLISON: And so just it’s a little bit janky. I think consumer—part of it is just consumer awareness and adoption. People are starting to get more familiar with everything. But I think, as the actual wallet experiences get better and as just people get more familiar with everything, there’s also the—I think it’ll be helpful that there’s a new regulatory construct coming in the US. 

There’s the bills under debate right now in the House and the Senate. I actually don’t think those are the main things holding up adoption, but I think they’ll be a helpful tailwind from a consumer conference point of view. But yeah, I would say making the experiences less janky and just more people getting familiar with them. Right. 

PATRICK COLLISON: Yeah, I mean, obviously, as of very recently, there’s been a big barrier, at least in our view, at the platform infrastructure orchestration layer. Because if you have to go and set up a completely separate parallel stack for handling all this crypto stuff, it’s going to be really difficult to manage both of those together and to reconcile them and just to—it’s a big lift, and it’s not clear for most businesses that it’s worth it. 

We obviously want to make that extremely straightforward, so you don’t need to kind of decide, "Am I a crypto business, or am I a fiat business?"—and to make it easy to be both. Once we fully roll out this functionality, then, yeah, I think it's going to be on the consumer side along the lines of what you say. 

CLAIRE HUGHES JOHNSON: I would think consumers eventually— 

PATRICK COLLISON: By the way, I think it’s actually a big opportunity for a lot of companies here and a lot of other companies where—that’s not the core of our DNA and expertise, obviously—and I think we’re already seeing it, that there’s going to be, I think, a lot of stablecoin-native consumer financial services that do extraordinarily well. 

CLAIRE HUGHES JOHNSON: Right. And I think the consumer, in the same way that we’re going to meld sort of fiat and people don’t have to think about that, the consumer probably isn’t going to have to think, “in my wallet are these stablecoins”, they’re just going to be able to use them. 

JOHN COLLISON: And you see it in like the stuff we demoed yesterday in the product keynote. It’s very integrated—stablecoin, fiat, everything intermingled. And that’s very much a deliberate plan. 

CLAIRE HUGHES JOHNSON: And that’s what’s really going to get the adoption going. Alright. And because this is popular, we have another question. I don’t know if we want to put it up for everyone to see. But is Stripe planning on launching its own stablecoin? 

PATRICK COLLISON: It’s not what we’re focused. We’re focused on catalyzing stablecoin adoption. And that’s really where we’re working backwards from. And John showed the sad, embarrassing chart of Bridge’s transaction volume as compared to Stripe’s. 

JOHN COLLISON: Happy, embarrassing chart. 

PATRICK COLLISON: Exactly. Happy, embarrassing chart. And that’s our North Star. And if there’s a reason or some jurisdiction or some use case for which building a stablecoin might make sense, then we’ll certainly think about it. It is the case—and this was covered at the product keynote yesterday—that USDB exists. This is Bridge’s stablecoin. 

JOHN COLLISON: We kind of have our own already. 

PATRICK COLLISON: Exactly. So in that sense, one exists. And by the way, Bridge makes it easy for others to launch stablecoins. And so if any of you want to create your own stablecoin, you can do that today with Bridge. But from a Stripe standpoint, we’re focused on just growing usage. 

JOHN COLLISON: And the other thing I’d say is people often expect Stripe to be opinionated on things where—you know, which payment method do you want people paying with and things like that. And Stripe basically, broadly in all these things, just wants to support whatever businesses want. And that’s true with kind of existing fiat payment methods or things like that. 

We just support businesses as they operate online. And so similarly with stablecoins, our prediction is that there’s going to be a few big stablecoins in the world. And we’re going to support all of the ones that businesses want to use. And so we have a great relationship with Circle right now for USDC. And we do lots of USDC orchestration. There’s USDG out there. There are all these different stablecoins that exist. And I think we just very much follow the pattern of what people want to use. 

CLAIRE HUGHES JOHNSON: All right. Great. We’re going to switch gears. A favorite that—I also get this question. And this is from Mia. If you had to build Stripe again from scratch, is there anything you’d do differently? This is where I think we all laugh. Yes. 

JOHN COLLISON: Huge amount. Very little we’d do the same. [laughter] 

CLAIRE HUGHES JOHNSON: What? No. We did some stuff pretty well. 

JOHN COLLISON: You should talk about some of your original sins and like the stuff we’re finally fixing yesterday. 

PATRICK COLLISON: Oh, yeah. [laughter] Look, honestly, this event has actually been quite moving for me where—you didn’t see me in the product or see us, indeed, in the product keynote yesterday. And often, we’re involved in various parts and so forth. And obviously we’ve been working closely with many of the teams who put those products together. 

But yesterday, I just felt so proud of the Stripe team. And as I think about the idea of trying Stripe again and doing things differently, my emotional reaction to that is recoil because I don’t think it could possibly have ended up working out as well as it has. I think people did such an amazing job. [applause] 

CLAIRE HUGHES JOHNSON: Yeah. I mean, so when you see what happened yesterday, you think, well, even if we would’ve done some things differently, it turned out pretty well. 

JOHN COLLISON: That’s a lovely sentiment but does not answer the question. [laughter] 

PATRICK COLLISON: No, but we made so many mistakes. 

CLAIRE HUGHES JOHNSON: Or what John said. 

PATRICK COLLISON: You know me. I love talking about my feelings. That’s the first and last time it’ll ever happen. But every time we've built something where it’s a one-to-one or one-to-n mapping, that’s been a terrible mistake. Everything in the world, even the things that feel like they shouldn’t be, are actually like N to M. And often fixing those is painful and complicated. 

And you think, well, a customer could only be in one country. Nope. Or you know, even countries split their—everything just sort of breaks. And so, everywhere, you need an N-to-M relationship. And then I suppose the other kind of structural mistake that—I was asked separately yesterday which Stripe launch I’m most excited about. 

And you can’t choose favorites. But secretly inside, what I’m thinking is the stuff we showed around the multi-currency accounts, which in some sense that’s small in that, okay, fine, you can store multiple currencies in a Stripe account. Great. But what’s actually happening there is we originally thought of Stripe as this kind of thin conduit for accepting money from customers and quickly transferring it somewhere else. 

What we’ve heard from so many of you over the years and honestly where Connect has kind of led us as we’ve listened and learned is that actually there’s so much that you all want to do with the money management itself, with the treasury functionality and then having kind of tight and granular control over how the money is dispatched. 

And so you can kind of think of money in approximately three buckets: it’s like, they’re receiving it, there’s how you manage it in the treasury, and then there’s how you orchestrate sending it somewhere. And that multi-currency account functionality, I think, enables a lot in categories two and three there. And so in our conception of Stripe, it’s a very fundamental structural change. And so if we’re building Stripe, again, from scratch, I think we would think much more about categories two and three from the beginning. 

CLAIRE HUGHES JOHNSON: Anything on your list? 

JOHN COLLISON: I have been struck by how much—basically, it took me a while to appreciate this aspect of Stripe that people really care about, which is it’s often not the shiny new ship or the new functionality. It’s the fact that Stripe is really reliable and always there for you. And you’re running your entire business on it. And it just needs to be fast, performant, reliable, available. 

And I think, as Stripe has grown and scaled, oftentimes we have people coming to us or people switching to Stripe because it just works. And their previous provider was going down on them or, you know, all these sorts of issues. And it’s interesting because, you know, if you talk to a customer, generally they’ll talk in terms of features and things like this. And it probably was—maybe 10 years in, we really started seeing a lot of customers coming to us for this reason. 

And almost they wouldn’t kind of say that they were coming for this reason. But I now increasingly think of Stripe as we should simply be the most reliable infrastructure partner that people have. And they can always depend on Stripe always working. And I think I would have made that more of the Stripe self-concept and self-identity— 

CLAIRE HUGHES JOHNSON: Earlier. 

JOHN COLLISON: —two years in rather than— 

CLAIRE HUGHES JOHNSON: Yeah. 

PATRICK COLLISON: Okay. I have another one. Sorry. You’ve really got me going now on all Stripe’s problems and mistakes. 

CLAIRE HUGHES JOHNSON: All right, but we do have to keep going. 

PATRICK COLLISON: Okay, fine. But just one other one—so if we’re doing it again, part of—there isn’t that much self-serve financial infrastructure in the world. There’s lots of financial services companies, and most of them don’t have self-serve access to their functionality. 

JOHN COLLISON: Everything is behind a sales rep or get in touch. Or— 

PATRICK COLLISON: Right. And we just believe extremely strongly in the importance of self-serve access. I mean, it’s less hassle. Like— 

JOHN COLLISON: It’s the Internet. That’s how the Internet works. 

PATRICK COLLISON: Exactly, it’s much less hassle. And a lot of people just, like, don’t want to have to talk to somebody in order to get going. You want to do it immediately. But there’s also something about the universality and the permission-less of it and kind of creating— enabling bottom-up innovation. But the mistake, the problem, and the thing I would try to do differently if we’re doing it all again is we haven’t always gotten—and in some cases, we’ve fallen, you know, for me, painfully short on support and some of our risk tooling and functionality, where every business is so important. 

And yeah, I mean, Stripe is one of the world’s largest bug-bounty programs in some sense where there are all these entities and people trying to defraud Stripe to the tune of millions of dollars every day. So it is a very difficult adversarial system. But whenever Stripe makes a mistake and we miscategorize a business as being fraudulent or something, you know, that really weighs on the Stripe team. 

And you know, three or four years ago, especially post the COVID ramp-up, you know, that was really hard for us. And the support teams and the risk teams have made a huge amount of progress over the last year where we have record-high support CSAT. We have record-low support dissatisfaction, both with record numbers of users. 

And any false positives or misjudgments we make with respect to risk where we—you know, some legitimate business suffers some burden—those have fallen enormously over the last year. And I receive a bunch of emails every week now going through each mistake, why it happened and sort of how it can never happen again. So I— 

CLAIRE HUGHES JOHNSON: But you’re saying the thing you want to do differently— 

PATRICK COLLISON: I wish we’d been able to do all that much sooner. 

CLAIRE HUGHES JOHNSON: Earlier, yeah, because it’s really part of the product. It’s part of the experience. 

PATRICK COLLISON: It’s part of the product. Part of what we try to communicate to people who join Stripe is like, if you’re running—if you’re YouTube and you erroneously ban some user, it’s like, okay, that’s unfortunate. You can’t watch your videos. But it’s presumably not the end of the world. There are other places on the internet to watch videos. But with Stripe, each individual business is so important. And each mistake really matters. So like every time we get something wrong, it really pains us. 

CLAIRE HUGHES JOHNSON: Yeah. Well, as the recipient from early on in my tenure at Stripe of any number of either a customer issue or a risk issue, Patrick, you always paid attention to them, I think, for the founders in the room. It was known internally that this is not okay, and we need to address these. But I think the systematic addressing of them and building more product thinking, right, into support and risk is something we came to too late. 

PATRICK COLLISON: I’m really excited to try Verified. This is something that— 

CLAIRE HUGHES JOHNSON: Absolutely. 

PATRICK COLLISON: —has been an idea from many of you. 

CLAIRE HUGHES JOHNSON: Yeah. We’ve been talking about some of the launches yesterday for years, the Verified piece in particular. All right. We’ve got to keep going. This one—I don’t know if you heard. People want to talk about AI. So as we’re training bigger AI models, we launched the Payment Foundation model yesterday—how do you balance accuracy with the need for sub-100 millisecond latency, the compute cost, the carbon footprint? We obviously have our Stripe Climate Project. These are tricky things. How do we think about it? 

JOHN COLLISON: I would say, I mean, one of the most interesting Stripe-specific AI aspects is this need for low latency. Patrick talked about self-serve as one example of this. But when you think about it, the reason that lots of other companies don’t do self-serve is not because they never thought of the idea. It’s that decisioning whether a customer is legitimate or is a fraudster in the time that they are signing up is actually a pretty interesting AI and fraud-detection challenge. 

And so there’s a ton of work behind the scenes that doesn’t tend to end up in the product keynotes because it’s the behind-the-scenes special sauce. There’s a ton of work that goes into that. Similarly, whenever anyone makes a charge on Stripe, we are taking into account everything that happened in the Stripe universe up to the last 100 milliseconds. Was this card or IP address or device or anything like that used in previous transactions? But we have a view of the entire Stripe universe as of 100 milliseconds ago that is taken into account in those fraud decisions. And that is different to how lots of other people are doing. 

PATRICK COLLISON: When you talk to ChatGPT, you know, it tells you that— 

JOHN COLLISON: “Sorry, I can’t—you know, that’s outside my knowledge window.” 

PATRICK COLLISON: Exactly. You know, there’s a knowledge cutoff. Like if Stripe had a knowledge cutoff in the same way, then that would be a massive vector for fraud. It would make a lot of the authorization improvements, you know, much worse because we don’t realize that a bank is down or something. So we can’t have a knowledge cutoff. 

CLAIRE HUGHES JOHNSON: Right. But that means it’s more expensive for us. 

JOHN COLLISON: I think you’ve got to get clever on the engineering side. You can do it pretty efficiently. There’s some good things going on there. 

CLAIRE HUGHES JOHNSON: Some leprechaun magic situation. 

JOHN COLLISON: Offensive, but yes. [laughter] So anyway, this is one of the Stripe-specific AI things— 

CLAIRE HUGHES JOHNSON: I’m pro-leprechaun. I don’t know what— [laughter] 

JOHN COLLISON: —is doing lots and lots of real-time AI. And then there’s also aspects where it’s kind of like test-time compute in AI models, where you can have kind of a quick answer or have the thinking for a little while, like you get in ChatGPT or something. And it goes off and gives more deeply reasoned answers. There’s simply lots of places where we can either give a fast, off-the-cuff, pretty good answer, or go away and do, kind of, much more thinking and get to a better quality answer. And we’re starting to introduce more of that as well. I don’t know what you’d add. 

PATRICK COLLISON: No, I think that’s right. And I think that—I mean, I think you’re right and Deepit, who asked the question, is right that it is very hard to do this super quickly and super efficiently. The teams at Stripe have done great work. [crosstalk] It is now super cheap and efficient. And if any of you have friends who want to work on these problems, we’re hiring people. 

JOHN COLLISON: It’s one of the most interesting AI challenges because it’s so adversarial. Like, any time we have made the mistake of having any kind of a hard dollar threshold, or you can do these 10 actions before we review whatever, the fraudsters instantly manage to reverse engineer it like in a matter of days. And so it’s this actually pretty fun cops-and-robbers dynamic of the system that you are trained— 

PATRICK COLLISON: John, it’s not fun. They’re bad people, and they should stop. 

JOHN COLLISON: I mean, they are bad people, and they should stop. But it makes for very interesting challenges because you have this kind of responsive adversarial dynamic. 

CLAIRE HUGHES JOHNSON: But I do think—my observation—and now, I’m a little bit farther removed,is this is multiple teams in different parts of Stripe who have goals that they’ve just been chipping away at for years, right, to get the cost down, to get the latency down, to, you know— 

PATRICK COLLISON: What we’re really noticing now is— 

CLAIRE HUGHES JOHNSON: —achieve that balance. 

PATRICK COLLISON: —especially working with our platforms and everybody who’s using Connect, like every marketplace, every platform, every ecosystem of economic activity has fraud. And it’s not just payments fraud. If you’re facilitating an ecosystem in which money moves, there are going to be fraudulent sub-accounts, fraudulent sub-merchants, you know, whatever. 

There’s always people trying to defeat it. And it’s been interesting over the last year. And now as we announced yesterday with Radar for Connect, to see how some of these same techniques can be brought to bear not just kind of on the payment itself but on these sub-ecosystems, and we’ve—you know, maybe it’s obvious in hindsight—but it was kind of an epiphany for us that every ecosystem has its own unique kind of fraud. But you can use these generalized approaches to quite robustly detect it. 

CLAIRE HUGHES JOHNSON: But I think, just to zoom back, the question—it’s interesting to me. You both decided to talk about fraud because AI models, for us, that’s the most—I mean, when I joined the company, the ML work we were doing on fraud was the most interesting, I thought, technically, and sort of some of the secret sauce—I won’t say anything about leprechauns ever again—but some of the secret sauce in the fact that we didn’t have—we had 160 people, and we were actually doing quite a meaningful business already. 

So that is—but the payment foundation model is not just about fraud. Right. I mean, we’re seeing great conversion optimization, auth optimization, revenue lift. So I think that combination can be very powerful, and it’s worth it then to keep chipping away at the challenge that’s being described in the question, which is we’re going to make more and more investment then. And we want to be able to do it efficiently. 

PATRICK COLLISON: It’s true. It’s true. We’ve quoted publicly the—my mind does just go to these fraudsters. They’re so pesky. But we’ve mentioned publicly the fact that businesses that use Optimized Checkout Suite see on average 11.9% more revenue. And when we first—and that’s the product of dynamic ranking of every part of the checkout based on— 

JOHN COLLISON: Mark described this a little bit actually in the fireside, remember, with the advertising— 

PATRICK COLLISON: Right. 

JOHN COLLISON: —where he was saying previously people do targeting where it’s like 18 to 28-year-old women. And it just turns out that you kind of have a model of who the consumer is. 

CLAIRE HUGHES JOHNSON: You think you know. 

JOHN COLLISON: You think you know. And I think that’s certainly our experience on all of these optimization tasks is, you think you know. And businesses will say, oh, we should have this as the top-ranked payment. It’s just like— 

CLAIRE HUGHES JOHNSON: No. 

JOHN COLLISON: —let the AI figure it out. 

PATRICK COLLISON: Exactly. So 11.9%—when we first—when the A-B test yielded that result, we didn’t believe it. We thought there’d be an uplift, but 11.9% just seemed implausible. And so we re-ran it, and we actually got a higher figure the next time. Like it is true. We’ve now run it with kind of—you know, focused on different sub-segments. 

We can now kind of isolate some of the causality within it, so it is true. And Will mentioned yesterday, you know, the fact that a significant majority of the enterprises going live on Stripe are now doing so with the Optimized Checkout Suite, you know, basically because of this. 

CLAIRE HUGHES JOHNSON: All right. Now, we’re excited. Okay. 

PATRICK COLLISON: And, yeah. It’s just like, well, what else can you do in a business that increases revenue by 12%? Like that’s faster than a lot of businesses grow every year, you know. So we’re into it.

CLAIRE HUGHES JOHNSON: We are pretty excited. All right. Not just combating the fraud, but raising the revenue—all right. So let’s talk about the—I get this question myself all the time, which is that Google and Microsoft recently said that about 30% of their code written in the last year was written by an LLM. I don’t think it’s actually exactly LLM. But yes. And what is that stat for Stripe? What would be a good target percentage? Do you benchmark? Do we measure this? 

PATRICK COLLISON: We don’t track this. And I don’t know that you could even have a good target, because the world today is so different to the world a year ago, will be so different in a year, that any target would be so, like—

JOHN COLLISON: And obviously, it’s a spectrum where, if you use VS code, does that count? If you plug the code that you handwrote into an LLM and ask it to optimize it, which lots of people do, does that count? And so I think it’s all on the spectrum. 

PATRICK COLLISON: It feels like a vanity metric. The thing we do track closely is pull requests per engineer per unit time. And that is up around 30% year over year. 

CLAIRE HUGHES JOHNSON: So productivity. 

JOHN COLLISON: Well, a proxy for productivity. 

CLAIRE HUGHES JOHNSON: Proxy. Proxy. 

PATRICK COLLISON: It’s the best proxy we’ve found. Yeah, and you could definitely ‘good heart’ it by—I don’t know—giving out free cookies at the end of the day, you know, if you write more pull requests, so there’s risks. 

CLAIRE HUGHES JOHNSON: You’re not going to find us posting what percent of code was written by— 

PATRICK COLLISON: I mean, we’re not morally opposed. As John says, it’s like— 

CLAIRE HUGHES JOHNSON: It’s hard to measure. 

PATRICK COLLISON: Yeah, it’s a symbiotic thing. It’s hard to— 

JOHN COLLISON: Yeah, like there’s very few people handwriting their code on a legal pad these days. I think there’s actually very few people using no AI to write their code these days. And so— 

CLAIRE HUGHES JOHNSON: All right. So the answer is we do not know. 

JOHN COLLISON: Yeah. 

CLAIRE HUGHES JOHNSON: All right.

JOHN COLLISON: But I mean, I think we don’t know because— 

CLAIRE HUGHES JOHNSON: I think that’s fair. 

JOHN COLLISON: —we don’t think it’s the right metric. 

CLAIRE HUGHES JOHNSON: We don’t think it’s the right metric, and therefore we’re not measuring it. 

JOHN COLLISON: I think what is true is— 

PATRICK COLLISON: But here,many people at Stripe love Cursor. And our internal LLM thing has—I mean, it’s used every week by almost everyone at the company. 

CLAIRE HUGHES JOHNSON: I think a very high daily active user of our internal model. That’s true. 

JOHN COLLISON: I think what is true is just the way of writing code inside a large production code base like Stripe is changing pretty significantly. And so it’s quite different now from this time last year where, yeah, there’s much more LLM usage and much more Cursor. And I’m guessing, when we’re back in a year or two, if we’re to provide an update, it’ll be pretty different still. And so it doesn’t feel to me like we are at the end state of writing AI-enabled code. It feels like we’re in a—you know, we have the microcomputer. We’re in a very transitional moment. 

CLAIRE HUGHES JOHNSON: Yeah. All right. Good. So we sort of already answered this one—so maybe let’s go quickly —which is, what’s the most interesting. So we’ve mentioned some AI capabilities. What have we not mentioned that we think is very interesting that we’re working on? Anything you want to add? 

PATRICK COLLISON: Well, we haven’t mentioned the—I don’t know that agentic commerce is the right term actually but that whole side of things. 

CLAIRE HUGHES JOHNSON: Yes. 

PATRICK COLLISON: And– 

JOHN COLLISON: It’s a complete rewiring of how all commerce on the Internet works. We haven’t— 

PATRICK COLLISON: Exactly. But that’s probably big. And you know, you saw in the demo how, in some of these tools we think it ought to be possible and will be possible to transact in line, whether that’s on Will’s skincare website or making a purchase from within Cursor. And Andrej Karpathy wrote a very interesting—Andrej Karpathy of John’s vibe-coding fame. 

But he wrote an interesting blog post about a week or so ago describing his experience writing a kind of cool menu app. So you have some menu and it’s using all these fancy French words or Italian words or something, and you’re like, what’s a—god knows what. And so with his app— 

JOHN COLLISON:Aglione. 

PATRICK COLLISON: Exactly. Yeah. Is that like a fish or a pasta? 

JOHN COLLISON: Exactly. 

PATRICK COLLISON: So you point your phone at it, and it hallucinates, generates, renders an image of every description of every dish on the menu—very helpful. So he built this, and his reflection was having AI write the code for this was a quite frictionless and straightforward experience. But wiring together all the different services involved and the domains and the rate limits and the authentication and the project and the organizations and just that side of things was what was most frictionful in creating it.

JOHN COLLISON: Was he kind of arguing for much more MCP out there in the world? 

PATRICK COLLISON: Effectively. Clearly, there’s a need. Now that AI can generate the code, there’s clearly a need for some kind of higher-level orchestration. I sort of think of it as like terraform for Internet services that kind of doesn’t exist yet. And many of those linkages are going to involve some kind of transaction, some kind of commerce, some kind of billing relationship. 

And so we’re working with a number of companies today on figuring out how to enable this. And I don’t think it’s going to be the next Sessions before people see the results of this. I think you’re going to start seeing it in the coming weeks. But I think it’s going to be a really big deal. 

CLAIRE HUGHES JOHNSON: So MCP, some form of network for transactions— 

PATRICK COLLISON: Well, I don’t know if MCP is specifically the protocol but kind of in that conceptual direction of how do we enable these... 

CLAIRE HUGHES JOHNSON: It’s a little bit of remaking some of the way the Internet— 

PATRICK COLLISON: Yeah.

CLAIRE HUGHES JOHNSON: Yes. That is pretty interesting and important to work on. Okay. This one was a question I asked during my interview process. I did not get a satisfactory answer, so good luck, Lloyd. How does Stripe balance investment in existing products versus growth opportunities? I didn’t ask the second part about the non-sexy stuff. 

PATRICK COLLISON: Well, I just think it’s awesome that, implicitly, Lloyd thinks that all the things—like I don’t know that everyone in the world thinks that multi-currency financial accounts are so sexy. But clearly, Lloyd does. 

CLAIRE HUGHES JOHNSON: It’s all very sexy, right, payments, infrastructure. Yes. Okay. So we’ve got existing products that need investment and then new opportunities, new spaces. We just talked about one. How do we balance? What’s the secret there? 

JOHN COLLISON: [sighs] [laughter] Look—no, I sigh because I don’t think anyone in the world has a satisfactory answer. And you always end up with these, like, really dumb-sounding rules of thumb like the—you know, there’s the Geoffrey Moore one that everyone likes of the Horizons framework where it’s, you know, 70% of your effort on near-term stuff, 20% on soon-horizon stuff like that’ll, you know, pay off in a year or two, and then 10% on the super long-term stuff. 

And that’s both like, obviously—it was actually—it wasn’t invented by him. It was invented by McKinsey, and it sounds like a McKinsey type framework. And yet it kind of maps to what a lot of sensible companies do. Like the ratios seem roughly right. I would say my observation for companies is that it is always tempting to conflate staffing levels with relevance and importance of a project. And those two are not the same. 

Like, you can have something that’s really important for the future of your company. That should be a three- or four-person team until they get a really cracking, working version of it that people say, oh, obviously this works. I think, sometimes the temptation of companies is to give more resourcing to something because it’s important. It’s like, no, no, no. Because it’s important, you should get a really compelling working demo and get it launched in the hands of users and get them adopting it. But I feel like I see, you know, especially among the large tech companies, this really loading up of— 

PATRICK COLLISON: I think the ‘Moore/not-actually-Moore’ framework is a really bad framework because it’s insular. And it’s— 

JOHN COLLISON: Isn’t it too top down? 

PATRICK COLLISON: Well, and it’s not user-centric. It’s not customer-centric. Like, everyone at Stripe working on product development—like the way in practice we do this is we start with the customer requests, the customer feedback, the customer—you know, the perceived shortcomings in our products, or the $1 markets, as Jensen calls them, as in the markets that are —$0 markets, perhaps, is the phrase he used—but the markets that are [crosstalk]—yeah. The markets that are nothing today but can be enormous in the future. So much of what we build at Stripe is for a use case that is very nascent and very small today, but we have some conviction that it could be really large in the future. But I suppose my point is— 

JOHN COLLISON: What Lloyd is getting at is there’s like a subjective judgment at the end of the day between: there is a near-term customer request for like, "Hey, I would like to be able to do this with my Stripe account." And then, there’s a zero-dollar market idea where I don’t think crypto, when we started really kind of investing in some of the stuff that we demoed yesterday, I don’t think that merited the investment it got based on the number of customer requests. 

PATRICK COLLISON: But you know, Connect is a mature product. Connect has been in the—we introduced Connect in 2014, if I—maybe it was ’13, if I recall right. And so in that sense, it’s an existing product. But it’s also very much a growth product. It’s one of our fastest growing products. And a lot of the functionality that we’re adding, global payouts or whatever, is I think going to expand Connect’s applicability to many new use cases. 

Will mentioned yesterday that everything basically that we’re building across Stripe, we are making available in Connect so that others and you all can go and sort of embed it into your marketplaces. This is why I don’t like the framework. It’s not user-centric. I dispute the premise of the question. [laughter] 

JOHN COLLISON: So what you’re hearing here— 

CLAIRE HUGHES JOHNSON: So I warned Lloyd about this. So you dispute investment. How do you define investment? Is it number of people? And you dispute—Horizon seems terrible because it’s not user-centric or macro-centric, frankly. I mean, I think one of the challenges with the Horizon framework is it assumes you have a few years. Right now, I don’t think you have a few years. 

PATRICK COLLISON: It sounds like something invented by McKinsey. 

CLAIRE HUGHES JOHNSON: Well, and I think— 

JOHN COLLISON: It was. 

CLAIRE HUGHES JOHNSON: They, like, celebrate that. Yes. But— 

JOHN COLLISON: I think what we’re saying is maybe— 

CLAIRE HUGHES JOHNSON: But there are trade-offs that need to be made. 

JOHN COLLISON: Absolutely. No. And I think—look, what it boils down to is, I think the reason you guys are in positions of responsibility on this stuff is because this is the fundamental nuance of running a product or running a company. 

CLAIRE HUGHES JOHNSON: This is the part. 

JOHN COLLISON: And I remember asking Aaron Levie this question. He’s like, "Yeah. You make a big list of the things you want to do and [stack rank] them by which will have the most impact." But it’s, like, fundamentally a very subjective determination. And I think maybe we only have thoughts on what’s bad. You know, Patrick proffers that he thinks the 70-20-10 framework is bad. I will proffer that people try to get super financialized on this stuff— 

CLAIRE HUGHES JOHNSON: Yes. 

JOHN COLLISON: —and do payoffs. And what happens is you just like—you know, if you ask people to kind of justify the ROI, what you will get is projections that magically justify the ROI. And if you say, oh, it needs to meet a 20% hurdle rate, it will magically have a 21% hurdle rate. And we definitely went through a period at Stripe— 

CLAIRE HUGHES JOHNSON: We did. 

PATRICK COLLISON: —[crosstalk] The corporate version of statistical significance below, you know— 

JOHN COLLISON: It’s P-hacking. Yeah. Yeah. Exactly. 

PATRICK COLLISON: It’s corporate P-hacking. 

JOHN COLLISON: And so we definitely went through a phase of trying to be really quantifiable about this. And I think it was a complete waste of time. 

CLAIRE HUGHES JOHNSON: Didn’t work. Didn’t work. All right. Art. It’s the art. It’s actually the job. All right. 

JOHN COLLISON: It’s the job. 

CLAIRE HUGHES JOHNSON: That’s the answer. Okay. We’re going to—now a favorite,  speaking of user feedback—LPMs. So, in Southeast Asia, China, India, digital commerce runs on 700+ local payment methods. How are we thinking about offering these to merchants for online and in-person transactions? 

PATRICK COLLISON: Well, we announced yesterday that we now support 125 payment methods, which I think is the most that anyone supports. But more importantly, we’ve done two things. One, we’ve built a—we’ve kind of re-architected the platform internally so that we can add them much faster. And one of the things, I think, Will mentioned was that we’re now able to ingest a specification and have an LLM build an integration for payment methods. 

JOHN COLLISON: We’ve kind of built device drivers for payment methods. And I actually kind of spoke to some payment methods yesterday—I was in the expo hall—who are using this device driver to kind of plug into Stripe. 

PATRICK COLLISON: Exactly. And— 

CLAIRE HUGHES JOHNSON: To help them plug into us. 

PATRICK COLLISON: Exactly. And so more importantly, this is an idea that, again, if we were doing Stripe again, we’d have done even sooner. I remember talking about this in 2014 though we didn’t do it in 2014. 

JOHN COLLISON: Well, that was a good— 

CLAIRE HUGHES JOHNSON: We’ve been talking about this one for a while. 

PATRICK COLLISON: Exactly. But yeah, we have finally built third-party device drivers for payment methods where others can plug into Stripe. And so I hope in the not-distant future there are thousands of payment methods on Stripe. And just payment method coverage is no longer a topic of discussion because Stripe has become so ubiquitous. 

CLAIRE HUGHES JOHNSON: All right. So, we’re working on it. We have been for a long time. All right. But some hope there. All right. This one is very nice and specific. Patrick likes very specific customer complaints. And so we’ve got one here. Pending payments—are we looking to resolve pending payments more quickly? But this particular complaint is about SEPA. And Erwin says we wait, on average, nine business days for payments to settle. How do we feel about this? Are we trying to get this to settle quicker? 

PATRICK COLLISON: Not only do we like talking about specific points of product feedback, but SEPA—we had a board meeting last week. And SEPA was a topic of discussion at our board dinner, where these issues with the payment scheme were top of mind for the whole board. And it’s—SEPA’s just not designed for Internet transactions. 

JOHN COLLISON: For remote, low trust— 

PATRICK COLLISON: Exactly. 

JOHN COLLISON: —transactions in a way. 

PATRICK COLLISON: Yeah, and so it’s very—the fraud characteristics are really poor. The banks often haven’t built the kinds of fraud systems that they have built for card payments for SEPA payments. And the reason for the delays is primarily so that, you know, we’re trying to, you know, perform this balancing act of protecting the businesses because we don’t want people saddled with a whole bunch of SEPA reversals and, you know, chargebacks. 

So we’re trying to protect the business, but also we don’t want people waiting, you know, forever for their funds. So we don’t have an answer here yet. I think it’s possible. I shouldn’t talk about things we haven’t done yet, but I think it’s intrinsic to some extent to the payment method, and it’s top of mind. 

CLAIRE HUGHES JOHNSON: And that’s challenging, yes. Thank you. All right. We only have a few minutes, so we’ve got to speed through some stuff, maybe have a lightning round. So easy question, what’s the hardest challenge that Stripe is facing today? Speed—lightning answers, hardest challenge. 

JOHN COLLISON: I just think it’s the upending of how Internet commerce works, where I said in the talk. But we came from a very browser-centric world. And many of you are already—there are lots of things that you were previously doing in a web browser where now it just completely skips the web browser. We think commerce is next. It hasn’t happened. We think it probably will. 

And so it’s not really determined what that looks like, but it feels very Internet—like people always use, say, intranet in 1994. You know, that phrase has a pretty steady rate of usage on Google Trends, but that’s what it actually feels to me right now. 

CLAIRE HUGHES JOHNSON: Anything—we have another question that came in which is, what keeps you up at night? Is that the same answer? What are the threats to Stripe in the long term that you worry about? 

PATRICK COLLISON: I mean, intellectually, I worry about us becoming insular and bureaucratic and ossified and sclerotic and calcified. 

JOHN COLLISON: You’re talking about your posters, your poster campaign. 

CLAIRE HUGHES JOHNSON: Like super unattractive. 

JOHN COLLISON: Describe your poster campaign at the office. 

PATRICK COLLISON: Yeah, I mean, like, I—

JOHN COLLISON: —your motivational posters.

CLAIRE HUGHES JOHNSON: Motivational posters. This is very motivational.

JOHN COLLISON: Well, just briefly describe them. 

PATRICK COLLISON: Hubris is such a problem for organizations that—and look, Stripe is not the world’s biggest or best or anything. But just, like, you know, as soon as anything gets any traction, it’s so easy to have it go to your head. And so we’re putting up posters around the office that I think are very inspiring of companies that were really successful and at the apogee and peak of some industry that are now bankrupt. [laughter] 

CLAIRE HUGHES JOHNSON: “The companies that have failed” posters. 

PATRICK COLLISON: But I think it’s really important to understand that, unless we are agile, unless we are intensely focused on serving you guys more effectively and on being the best way for you all to grow your businesses, like, Stripe has no right to exist. You know, we could be dispassionately hauled away by the garbage collector of capitalism. So intellectually, that’s the thing I worry about. But emotionally, the thing I worry about like when I’m actually lying awake in bed, it’s the security, reliability, compliance, just like the treasury—

JOHN COLLISON: Keeping the “most reliable” part. 

PATRICK COLLISON: Yeah. Exactly. People at Stripe just take so seriously the custodial obligation of the businesses and the downstream people that we serve. 

CLAIRE HUGHES JOHNSON: Responsibility. 

PATRICK COLLISON: We mentioned the 1.3% of global GDP stat. And on some level that’s cool, but it’s also a crushing obligation and to just make sure that Stripe works properly every day. 

JOHN COLLISON: I think, on your first point, there’s a market structure aspect to it where—part of the reason we worry about that is—Warren Buffett, who just stepped down obviously after his 55-year run running Berkshire, talked about wanting to invest in ham-sandwich businesses where they could be run by a ham sandwich, and they’re kind of on autopilot. I think he specifically used Coca-Cola as his example. And I think the Coca-Cola CEO was really offended when that comment got back to him. 

But you know, Coke—everyone knows Coke, and they just do their polar bear ads and everything. And everyone knows it, and the whole business works whereas the business Stripe is in is being the best way for people to operate their businesses online. And we sell to businesses. We sell to very informed buyers, who have lots of other options for every single thing that we want to—like for every single service that we offer. 

And so as a result I think that’s what keeps one feeling fresh definitely. It’s because you know that, ultimately, it’s only by having the most useful product in that space that people will actually stay coming to us. And I think we feel that very acutely. It is not quite a ham-sandwich business. 

CLAIRE HUGHES JOHNSON: Not a ham sandwich, a responsibility and a pressure. I think that’s a good thing to end on, also with a thank you to everyone who came to Sessions 2025. This is our last main stage session. But there’s a lot of breakouts happening today, I think, into the afternoon. And we’ve just opened registration for Sessions next year, which is April 28 to 30, 2026. 

We actually—yeah. This is exciting for me because announcing something that was happening a year away is something neither of these guys likes to do. So I’m thrilled to end with that power move. And thank you all. Have a wonderful rest of your day. Thank you, John and Patrick. 

PATRICK COLLISON: Thank you.