Inside Cursor: The Anatomy of a Generational Startup

0:00We don't need to compete with Anthropic and OpenAI on models right now. The interface between the human and the model is the key thing. If you looked at the competitive landscape, it was almost silly. I asked Michael, I was like, what do you think about Cloud Code? And he said something to the extent, look, we are going after the biggest market in the world. You're always going to have favorable competitors. That does not scare us. As founders, we all want to be ambitious and we want to push to the maximum point of Pareto Frontier, but you have to kind of know what the curve is. For us, the decision to invest in CurseRate was actually pretty obvious.

0:30Remember Andre Karpathy was using it. It was clearly a phenomenon. It was clearly like a known brand. They went and got all the users in record time. Now they have the asset, the data, the know-how to build their own models. You really couldn't do the flip of that unless you started as a frontier lab. They were definitely a couple of old moments. Like, remember when? Two years ago, betting on an independent AI coding company looked almost irrational. Microsoft owned GitHub, VS Code, Copilot, and had access to some of the world's best AI models. And yet, Cursor broke through. In this episode, Martin Casado, Sarah Wang, and Matt Bornstein look back at the decisions that shaped Cursor from its earliest days, building a standalone product instead of a plugin, resisting the push into enterprise sales too early, and staying relentlessly focused even as the competitive landscape changed around them.

1:24They also unpack how the company repeatedly reinvented itself as AI advanced, built a culture around speed and taste, and eventually applied the same intensity it brought to product to hiring, sales, and M &A. It's a conversation about cursor, but also a broader look at what it takes to build a defining company in a market where the technology, competition, and conventional wisdom change almost overnight. So to kick off, I actually wanted to take us back to early 2024. And obviously, you guys led the deal in, I think it was May of 2024 that year. But can you walk us through what was going on in the Q1, Q2 of that year, and maybe how you guys started working with Cursor well before we ever partnered officially with them?

2:16So you have to sort of transport yourself back to late 2023, early 2024. I went back and looked it up. The leading models at the time were GPT-40, Cloud 3, and Llama 3. That feels like forever ago. Llama 3? Yeah. That was actually pretty good. That was great. Yeah, it was a great model. Change in strategy, obviously. Now, the leading coding harness was Copilot by like a mile, right? This was like the consensus thing. Microsoft was doing their Microsoft thing, right? They had Copilot. They had Office. They had VS Code. All of these things combined. They were going to somehow win AI. And so it was clear that AI coding was sort of working, but it was a different universe compared to today, right?

2:52Like you had the sort of early signs that you could do meaningful work with AI, but agents didn't really exist. Loops didn't really exist. Reasoning didn't really exist at the time. So it's sort of a crazy time. There was a continuum of like companies going after coding, right? So on one hand, you had those that are like, the only way to win coding is to build a base model, and it has to be a coding-specific one. And on the other hand, you had, very presciently, by the way, companies just focus on agents, right? And like Cursor was somewhere in between. And not only models, there were a million companies doing plugins.

3:29And both of these were very reasonable things to do, done by very smart people, right? Like that's what's so interesting about these new spaces. It's like smart entrepreneurs show up and do really interesting things. And in this case, model training was a very reasonable thing to do because the models weren't that good at coding yet. And there was sort of a theory that if we focus on this into a coding-specific model, that's the breakthrough. People doing plugins on the other side were sort of saying, hey, let's not toss out the whole IDE. That would be a crazy thing to do. But the interface between human and model is really important, so plugin is a great place to go.

4:01You know, it just kind of turned out that the thing that Cursor was doing was the right thing to do, right? Like us, they were very bitter lesson-pilled. At the time, they would have said, oh, there's no need for us to go train our own model. Of course, that changed later for different reasons, which we're going to get into later in this podcast. But it's like, we don't need to compete with Anthropic and OpenAI on models right now. The interface between the human and the model is the key thing. And this is something that I remember Michael and Amon saying a lot and just saying very clearly from the earliest days of Cursor, saying code in the future will look like pseudocode.

4:35I remember Michael making this point a lot. Like when you look at old like computer science papers, there's a section. Here's the algorithm, not in code, but in like pseudocode, which I'm sure many people are used to. And we don't exactly write in pseudocode now, but probably you could. Probably you could just take a pseudocode block from an old paper and like the models now would implement that. This is like really what the cursor guys thought from the very beginning. And they've proven to be exactly right about this, right? It's like this sort of natural language-ish specification, like pairing the programmer's intent down to the minimum possible spec.

5:05That's what really sort of matters now. So anyway, we were in this world, as Martin said, we were talking to a ton of very smart, very interesting people working on coding in various ways. The cursor team kind of stood out because not only did they sort of align with the things that we thought about the world, more importantly, smarter people than us and engineers working. At the time, the leading AI companies like OpenAI, MidJourney, Replicate, a bunch of others all kind of used this and subscribed to this model. And the team, frankly, just seemed kind of special, right? They had an unusual degree of focus.

5:35I remember we got Michael to come pitch our GP group. And 90 % of the meeting was him saying no to things, which is just like for founders out there working on products that you love. Like VCs will always ask you kind of dumb questions about other things you might do or things you're related to or whatever. And Michael literally just sat there and like very politely listened to all the questions. And said, hmm, interesting. No, we're not going to do it. Wait, wait. Actually, on that point, can I ask you guys a question? because I remember that meeting so vividly when you brought Michael in for the A.

6:05And one of the things that he said no to was actually, and to your point on there were a lot of plugins, right? And so there was this maybe more consensus thought that a VS Code fork versus the plugin into VS Code would, one would be better for enterprise and it wasn't the former, it would be the latter. And I recall, I think it was you, Martine, or maybe it was both of you asking him, well, are you going to get enterprise customers with this medium, right? And he looked at all of us in the eye with high conviction. And, well, I'm going to turn it back to you. And what did you think of his answer?

6:40And what gave you guys the conviction to take a pretty contrarian bet at the time? They actually had a very reasoned articulation for why, where they run the design spade makes the most sense. So, for example, why wouldn't you build a coding-specific foundation model? So what they said at the time was, it turns out that you don't speak to these models in code. You speak to these as humans in natural languages. Therefore, the model has to understand the breadth of human experience and language, right? And so that means in order to build a coding model, you're actually building a frontier language model.

7:12And that's just a very hard thing to do. And then, of course, there are other companies doing it. And there are much more focused problems like TAMC completes to do. So I think every one of their decisions was very reasoned, in our opinion. They were also very product focused, which you actually don't see a lot in AI. So they really believe that you can build a product, the product, if it's really good, will be adopted, which, by the way, is very similar to how Elon thinks. And that dictated a lot of their decisions. If you really believe in the product, you would never do a plug-in because then you're part of somebody else's product.

7:40If you really believe in the product, you wouldn't do enterprise early, you do it later because the product itself is to kind of go to market motion. And so there was, I would say, and Matt, let me know if you would agree, tremendous clarity on exactly what type of company they were and the decisions that they said all fell from that consistently. Where many of the companies we saw were kind of these pastiches of, we'll try all of these different things. We're not quite sure what's going to work. So their plugins and their services and their whatever, whatever, whatever. This was a product company going after what they considered to be a product problem.

8:11Yeah, no, I think that's exactly right. And I think they had the ambition and the courage to kind of do the maximal form of what the product would look like, right? I think, you know, as founders, we all want to be ambitious, right? And we want to push to the maximum point, the Pareto frontier. But like, you have to kind of know what the curve is. And they just kind of knew what the curve was, right? And then get distracted by the sort of non-product things. I remember when we were trying to convince the Cursor team to let us invest in their company, I asked Michael once what he did for fun outside of work.

8:40And I was like, he didn't understand the question. He's like, what do you mean? I'm like, when you're not in the office. He's like, what do you mean? I also remember, by the way, when we had the lunch after we did the investment, and I was talking about something about internet days that were similar. He was like, I was five.

8:57And we're like, oh. But I mean, what's amazing about not just Michael, but all of the founders is even though they were pretty young when a lot of these things happened, they actually were kind of students of all this history, which is a trait that we see, Al Trampel talks about this a lot. We see this across all of our top founders that they're like able to study and really learn from what's kind of happened before. I'm going to turn the question back on you. I think Matt would agree. For us, the decision to invest in Curse Rathay was actually pretty obvious. The team was phenomenal. It was incredibly focused.

9:24All of us were users. Half the team actually used Cursor. Remember Andre Karpathy was using it. It was clearly a phenomenon. It was clearly a known brand. The growth was asymptotic. So from an early stage, investors are pretty straightforward decision. It was a much less obvious decision for a growth investor. And so like, I think probably what you heard in that first meeting, maybe it was a little bit dissonant with how you'd growth invest, but then you very, very quickly led the next round. So maybe talk through how you kind of got past the typical growth evaluation and got conviction. Yeah, I mean, I think the timelines are so compressed here, it's crazy to think back on, because I remember you guys led the A in May, June, right, early summer.

10:06And by October, we did the B. And so it was clearly just vertical liftoff. And there were a couple of things that summer, right? You guys talked about Andre Carpathy tweeting about them, but they went on Lex. Incredibly thoughtful discussion of how they thought about their vision for coding going forward. And so I think it comes back to, Martine, something you said, which is they were just reimagining everything. Right. And, you know, you just talked about this as well, Matt. But you can't you can't write growth memos that way. Never mind you. So, you know, you've got the bug, you know, because you were calling us and we were so excited.

10:44How did you convince David George? Like, don't you have to show like, you know. Yeah. You know, it's so funny because and I think you had to think out of the box for this one. I mean, of course, the momentum was vertical, but there were so many things that they defied conventional thinking where I remember I the first time, you know, Martin, you introduced me to Michael. We got a coffee that day, you know, as Mike, you know, they move quickly. So I pulled some stuff together for them. Right. Because we're, you know, we're trying to show them, hey, this is the next stage of what growth looks like.

11:16And there's so many conventional pieces of conventional wisdom that they just completely broke. Like we would say, hey, we'll help you find a sales leader because, you know, previously it made sense to start layering a sales leader at 25 to 50 million of ARR when self-serve starts to peter out. Michael looks me dead in the eye and is like self-serve is not petering out. Right. And so then you start to pull these drivers that you're like, oh, these assumptions that would go into any growth model historically. You can't have that asymptoting at 25 percent growth in five years. Right. And so I think a lot of the growth investment decision at the time was throwing out some of the assumptions that we would typically make on how a company would, you know, eventually start to slow growth.

12:02I'm very glad that we did that. And then the second piece is we should talk about the competitive landscape. Right. Because you mentioned Copilot. I was actually going to say, and actually, I think Matt deserves a ton of credit here, which is what we talk about now in retrospect, how it was a pretty easy decision on the Series A. But, you know, Matt led the early deal and he saw something a lot of people didn't, which was this, which is if you look to the competitive landscape, it was almost silly, which is you had a large incumbent, which is Microsoft, that had a company at scale that had the opening I wait and they own VS Code.

12:36so everything you know everything would say that there's just no way that you could survive relative to this this one single competitor we'll talk about the other ones um later and i and i think that also was the thing you said about michael was what i witnessed it'd be great to hear matt which um i even remember when matt said this early on he was like you know they have a good answer for everything which when you have founders that are students of what they do they're very consistent in their answers. And, you know, and this is not, I think when we did the deal, it was pretty obvious, but it wasn't so obvious six months before.

13:13You know, and Matt and the team had that deep conviction way back then. And I think for every one of these races, there are mirrors of that. Anyways, is this kind of fun talking? Yeah, no, totally. And look, I mean, people think venture capital is like an individual sport. Like, it's not at all, at all. I mean, our whole team was working on this at the time. You know, I think it was over a year. We literally had a cursor team on our team. I remember that, yeah. Marco was great. Rachel was great. Yeah, yeah, yeah. You know, you led the team. I mean, it was great. So, but yeah, no, you're exactly right.

13:43I mean, these guys are crazy smart, right? They're like off the charts, sort of just raw intelligence. Yeah. They're able to do this thing we talk about a lot, like context switch between tech and markets really, really seamlessly. and they like ingest data, right, whether it's historical data or like things happening in the market and act very, very quickly, which you said before, Sarah. But I agree with that, but specifically, it was just kind of crazy to think you're going to fork something from Microsoft who owns the thing and then you're going to, you know, compete with them with smaller models even they own the opening eye weights and they have enterprise, Salesforce, and they have...

14:24The greatest, yeah, enterprise distribution. Yeah, ever. You know, it's so interesting, right? It's like... They have GitHub. Yeah, they have GitHub. Just 100 million developers. They literally own every piece of that. You know, it's so interesting, right? Because, like, that's definitely the conventional wisdom. And, like, it's usually correct, the conventional wisdom in many of these cases. You know, one of the data points at the time was actually Marco and our team set up a dinner where Amon sat right next to John Schulman. And they talked, like, a lot about Cursor. and I think John's thing was like, oh yeah, I tried Cursor, but like our code base, I think he was at Anthropic at the time, like our like monorepo just like immediately killed it, right?

15:06It's like I couldn't even load the code base and like I think like the next day Amon came back, he's like, oh, we made some changes, like here you can look. And not only is it like speed of iteration, but like, you know, we just look for these little data points where it's like actually like doing a plugin or just like writing on top of VS Code is clearly not going to work because like the smartest people in the world have tried it and it doesn't work, right? So it's like, so these guys were sort of listening to it. And historically, we as a team have always bet on the independent in a large market, right?

15:35If there's a leading independent, we'll always bet even if the incumbent seems very scary. I think it's worth talking about competition because my experience with all of our experience with Cursor is that there was always like a looming existential threat and we always forget about it. But like, you know, of course, Copilot was the big one. And then remember the whole Windsurf thing? So, like, Windsurf actually did this very smart thing where they went to YC and there was a lot of YC users using it. So, you kind of had this kind of whole movement on Windsurf. And, you know, Cognition was a phenomenal company.

16:05Actually, it was making real waves with agents and they've done a great job of that. And then you had the first version of Cloud Code that came out. And, like, that was kind of like, you know, a competitive thing. And so, like, you kind of go through, like, there was always some competition. And one thing I would say, and I would love, again, to actually hear your perspective on it, is like, it is such a pleasure to work with founders that are entirely unfazed. You know, like normally, like, you know, founders of this thing's happening are calling and they're kind of freaking out and whatever.

16:35They were 100 % unfazed by the competition. Yeah, paranoid but unfazed. I'll never forget, you know, and we've been lucky to do, I think, all of their rounds starting from the A. Is that true? Did we invest in every cursor round? We did, yeah, with increasing amounts of dollars. Sarah co-led the B and the C. That's amazing. That's real conviction. That's amazing. That was a harder conversation to convince the team for the C. You know, at the B, you could argue, God, just bet on vertical momentum. That was actually obvious. Who grows from 4 to 50 in like four months or whatever. But the, or actually no, it was the D.

17:14So the Cloud Code, not to take us on a time travel trip again, But Cloud Code came out May 2025 with a splash. You know, it wasn't, the pickup wasn't right away, but obviously formidable. And I remember there was around, I think a couple months after, three months after that. And I asked Michael, I was like, what do you think about Cloud Code? And he said something to the extent, actually, you know, exactly what you're saying here, which is, look, we are going after the biggest market in the world. There's going to be competitors. In fact, there's been a rotating cast of characters from the very beginning.

17:51Microsoft Copilot was the first one. Claude Code is one of a rotating set of characters. And, you know, we think in big markets, you're always going to have formidable competitors. That does not scare us. and you know I just remember being somewhat blown away by the clarity of that comment the lack of fear right there's humility as well like it's humility mixed with bravado and I just think that is kind of the winning combo when you're going after these juggernauts like that. And I know it sounds vague but like things like this it's almost a Bezos thing where it's like if you're smart and right like it actually like gives you I think the confidence and the focus to like have that kind of attitude when when you have this like mega competitor sort of coming after you.

18:35Yeah. That I'll say, I will say, they were definitely a couple of oh fuck moments, but they were always around models. Like remember when Opus 4 or 5 came out? Like we're all on vacation and I think everybody in the world was like, oh fuck. That's true. You know, these things are moving much, much faster than any of us thought. And then I think the same thing when the pre-release of Mythos came out and people had access to it. These models were getting so good so fast. And I think another thing that the Cursor team probably doesn't get, you know, the level of credit for is to the extent they cannibalized themselves so dramatically in like a Reed Hastings type way in a matter of two years, right?

19:19So they were first an IDE and then they moved to an agent platform and then they moved to a model platform. And it's very, very tough for, you know, founding teams to actually have to discipline. Yeah, tab was the big thing. You know, Karpathy had that famous post. I guess he has a lot of famous posts where he's like, tab, tab, tab. Yeah, exactly. Like that kind of doesn't matter that much anymore. Yeah. I actually think that's a good segue to maybe some of the other strategic decisions they made along the way around business model. And to your point on oh fuck moments, I don't know if they would consider this that because I don't think the vagaries of Twitter impact them as much.

20:02As the investors. Yeah, as the rest of us just like scrolling all day. But it was probably summer of 25 where there was a lot of shit written about their gross margins, etc. And then, of course, you know, post the Opus 4-5, maybe the January period. I mean, we should all talk about that because that was kind of those are both different but interesting times. Yeah. I mean, one thing I'll say is I just feel like X is driven by like tail chasing, adrenaline sinking in schadenfreude. And there tends to be a massive disconnect by what people say on X and like how the business is actually doing. And so even though there was a lot of this, the business tended to do very well.

20:46and it's so interesting is so much of this often comes from investors, especially things talking about like business quality. But tech transformations always proceed figuring out the business. Like you guys remember the internet? Like we didn't even figure out how to monetize it for a long time. Forget margins. Making any money at all, right? And here you've got like unlimited demand, unlimited growth and margins. So, you know, I think this tends to be, you know, A, X is its own kind of echo chamber as people say, but also B, every single wave you have investors trying to say something that sounds smart, which is like picking apart, you know, the business model of a nascent tech transformation.

21:23And I just think historically, they've always been proven wrong. Yeah. Yeah, for sure. I mean, you know, I think I'm going to borrow this quote from something you said, Martine, but, you know, first comes the technical transformation, then comes the business model. And in this case, like, I think the strategy where they went and got all the users in record time, now they have the assets, the data, the know-how, et cetera, to build their own models, right? You really couldn't do the flip of that unless you started as a frontier lab. And it, you know, it's clearly paying off in this combo that we'll talk about later.

22:01But I think seeing that kind of backdoor strategy was really inspiring and they took a lot of heat for it. So Martine, how do you think Cursor really became a sales-driven company or a company that's good at sales? Because, you know, early on, I think I told Michael a few times, too, it's like, oh, you better invest in sales. He's like, nope, not going to do that. They literally set up enterprise at cursor.com. It's like, if you're an enterprise, send your inquiries to this. But like, but obviously now they're very good at sales, right? And go to market in generally. Like, how do you think they made that jump?

22:30You know, it's a great question. So here'd be my best guess, which is they are very much a student of what they do. They're very thoughtful. Every raise, you know, every product decision. And I think as they learn more about the market, they realize that the margins were in the enterprise. So very classically, if you do your competitive set, say the large labs, the large labs will basically subsidize or give away the single-served tier. But then they actually have, they command great margins in third party and they command great margins in the enterprise. And, you know, of course, it was really a scale.

23:01And as the company has always gone, they decided, okay, like, so, you know, for future viability, we need to go into the enterprise. That's where A, the bulk of the spend is, but also a lot of the margin. and the margin tends to be the value here. We've created a great, you know, top of funnel engine. And so, you know, this is existential and this important. And then when they decided to do it, they really went in. And that may be the fastest route of a sales team in the history of the planet, right? I mean, you know, you were very involved, Sarah, for example, on recruiting. So maybe talk about that actual expansion because it was.