Inside Cursor: The Anatomy of a Generational Startup

Key Takeaway

When building in a crowded market, choose a narrow product thesis and use it to say no to distractions. Write down the one user problem your product must solve better than anyone else, then review every feature, channel, partnership, and hire against it this week. Cursor’s advantage was not avoiding competition; it was committing to a complete product experience, iterating rapidly on user feedback, and delaying adjacent opportunities until they reinforced the core.

Episode Overview

A16Z investors discuss why they backed Cursor early and how the company emerged in an intensely competitive AI coding market. They highlight Cursor’s product-first conviction, rapid iteration, willingness to cannibalize earlier products, disciplined hiring, and expansion from self-serve adoption into enterprise sales and strategic acquisitions.

Full Transcript

Transcript of Inside Cursor: The Anatomy of a Generational Startup from A16Z. Auto-generated from episode audio; may contain minor errors.

We don't need to compete with Anthropic and OpenAI in the field of models right now. The interface between the human and the model is the main thing. If you look at the competitive environment, it's almost pointless. I asked Michael what he thought of Claude Code. And he answered something like this: " Listen, we're entering the largest market in the world." You will always have formidable competitors. It doesn't scare us. As founders, we all want to be ambitious and maximize on the Pareto frontier, but we need to understand what that curve looks like.

For us, the decision to invest in Cursor was pretty obvious. I remember how Andrey Karpaty used it. This was an obvious phenomenon. It was already a well-known brand. They engaged all users in record time. They now have the assets, data, and knowledge to build their own models. Doing it the other way around would be impossible if you didn't start as a cutting-edge lab. There were definitely a few moments. Remember how I To begin, I would like to go back to the beginning of 2024. And as you know, you led the deal, I think, in May 2024.

But can you tell us what was happening in the first and second quarters of that year, and how you started working with Cursor even before we officially became partners? You need to move to the end of 2023— beginning of 2024. I checked it. The leading models at the time were the GPT-4o, Claude 3, and Llama 3. Wow, that seems like an eternity ago. Llama 3? Yes, yes, yes. She was actually pretty good. Wonderful. Yes, it was a great model. We are changing our strategy, obviously, now.

The leading coding tool was Copilot, by a wide margin, right ? This was a common choice. Microsoft was doing their job, right? They had Copilot, they had Office, they had VS Code. All of these things together—they had to somehow win in AI. It was clear that AI coding worked, but it was a completely different world than today, right? You saw the first signs that meaningful work could be done with AI, but agents didn't actually exist back then, cycles didn't exist, reasoning didn't exist either. So it's kind of , you know, a crazy time.

Although there was a whole range of companies involved in programming, right? On one hand, there were those who said, " The only way to win at programming is to build a base model, and it has to be specialized for the code." And on the other hand, you had, by the way, very visionary companies focused only on agents. Is that true? And Cursor was somewhere in the middle. Sorry. And not just models, there were millions of companies making plugins. That's right. Yes. And both of these approaches were quite logical, implemented by very smart people, right?

Here's what's most exciting about these new areas. Smart entrepreneurs are emerging and doing really interesting things. In this case, training the models was a very smart move, because the models weren't that good at code yet, and the theory was that if we focused on that and created a dedicated model for the code, that would be a breakthrough. Those who created plugins, on the other hand, seemed to be saying, "Hey, let's not throw out the whole IDE." That would be madness. But the interface between the human and the model is really very important.

"That's why the plugin is a great choice." Well, you know, it just turned out that the path Cursor took was the right one, right? They, like us, were supporters of the " bitter lesson," you know, then they would say, "Oh, we don't need to teach our own model." Of course, this changed later for other reasons that we'll talk about later in this podcast, but the bottom line is that we don't need to compete with Anthropic and OpenAI in models right now. The interface between the human and the model is the main thing.

And this is something that , as I recall, Michael and Aman were very clear about from the very early days of Cursor: code in the future will look like pseudocode. I remember Michael often emphasizing this. When you look at, you know, scientific articles in computer science, there's a section: "here's the algorithm not in code, but in pseudocode," which I'm sure many people are used to, and, you know, we do n't exactly write in pseudocode now, but we probably could. You could just take the pseudocode from an old article and modern models would be able to implement it.

This is exactly what the Cursor developers had in mind from the beginning. And they proved that they were absolutely right. Is n't that right? This is a kind of natural language description, a reduction of the programmer's intentions to the minimum possible specification. That's what really matters now. So, whatever . We live in a world where, as Martin said, we interact with a lot of very smart and interesting people working on programming. The Cursor team stood out because they not only shared our view of the world.

More importantly, smarter people than us—engineers who were working at leading AI companies at the time, like OpenAI, Midjourney, Replicate, and others —all took this approach and followed this model. And the team, frankly, seemed special. They had an unusually high level of concentration. I remember when we invited Michael to present his project to our GP group. And he spent 90% of the meeting giving up unnecessary things. This is a useful lesson for founders: venture capitalists will always ask stupid questions about what else you could be doing or what it’s all about.

And Michael just sat there, very politely listening to all the questions and answering: “Hmm, interesting. "No, we won't do that." Wait, wait. By the way, can I ask you a question? Because I remember so vividly that meeting when you brought Michael in, and he rejected, in particular, the idea with plugins. There was an opinion that a VS Code fork, unlike a plugin, would be better suited for the corporate segment, but it turned out that this was not the case. I remember you, Martin, or maybe both of you, asking him, "Well, will you get corporate clients with this tool?" He looked each of us in the eye with great confidence, and you know , I'll take that question back to you.

What did you think of his answer? And, um, what gave you guys the confidence to make a fairly, I would say, fairly controversial bet at the time? They actually had a very well-founded explanation for why their place in the design space was the most logical. So, for example, um, why don't you create a basic model that specializes in programming? So what they said then was: it turns out you don't talk to these models with code. You talk to them like humans, in natural languages. So the model has to understand the full breadth of human experience and language, right?

And so, you know, that means that to build a model for coding, you're actually building a front-end language model, and that's very difficult to do. And then of course there are other companies that do this, and they focus on much narrower problems, like code autocompletion. So, I think every decision they made was very well-considered, in our opinion. They were also very product-oriented, which is really rare in the AI ​​field. So they really believe that you can create a product, and if it's really good, it will be accepted, which, by the way, is very similar to how Elon thinks.

And this determined many of their decisions. For example, if you truly believe in a product, you would never make a plugin because then you become part of someone else's product. If you truly believed in the product, you wouldn't be doing the corporate segment in the early stages, you would be doing it later, because the product itself is the way to go to market. And so there was, I would say, and Matt, let me know if you agree, a tremendous clarity about exactly what type of company they were, and all the decisions they made consistently flowed from that.

Okay, whereas a lot of the companies that we saw were kind of like, "we'll try this, we're not really sure if it's going to work," you know, so it's plugins, services, and stuff like that, they were a product company solving what they thought was a product problem. Yes, no, I think that's absolutely correct, and I think they had the ambition and the courage to make the most complete form of what the product would look like, right? I think as founders, we all want to be ambitious, right?

And we want to squeeze the maximum out of the Pareto frontier, but we need to understand what that curve looks like, and they just knew what it was, you know? And don't get distracted by things that aren't related to the product. I remember when we tried to convince the Cursor team to let us invest in their company. I once asked Michael what he did for fun, well, outside of work. And it felt like he didn't understand the question. He's like, "What do you mean?" I say, "When are you not in the office?" He said again: "What do you mean?" I also remember, by the way, our lunch after we formalized the investment.

I was talking about some things from the early days of the Internet that were similar. And he's like: " I was five then." But you know what's amazing — not just about Michael, but about all the founders — is that even though they were quite young when a lot of things were happening, they actually studied this history, and that's a trait that we see a lot. Al Trapp Hill often talks about this. We see this in all of our best founders: they are able to study and learn from the lessons of what has come before.

Now I will forward this question to you. I think Matt would agree that the decision to invest in Cursor was pretty obvious for us. The team is simply phenomenal. She is incredibly focused. We were all users. You know, half of our team actually used Cursor. Remember, Andriy Karpaty used it. This was definitely a phenomenon. It was already a well-known brand. The growth was asymptotic. So for early-stage investors, the decision was quite simple. But for a growth-stage investor, the decision was much less obvious. So I think what you heard in that first meeting may have been a little dissonant with how you typically invest in growth.

But then you took the lead in the next round very quickly. So, maybe you can tell me how you overcame typical doubts about assessing growth and what you learned. Yes, I think the time frame here is very tight. It's even strange to remember, because I remember you led Round A in May- June, right? At the beginning of summer. And already in October we did Round B. It was a real vertical takeoff. A few important things happened that summer, right? You guys were talking about Andriy Karpaty tweeting about them, but they went on Lex, uh, an incredibly thoughtful, uh, discussion about how they see their vision for the future of programming.

Uh, and so I think it goes back to Martin, to what you said, which is: they were just reimagining everything, right? And you, you know, you were just talking about it too, Matt. But you can't, you ca n't write growth reports like, "Okay, forget it, you, you know, you caught this bug because you called us and we were so excited." How did you convince David George? And don't you need to show, like, well, you know? Yeah, you know, it's so funny because I think you had to go beyond your usual thinking to do that.

I mean, of course, the dynamics were vertical, but there were so many things where they went against the grain that I remember the first time, you know, Martin, you introduced me to Michael, we had coffee that day, you know, this is Mike, they move fast. Uh, so I put something together for them, right? Because we, uh, you know, we're trying to show them, "Hey, this is the next stage of what growth looks like." "And there are so many accepted truths that they just completely destroyed." For example, we would say, "Hey, we'll help you find a sales manager." " Because, you know, it used to be logical to start bringing in a sales manager at 25–50 million in annual revenue, when self-service starts to fade away." Michael looks me straight in the eye and says, “ Self-service is not going away.” "Yes?" And so you start using these levers, thinking, "Oh, these assumptions that have historically been built into any growth model, you can't let this stagnate at 25% growth in 5 years." "Yes?" And so I think a big part of the decision to invest in growth at that time was to let go of some of the assumptions that we typically made about how the company would, you know, eventually start to slow down in growth.

Uh, I'm really glad we did it. Uh, and the second point: we have to talk about the competitive environment, right? Because you mentioned Copilot. I was actually going to say, and actually I do, I think Matt deserves a huge amount of respect here, which is what we're talking about now, looking back. It was a pretty easy decision for a Series A, but, you know, Matt led that early deal, and he saw something that a lot of people didn't see , which is that if you look at the competitive landscape, it almost looked like a no-brainer because you had a big incumbent, Microsoft, a company of that scale that had the clout of OpenAI and owned VS Code.

So, everything indicated that it was simply impossible to survive alongside this one and only competitor. We'll talk about the others later. And I think one of the things you said about Michael is something I witnessed, and it would be great to hear Matt, I even remember when Matt said it at the very beginning. He said, "You know, they have a good answer for everything." When you have founders who study their business deeply, they are very consistent in their answers. And you know, it's not... Listen, I think when we made the deal, it was pretty obvious, but 6 months before that— not at all.

Matt and the team had this deep conviction back then. And I think each of these funding rounds is a reflection of that. Anyway, is this guy real? No, completely. And look , people think venture capital is an individual sport. This is not true at all. I mean, our whole team was working on it at the time. I think it lasted over a year. We literally had the Cursor team as part of our team. Michael was wonderful. Yes, yes, yes, yes. You led the team. I mean, it was great.

So yes, no, you are absolutely right. I mean, these guys are incredibly smart, right? Their intelligence is simply off the charts. Yes. They are able to do what we keep talking about— switch between technologies and markets very easily. And they absorb data, right? Whether it's historical data or market events, they act very quickly, as you already mentioned, Saharan. But I agree with that, however, in particular, it was just crazy to think that you were going to fork Yes. something from Microsoft, which owns this thing. And then you're going to, you know, compete with them with smaller models, even though they have the OpenAI scales, they have the corporate sales force, and they have the best, yeah, corporate distribution.

Yes, the best in history. You know, it's so interesting, right? This, this is like they can, yes, they have you 200 million developers. They own everything. They literally own every part of it. You know, it's so interesting, right? Because that's definitely a common opinion, and it's usually correct in many such cases. Um, you know, one of the facts at the time was that Marco from our team organized a dinner where Aman sat next to John Shulman. And, um, they talked a lot about Cursor. And I think John said something like, "Oh, yeah , I tried Cursor, but our codebase," I think he was at Anthropic at the time, "our monorepository just killed it instantly , right?" It's like I couldn't even download the codebase.

And it seems like the next day, Aman came back and said, " Oh, we made some changes, here you go, take a look." And it's not just about the speed of iteration, but also the fact that we were just looking for these little facts where it turns out that making a plugin or just writing on top of VS Code—that's clearly not going to work, because the smartest people in the world have tried that, and it doesn't work, right? So these guys were actually listening to us.

And historically, as a team, we've always relied on independent players in a big market, right? If there is a leading independent player, it will always be there, even if the competitor seems very scary. I think it's worth talking about competition because my experience and all of our experience with Cursor shows that there was always a kind of looming existential threat, and we always forgot about it. But of course, the big problem was the Co-pilot. And do you remember that whole story with Windsurf? So Windsurf did a very smart thing by going to YC, and there were a lot of YC users using it , so there was a whole movement around Windsurf.

And you know, Cognition was a phenomenal company that really made a splash with agents, and they did a great job with it . And then, yeah, the first version of Cloud Code came out, and it was, well, you know, kind of a competition. And that's how you go through it, meaning there was always some competition. And I would like to say one thing, and again, I would be interested to hear your opinion: it's nice to work with founders who are not embarrassed by anything. Yes. Usually, the founders of such projects have their own calling, they get nervous and stuff.

And they were not at all afraid of competition. Yes, paranoid, but steadfast. I'll never forget, and we were lucky enough to participate in, I think, all of their rounds, starting with the Series A, but the one that... Did you invest in all of Cursor's rounds? We invested, yes, I think, with increasing amounts, but... Sarah was a co-lead in the C round. It's incredible. This is true confidence. It was a tough conversation to convince... I wanted to say, round C. the team about round C. You know, in round B you could say, "Oh my God, just bet on vertical growth ." This year it was obvious.

Who goes from four to 50 in four months or something, but... No, actually, it was round D. So, Quad Code—I don't want to do a time travel thing again—Quad Code came out in May 2025 with a lot of fanfare. You know, it didn't catch on right away, but it was obviously serious, and I remember there was a round , I think, a couple of months, three months after that, and I asked Michael. I said, "What do you think of Claude Code?" And he said something like , well, exactly what you're saying here: " Listen, we're entering the biggest market in the world." There will be competitors.

In fact, a whole range of players have been changing here since the beginning . Copilot, Microsoft Copilot was the first. Claude Code is just one of a number of players, and we believe that there will always be serious competitors in large markets. It doesn't scare us. And, you know, I remember being struck by the clarity of that comment, the lack of fear, you know? There is also humility there. It's a mixture of modesty and courage, and I think that's exactly the winning combination when you're challenging such giants.

I know it sounds vague, but things like that—it's almost Bezos-esque: if you're smart and you're right , it gives you the confidence and focus to have that approach when you're being pressured by such a mega-competitor. Yes. Although I will say that there were a few moments of "this is a mess," but they were always related to the models. Remember when Opus 4.5 came out? We were all on vacation, and it seemed like the whole world unanimously said, "This is a mess." This is true. You know, these things move a lot faster than any of us thought, and the same thing happened when the pre-release of Mythos came out and people got access to it .

These models were getting so good so quickly, and I think another thing that the Cursor team probably doesn't get credit for is how dramatically they cannibalized themselves Reed Hastings-style in about two years, right? So, first they were in the IDE, then they moved to an outdated platform, and then to a model platform. And it's very, very difficult for founding teams to have the discipline to do that. "Tab" was something big when, you know, there was that famous post in the Carpathians. Apparently, he has many famous posts where he simply writes: Tab, Tab, Tab.

Yes, exactly. It doesn't matter that much anymore . Yes. I actually think that's a good transition to other strategic decisions they've made along the way regarding the business model, and to your words about the "this is the mess" moments. I don't know if they would see it that way , because I don't think the vagaries of Twitter affect them as much. Ahem. As for their investors, as for their investors. Yeah, like the rest of us who just scroll through the feed all day, but I guess that was in the summer of '25 when there was a lot of writing about their gross profits and stuff.

And then, of course, after the release of Opus 4.5, maybe in the January period. We all need to talk about this, because these were two different but interesting periods. Yeah, I mean, what I can say is that X is driven by pure adrenaline, hype, and glee. And there is usually a huge gap between what people say in X and how things actually are in business . So, despite all this, business generally performed very well. It's so interesting, and a lot of it comes from investors, especially when they talk about the quality of the business.

But technological transformations always outpace understanding of the business model. Remember the internet? We couldn't even figure out how to monetize it for a long time. Forget about profitability, about any earnings at all , right? And here you have unlimited demand, unlimited growth and margin. So I think it's, firstly, an echo chamber of its own, as they say, and secondly, in every technological wave, investors try to say something clever, taking apart the business model of a new technology, and historically they've always been wrong. Yes. Yes, definitely.

I think I'll borrow a quote you said, Martin: first comes the technical transformation, then comes the business model. In this case, I believe that the strategy by which they brought all the users on board in record time allowed them to gain the assets, data, and expertise to build their own models, right? You would n't be able to do it any other way if you didn't start as a cutting-edge lab. This obviously pays off in that combination, which we'll talk about later, but I think this workaround was very inspiring, even though they got a lot of criticism for it.

So, Martin, how do you think Cursor became a sales-oriented company or a company that knows how to sell? Because, you know, in the beginning I also told Michael several times . Like, "You should invest in sales." He answered: "No." "I won't do that." They literally created the email enterprise@cursor.com. Like: " If you are a corporate client , send requests here." But they're obviously selling really well right now , right? And in general, in all market entry strategies. How do you think they made this leap? Yes, that's a great question.

So, I assume they study everything they do very deeply. They are very balanced. Every investment round, you know, every product decision. And I think that as they learned more about the market, they realized that the profits were in the corporate segment. A classic example: if you look at competitors, say, large laboratories, they actually subsidize or provide basic services for free. But then they get high profits from third-party orders and excellent profits from the corporate sector. And of course, it was true scalability. And as always happens in companies, they decided: okay, for future viability we need to go corporate .

This is where the main costs are concentrated , but also significant profits. Profitability is the main value here. We have created a great customer acquisition mechanism, so it is a matter of survival and strategic importance. And when they decided to do it, they took it seriously. This was probably the fastest sales growth in the history of the planet, right? I mean, Sara, you were very involved, for example, in the hiring process. Can you tell me about that expansion, because it was... Yeah, it's so interesting to watch, you know, as they say: if you do one thing, you do everything.

It's really noticeable. A lot of what you're talking about and how they approached the product and so on was reflected in how they hired employees, especially for go-to- market. We must pay tribute to Jordan, Roman, and their team, who built this mechanism from scratch in the initial stages. These are simply incredible people, some of the best I've ever worked with , and with whom we all worked. Well, they also spent 40% of their time on recruiting. Yes, yes, exactly. What is incredible for tech founders of product companies is spending 40% of their time on hiring.

I've never seen anything like this before. And it's so effective because the company is really a team that you build, and they took that as a foundation. Everyone is talking about it. Everyone talks about bringing in the best founders or whatever, but it's a whole other thing when Oscar says, "Hey, I'm on an overnight flight to Europe to close a deal." I literally, well, see him there. It seemed like it was constant . Absolutely. And you know, there was that great article that Brie wrote about their hiring methods that everyone should read .

Um, a lot of it has to do with their engineering hiring, like, internships, all that stuff , right? But even in hiring go-to- market specialists, how much thought did they put into hiring a single account manager, right? And, of course, this is not news, but it is this approach of " informal channels, channels, channels," that they have perfected . They knew exactly who they were looking for as the first account managers. I remember we would come there and sit in their office for 8 hours, helping them find people and all that.

And they didn't say, "Oh, we want to do everything the old way." We said, “This won't work. We are building a company in a new way, right? Therefore, we need very specific account managers who fit this profile. We're going to go and find them. We won't just wait for people to come to us. All the things that have already been written about a little bit regarding the engineering side. They did this on the go-to- market side as well, and obviously with phenomenal results. That is, they have reached over 50% of Fortune 500 companies, it seems, faster than anyone else we have seen.

Can you talk specifically about how they applied the lessons from engineering hiring to, say, hiring account managers? Because this is something that I think our founders face all the time. Well, you know, I personally have often encountered this. How, well how, how do you actually make this leap and what lessons actually carry over? Yes. Or what you saw in Karat. No, yes. And you know, hopefully this has been written about before, so hopefully I'm not sharing classified information, but how they think about eras in a person's or company's career, and who was the main lever for making it happen.

I love this period of intensive research where they say, “Hey, here are the top 10 companies that have been successful in this type of sales. And then let's dig deeper. Here are the top 10 teams in these companies, and then here are the same number one and number two, the specific account managers who enabled these changes and took into account, again, informal channels, channels, channels, right? As their manager, their manager's manager, and their manager's manager's manager would say: " This is superstar AE." And, of course, at the initial stage, the question was: will they be able to work in conditions of some uncertainty?

After all, you don't need someone who can only follow instructions. Now , maybe they can hire them, but in the beginning, you know, the original AE lineup was incredibly strong. And I think understanding what the best outcome looks like over a period of time and who delivered it is incredibly important for both engineering and product creation, as well as going to market. And I think finding the same star who did it at the right moment is really hard. This significantly narrows the pool of candidates you can choose from.

But because Jordan Roman and the initial team made the right choices, they were able to scale quickly without losing the Cursor culture. Yes. Which, as you know, has been the subject of many articles regarding Cursor culture. I'm curious, maybe you'll share your own impression of her. As I mentioned earlier, supposedly programming was all they did. That was my impression at the time, and the founders can, of course, correct us. In my opinion, they did n't do it then because they were obsessed with work. And I think they still don't do it because of their obsession with work.

They just literally love writing code and building systems. And I think the three of us can feel the same way sometimes , right? This is a completely understandable thing. And that seems to have been the main cultural stimulus at the time. The team is very small. Everyone had that attitude, and we were looking to the future: What will this actually become? From an engineering or product perspective, they were very narrowly focused on what they were trying to achieve and how to manage the team. But there is a huge paradox here, because I completely agree.

I agree 100%. However, if we analyze their actions in detail later, the conclusion is completely different. What did they spend their time on? For recruiting, market entry, understanding business models, business development. So , to some extent, and again, I completely agree. And again , on one hand, you have a very product-oriented team coming from the tech industry who literally wanted to build products that they themselves would want to use. So, listen, I think strong founding teams are working on exactly what the business needs, and I think they've done very well.

And I think that allowed them to navigate the most competitive market we've probably ever seen. Was the market more competitive than coding? I mean, it's hard to imagine. There are so many companies. Database wars? Yes, I mean, you have very stiff competition in relatively narrow markets. Databases are a huge field, but compared to coding, I actually think it's a lot smaller, right? Well, like, coding is a kind of metacategory, right? It's like the entire expression of human thought in economic form is a kind of coding.

Imagine : starting with competition with Microsoft and ending with competition with, say, Anthropic. You know, both of these companies are kind of megacorporations with a huge amount of capital, power, and distribution. Meanwhile, you fundamentally change your approach to the product, your go-to-market, and then participate in the most complex M&A deal of all time, all in just 2 years. I think this is crazy, right? That's it, in 2 years. Like this is the ultimate creation of startups, or ... No, I'm completely... There's definitely something to what you're saying, right?

In the beginning, these guys were incredibly focused on the product and literally improving their own lives. They literally coded for fun. When I finally got them to admit that they were doing something for fun, it was, "Oh, that's coding too." But yes, which, by the way, is the only one... Guilty. Guilty. Yes, that's the only pleasure we have left, by the way. You know, now that models have brought us a future without scarcity. It seems like coding is, you know , the latest fad. And, you know, but there was always a realization, or at least I always thought, talking to Michael, that he always understood the game, if that makes sense.

He really was, and they all genuinely loved coding and building great products. But they understood that building a long-lasting or great company is not just about obsessing over a product. I had the feeling that he always understood this and seemed to be calculating everything a few steps ahead. Yes, listen, if your goal is trading software, it is critically important to set up the business correctly, that's a fact. Yes. I wanted to add a story that's not related to, uh, their programming. But another thing that, you know, we've seen in a lot of founders , and Michael, Swale, and Amon embody, is this principle: if you do one thing, you do everything, right?

So , of course, this attention to detail in the product, a very artisanal approach, craftsmanship, right? The word "flavor" is overused, but they had plenty of it. But this extended not only to hiring or going to market, but also to the space they created and the events they hosted. I remember walking into their office and thinking, "Wow, in my 20s with a little one, I wouldn't have guessed to hang, I don't know, garlands here." They were known to walk without shoes, right? You came in, put on your slippers , and felt right at home .

That's why everyone worked there constantly, because they felt: "It's cool here, like at home." And they, uh, you know, everything was in the "hygge" style, Scandinavian comfort. I remember coming into their second office when they finally expanded and complimenting me on the sofa or something, because I was just furnishing my house, and they said, " Oh, yeah, we got it from a second-hand furniture dealer, very hygge-like." So they were aiming to achieve a specific look and atmosphere. And it struck me because I thought, "Wow, I don't remember guys in their 20s thinking so much about the space they're working in and how it affects the product and your well-being." Yes, there's actually a good connection here with a lot of their decisions.

So marketing was a pretty obvious example. They were very picky: "We use X, that's not our style, we don't spam." We gave so much advice. I remember them saying, "Oh, let's do this event." "No, that's not our style." So they very skillfully refused in order to preserve their essence. Hmm, I will say that they were also very, very good at it when it came to personnel. That is, if a person didn't fit, they made a difficult decision pretty quickly. Hmm, this is very , very rare to see in early-stage teams.

And they were very, very consistent about it. So I would say that this is almost a kind of unobvious, probably the biggest strength of the company: if they have made a decision, no matter how difficult it is, whether it is a decision not to enter the market or to limit the scope of their activities. They see things through to the end, or, you know, who stays in the company and who doesn't. So, you know, I give them credit for really having the courage and perseverance to make such difficult decisions.

Which is kind of a trait of Elon too. It's also very Po- Ilonian, yes. I actually think that these two companies—I mean, we just talked a little bit about mergers and acquisitions. Hmm, they fit together very well . They fit together very well on a technical level, do n't they? Elon has the computing power, they have the distribution and the data. Hmm, on a visionary level, they both believe that code is the way to change the entire field of computing and, perhaps, all of humanity. But also on a cultural level, right?

They are very, very, very focused on product engineering. Hmm, very focused on product engineering, iterating very quickly. You know, they make tough decisions very quickly. And that's perhaps why they haven't been involved in many mergers and acquisitions. And I would say that this is probably the best combination I have ever seen, all things considered . Yes, I completely agree. I mean, there's a funny analogy with our experience with SpaceX. I remember we met them in 2019. Back then, it was all about the launch business, but the SpaceX team was saying, "No, we're going to launch global connectivity.

We will provide internet to everyone in the world." They had nothing then, not a single Starlink satellite in the sky. And of course they did, right? And if you think about the Cursor team that told us, and, you know, they actually beat every prediction that they gave us, but oh, and by a lot. There are so many common traits in Elon's behavior : moving quickly, entering large markets, winning in them despite incredible competition, say, like with Comcast, or in the case of Starlink—there are countless parallels.

Of course, the spheres are completely different, but it's incredibly interesting to watch. There is one more thing I would like to add: in my opinion, both in the case of Elon and SpaceX, and with Chris, everyone around constantly predicted the imminent demise of both companies. Mhm. Yes. So I think there's a certain cultural resilience here in the face of skeptics, which I actually respect a lot in both cases. Actually, I have another important question that I wanted to ask you. Cursor is also very adept at M&A.

I mean, small acquisitions. You know what's funny? I was just about to say that. Yes, yes, yes. I'd like to know a little more from your perspective: why they did it, what their philosophy was, and all that. So, listen. There is such a battle for talent that companies that have learned to attract them through acquisitions gain a significant advantage. There are many significant advantages here. Therefore, their first steps were exactly this: they saw a great team and simply made the acquisition. What was really cool about Cursor was that they were never intimidated by operational complexity.

They just said: no matter how complicated it is, we will do it. They also were n't too concerned about their ability to integrate culture , because they were very good at it. It was simply part of their approach. Only later, for example with the acquisition of Graphite, did they begin to consider acquisitions as part of a strategic course. Of course, this was a much larger takeover. And, you know, I think the company has grown so quickly that who knows how they would integrate all of this and implement it in the future, because now they themselves have become part of a large acquisition.

But I will say one thing: they approached this the same way they approached everything else— extremely thoughtfully. They had a plan, they executed it quite well, and so this may be the most concentrated company in history in terms of the number of mergers and acquisitions they have done. Yes, yes. You know, M&A and everything else. Yes, I wanted to say that now it seems like a kind of badge of honor has appeared when the CEO says: "My company has X founders working." Yes. Um, and I think that's supposed to be a sign of respect, right?

I don't know what this indicator is in Cursor. Do you guys know? Um, I remember, they have so many great founders there, but I remember how they brought in Tito from Qualia and then Adam Ward to be on the talent acquisition team. Yes. In fact, seeing firsthand the benefit of bringing in someone who is actually the founder and CEO, but who has fit perfectly into Cursor's culture, lifted it to new heights, and accelerated its development. I think this strategy...I don't know, I'm sure they didn't invent it in this new era of generative AI, but they are definitely one of the best examples of how to integrate founders into your company.

Funny, right? Because Anthropic often gets that label now. Like, "Oh, the CEO of Company A just moved to an individual performer role at Anthropic." But Cursor did it first, and they didn't just put people in the roles of performers. They really build the company in a way that, to me, is very...it's very unconventional, but... to run something incredibly complex. Yes, that's it. This is a traditional list. Yes, that's right, yes. I think this is actually true. I mean, it's not that they did something that people think is hard, but it's actually easy.

I think it's really difficult, and they just managed to make it happen. So let's give them credit.

Go deeper

• All A16Z episode guides