Podcast Episode Summary
The podcast episode features a conversation between Jack Altman and Martin Casado, a General Partner at Andreessen Horowitz (A16Z). The main topic of discussion is the shifting dynamics of venture capital, the importance of media in the industry, and the rise of AI infrastructure.
Key Points:
Participants:
Podcast Episode Notes:
The episode is a conversation between Jack Altman and Martin Casado, with a focus on the shifting dynamics of venture capital and the rise of AI infrastructure. The conversation is engaging and informative, with Casado sharing his insights and experiences as a General Partner at A16Z. The episode is a great resource for anyone interested in the venture capital industry and the impact of AI on the market.
Jack Altman sits down with Martin Casado, General Partner at a16z, to unpack the shifting dynamics of venture capital and why media matters more than ever. They cover a16z’s evolution from generalists to specialized platforms, the rise of AI infrastructure, and why today’s fiercest battles are often for talent, not market share.
Timecodes:
0:00 Introduction
0:27 Importance of Media for VC
3:50 Evolution of a16z
7:00 Specialization
10:32 Value of Distribution
13:16 Staying Power in Infrastructure
19:49 The Conflicts Dynamic
26:32 State of Play in AI
30:48 The Future of Coding
34:58 Significance of Open Source
39:48 Marc Andreessen’s Leadership
44:02 The Only Sin in VC
48:37 Scaling a Lot of Board Seats
Resources:
Listen to more from Uncapped: https://linktr.ee/uncappedpod
Find Jack on X: https://x.com/jaltma
Find Uncapped on X: https://x.com/uncapped_pod
Find Martin on X: https://x.com/martin_casado
Stay Updated:
Let us know what you think: https://ratethispodcast.com/a16z
Find a16z on Twitter: https://twitter.com/a16z
Find a16z on LinkedIn: https://www.linkedin.com/company/a16z
Subscribe on your favorite podcast app: https://a16z.simplecast.com/
Follow our host: https://x.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Stay Updated:
Find a16z on X
Find a16z on LinkedIn
Listen to the a16z Podcast on Spotify
Listen to the a16z Podcast on Apple Podcasts
Follow our host: https://twitter.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Host
Today on the podcast, we're sharing a feed drop from Uncapped, where Jack Altman sits down with Martin Casado, A16Z general partner. They talk about the shifting dynamics of venture capital, why media now matters in a way it never did before, how A16Z evolved from generalists to specialized platforms, the rise of AI infrastructure, and why talent competition might be more fierce than market competition. Let's get into it.
Jack Altman
The market is so big and it's growing so fast. Even companies that seem like they're competing end up in totally different places just because so much white space is being created. But they're all competing—like totally different companies and spaces are competing for the same talent. So the first time I can remember where the actual talent competition is like way more fierce.
Martin Casado
Martina, I'm really excited to be doing this here with you today. Thanks for making time for it.
Jack Altman
One of the things I was just chatting with you and laughing about on my way in—there were like many other podcasts going. There's one like before and after us. I talked about this with Mark about how podcasts are like this future thing of media. My question for you is, as somebody who's been on the inside of a firm that's dominated this, and you do a lot of it yourself, what's your experience about the importance of media for venture capital?
Martin Casado
I think it's probably important to recognize that it's never been a thing really. If you look at a lot of historically good investors, they weren't very public. I think about the greats, like Moritz, Ping Li, Doug Leone, Benton, Mike Volpi. They're just not very public. Historically, there's been no correlation between being public or not.
I think a couple of things have changed. One of them is that traditional media just turned on tech and it hates tech. In the past, when I was a founder, to get a lukewarm to positive article was pretty straightforward, and the VCs would help with that. They would know the reporters. It was very easy. But now it's actually very dangerous because you go talk to them and who knows what they're going to say. So in a way, if you want to help a portfolio, you do want to build a bit of a platform. You do have to go direct.
The second thing is, if you're traditionally in enterprise, marketing has been something that you build brick by brick. You put content out there and people read it. It's durable over time. You get this compendium and you build a brand over time. It feels we're in an era now where it's become so episodic that if you don't understand the current zeitgeist, you can't even get a voice at all. By episodic, I mean today, GPT-5 launched. It was massive. If you didn't know that was going to happen, you would have been drowned out. If you did know, you could draft on it. For some launches, they're a big deal, and then they just disappear forever. So much of the nature of how we consume and think about content has changed.
I do think that venture capitalists, one, if they have a message they want to get out, they kind of have to go direct. If it's your own platform, it doesn't hate you. That's one. But also, to help your portfolio company, I think you need to build an in-house capability so they can know how to most effectively message. You can't really borrow a page from traditional marketing. This is from someone that's come very much from the age of traditional marketing. It's just different.
Jack Altman
One of the things I've been very surprised by is there's always room for another podcast. People consume a lot of this stuff. I think people in tech find it's almost like a halfway between working and watching Netflix. It's like, I'm passively learning, but it's low stress. People would rather consume a good podcast than a new Netflix show.
Martin Casado
For sure. There's always a concern that there's too much content. But that concern has been around forever. There's always been too many books to read, too much TV to watch, too many web pages to read. It's always been an ordered list that starts with the most important and goes to infinity. The question has always been, how do you be in the top 10 or the top 20? That changes all the time. I do think you're right. People like to consume things casually that are relevant to their interests. It's actually a great time now that you can be in that top 10 for the set of people that you care about.
Jack Altman
I want to talk about your time at Andreessen and what's evolved, which has obviously been a lot. Can you give the picture of what it was like when you joined?
Martin Casado
When do you want to start? When I was a founder or when I actually joined as a GP?
Jack Altman
Maybe when you joined as a GP, but then let's connect it back.
Martin Casado
Okay. When I joined, it was 2016. I've been here almost 10 years.
Jack Altman
That's a lot.
Martin Casado
In the wild. I think it was the ninth general partner. The firm had 75 people. Not only were we all generalists—you could do whatever you want, that was kind of part of the pitch—but most of us had done some pretty serious time operating. My journey with my startup was about 10 years. Many of us were so tired of the space we came from that we did something totally different. It was very, very different. Generalist, few people in the firm. The investing team alone, you really had GPs and we were all the same. Then you had relatively junior partners that couldn't write checks that would bounce between the GPs. There was no alignment at all. It was a very different structure.
Jack Altman
One of the things that's interesting over the evolution is that it started in this generalist version, and now you're running a distinct platform. The way the firm is shaped is there's a ton of autonomy. Mark was talking about this on the podcast—part of the idea was we can recruit these amazing GPs because they get huge autonomy, but we're going to have specialists, sort of leaders for things. What has that change meant for you? What's it been like to go from generalist to specialist?
Martin Casado
It's important to talk about why a change is necessary. The historical context is interesting. There are two things that are important. One is the model of venture came out when tech was like a non-market. It was total speculative stuff. When you meant tech, you meant everything from bio to software. Everybody was a generalist. It wasn't really a profession. It was like, if you wanted to play money, you do it. They made decisions that made sense at the time, but that no longer makes sense.
For example, it's a historical quirk why venture capital uses the same model you'd use for a dentist office—a partnership where everybody's equal. That makes sense for a small service organization, but you can never scale that. There are all these decisions that were made when the market was much smaller. As AUM grows and as the market grows, there are many more companies you could deploy in now. You'd have to restructure the firm. Our view is we definitely want to scale. We definitely thought we had the best platform for founders. The markets were so large, you didn't have to be a generalist. In 1980, if you did enterprise infrastructure software, how many companies could you invest in? Not a lot. Two or something. Now, someone can have an entire career investing in databases alone. As the market grows, clearly you have to specialize. When I joined, we were all generalists, often hating our own disciplines because we'd been through it.
Jack Altman
Do you have to become specialists as the market grows or as the firm grows? Is the specialization choice downstream from growing the firm, or is it downstream from the market growing?
Martin Casado
I think it's ultimately a function of the market. If you believe this stuff is competitive, which I do, then you need to end up with a product that is competitive. Because it's adaptively competitive—if you've got two firms competing, you're always looking at the weakness of the other one. If a certain firm can't do seed, you'll want to do seed. If they can't do large checks, you want to do large checks. What happens is everybody ends up getting as many products as they can so they don't have any weaknesses. That will naturally happen. You can only do that if the market is large enough.
Now you have a high AUM. You've got a lot of products. I've got a growth fund, a seed fund, a venture fund. Then you have to ask how do you scale that? My venture was not built to scale. I think this is why we've seen the industry go this way. The market has increased a lot. Funds want to be competitive. To be competitive, they have to find out what products they offer that are actually competitive. This drives to higher AUM. As a result, you have specialization.
Jack Altman
When you're... assuming you want to scale AUM independent of the market, you have to solve this problem because you can't scale a consensus org of generalists. It's just not something you can do. The people issues on the inside. You can't get through good decisions.
Martin Casado
Is that what you mean?
Jack Altman
Well, there are many issues. One of them is you wouldn't ever have a structured approach to tackling a market. You can never know that you've got good coverage because maybe everybody wakes up in the morning and they all like the same thing. I don't think you could actually, even from a numbers game, scale it because you're not carving it up enough where you know you've got uniform focus. You don't know if you are hiring people that can cover certain areas. From a strict number standpoint, it doesn't work.
How valuable is the specialist thing when you're in these competitive situations? I imagine a lot of the time when you're competing to win a deal, you're up against a firm or a partner that is more or less generalist. I'm curious how that plays out in the day to day.
Martin Casado
It's hard to answer how much it helps in the competitive situation. My experience is it's a lot more powerful that founders know I've been a founder. I know this is such a cliche thing to say, but I do feel that resonates much more than, "I got a PhD in computer science" or "I know infrastructure." The reality is most founders know a lot more than I do about whatever their area is, even if I've got a high level thing. In the competitive situation, it's not hugely helpful.
What I think is very helpful is that I am primarily a Series A investor. At Series A, you have to have some thesis on how the tech hits product and how the product hits the market. Unless you've been very close to both of these things, that's a hard thing to do. If I was a growth investor, it wouldn't matter. I just look at numbers. But for where I am, I think you kind of have to understand.
Jack Altman
An interesting offshoot question from that is, when we were talking earlier about how important media is, you said it seems really important, but there are a lot of great examples of investors who are all over media and social media. There are also phenomenal investors you never hear about on the internet.
Martin Casado
I don't know if they're correlated at all. Mostly the best investors I have known in the last 20 years had no media presence and no interest in it.
Jack Altman
Around being a former founder, as I'm thinking through names, I can think of a lot of examples of both. I definitely think founders appreciate it. Speaking as a former founder, I think that's a nice thing. But I wonder if that's also an uncorrelated thing, or do you think it has more of a correlation?
Martin Casado
I'm just going to guess. I would guess that founders really appreciate reach. I don't think a founder is like, "Martin, I saw you on that podcast. You seem smart." Because everybody sounds pretty smart on podcasts. I do think a founder would be like, "When you really believe in something, you talked a lot about it. I will have the opportunity to talk about it. You will help me break through the bootstrap problem of zeitgeist understanding and brand." I do think having a platform matters more and more. Again, a lot of this is just because the media has turned on tech so heavily.
Jack Altman
Totally. There just aren't a lot of options.
Martin Casado
I think sometimes we in VC overweight our importance in these things. Most companies with great brands did not do it through a VC firm. It's not like we can single-handedly make great brands, but we are an accelerant. We are a platform. There is actually a lot of signaling as a result of alignment with a good firm. I think all of that matters.
Jack Altman
There's this question of whether the top VCs get to do the great deals because they were the top VC, or are they in some sense helping make them? My own instinct is that for the most part, it's the former. Companies are almost exclusively made by the founders.
Martin Casado
I agree a hundred percent. The primary reason to create a distribution channel as a VC is so the portfolio can get out there and reach people. This is very hard what we do. It's not because Martin needs to be famous or Martin needs a brand. That never comes up in a closing situation. I've done so many deals. That has never been a thing. But I do feel that a number of our companies, once they're ready to launch, we can provide a benefit. That is ultimately the benefit to the portfolio. But I have never seen a company I lose by marketing.
Jack Altman
Totally. I want to jump over and talk about AI a little bit. I'm interested in talking about infrastructure because it's something I know relatively little about. First of all, what is it? If you could put a broad fence around what the term is.
Martin Casado
I do computer science infrastructure. I'm a computer science maximalist. I think it's the meta-discipline that you can solve other disciplines with. We solve graded unified field theory and physics goes away, and then we just go on to biology type thing. Infrastructure is the stuff used to build the apps. You sell to technical buyers, people that use computer science to solve business problems. If the company sells to marketers, that's not infrastructure. But developers, database administrators, networking—that's core infrastructure. Depending on how you count, this is a multi-trillion dollar industry. The important thing is the actual buying and use behavior is a very technical thing. That's our definition of infrastructure.
Jack Altman
Compute, network, storage, databases, now models, dev tools, that type of stuff. It seems to me that when these new paradigms come—cloud, mobile, AI—that's a very good moment for infrastructure because the board's shuffling a lot and new infrastructure is being laid. When you're looking at it, is there any broad way you think about whether this will continue to exist over time? Will the models or AWS do it? Will there be a need for a third party? How do you start to think about what will play out over time at a structural level in infrastructure?
Martin Casado
Can I say something that may not be true, but I feel very strongly?
Jack Altman
That's what we're here for.
Martin Casado
This is an inflammatory opinion that's self-serving that may not be true. But it's an observation. In software, the true differentiation is technical. There's brand stuff and business stuff, but if you have two products, it comes down to a technical problem. That almost always comes from the actual infrastructure the software is built on. If I built two dog-walking apps, the fact that one has three or four features is a very light differentiation. But one being super fast and one being super slow—that's infrastructure. The companies that provide infrastructure, I think they are the source of value, the source of differentiation. While there are fewer infrastructure companies, my bet is that they have better multiples and are more durable because they service everything above it, but they're the thing that provides it.
Sarah Wang, who is an investor on the growth fund here, and I did a relatively loosey-goosey public market analysis of what the multiples are of companies that are infrastructure versus apps. They just have higher multiples for this reason. Does that make sense?
Jack Altman
It does make sense.
Martin Casado
My view is infrastructure is where the value is. Every time you have a platform shift, you'll get a new set of infrastructure companies, and then a bunch of apps get built on top of those. The value is going to accrue largely to the infrastructure because that's where the differentiation ends up happening. You don't need a platform shift necessarily for important infrastructures with good multiples. I think it's a durable part of any sort of application. But the question is, what happens when it matures? The clouds do and it becomes an oligopoly, and private investors can no longer invest in it. But every time we've seen that happen, you see a layer of infrastructure evolve on top of it.
Let me say it this way. The way I view the world is you've got a bunch of app developers who are non-tactical. They want to develop apps to solve all sorts of consumer and business problems. Their goal is to build an app for a non-technical user. Why would they invest heavily in technology? They'll pick up whatever is easiest to use technically. The companies that fill that need provide a lot of the value of the true differentiation. Does this make sense?
I always think there will always be something that you can make faster, easier, more reliable. It'll always be the bedrock that apps get built on. Until people stop wanting to produce apps, you'll always need to produce. This is totally independent of macro shifts or platform shifts.
Jack Altman
How do you think about if or when the big players are going to decide to enter those markets and how that might impact things? In cloud, whether AWS was going to offer something directly. Now whether OpenAI or Anthropic offers something directly. Or are you like, if I'm investing at the Series A or B, that's not that important? You just have to think about great entrepreneur, big market, and that's okay.
Martin Casado
I worked at VMware for four years. We were the big incumbent. We were always worried about the shadow cast by these incumbents. The reality is it's not nearly as strong as everybody's worried about. It's very hard for these big companies to execute. re:Invent is the AWS conference. Being an infrastructure investor for so long, every time they have re:Invent, I have to play therapist to all the founders. They call and say, "They're entering our market. They're entering our space. They're doing all this competitive stuff." I still today can't really think of a company that AWS has put out of business, even though they entered the market. They kind of compete with everybody.
If the market will bear an independent company, then that requires your own sales force, your own focus on customers, your own support, your own technical differentiation. No big company can build a small company in a big company because they have too many centralized services. If the market won't bear an independent company, there's no company to build anyways. My view is, as long as the market is continuing to expand—which software is—if you enter an area that's viable, as it expands, you will fill that expanse. If that doesn't work, the market is either not big enough or it isn't expanding fast enough. If you take the historical view, this is the case.
Jack Altman
I strongly agree with that. When you think about markets in AI right now and how things are evolving, one of the things I thought was really interesting from talking to Mark was that as you all ambitiously grow the firm, one of the biggest issues is companies running into each other in this conflicts dynamic. You're becoming prominent within an area to a degree where companies, as they grow, grow into each other. What's your experience with that?
Martin Casado
It's such a complicated problem. You can try and do everything right and still end up with conflicts. It's pretty good to categorize the conflicts. Perhaps the most common conflict is one company that you've invested in pivots into another one. This one is basically impossible to do anything about because companies have to figure out the right business. You're on the board or not, and you don't control it. There's nothing an investor can do.
There's another more pernicious type of conflict with existing portfolio companies. We're seeing a lot of this now. Imagine you have a tech revolution like AI. You have a set of old companies doing things the old way and a set of new companies doing the new thing. The old company wants to pivot to using AI to do what they did before. But the reality is the old way is not the AI way. They're not AI native. Now there's this question: when we invested in this company, it was doing X. Now it wants to do X with AI. But the AI way of doing it is entirely different and they've got no chance. You have the dilemma of going to the founder and saying, "We're investing in one of the new space, but it's AI and you're not AI." That's not going to work. Or you just don't do the deal.
I run into this one a lot. We've got a very large portfolio. To date, we've just been like, "We're going to back the portfolio companies that we have." This happened last week. The founder called me and said, "You can't invest in this. This is the space we're going into." They haven't even done it yet. We try to do the right thing there. That one is the toughest one for all investors today because you never want to bet against your own portfolio. But the reality of them being actually competitive is very low.
Then there's one more, which is kind of a fun stage thing. We've got a growth fund. They do their own thing. We've got an early stage. They do their own thing. Sometimes the communication isn't always perfect, and you can end up in conflicts that way.
The one that we simply do not do—I always have this talk track. I borrowed it from Chris Dixon, but it's very effective. For any company that I'm an investor in, if I'm talking to another company that looks similar, I'll ask the founder, "Is this your mortal enemy? You only get one. You can't keep changing it. But if this is your mortal enemy, we'll do everything together to kill it. And we won't invest in that." At least that gives them the power to decide who it is, but not hamstring investment efforts.
Jack Altman
That makes sense. It's also funny because a lot of times, two companies that look the same but are serving very different segments of the market might sound like competitors, but they're never going to bump into each other. Versus two companies offering different products to the same customer are much more likely to bump into each other.
Martin Casado
In AI, it's even crazier than that. The market is so big and it's growing so fast. Even companies that seem like they're competing end up in totally different places just because so much white space is being created. But they're all competing—like totally different companies and spaces are competing for the same talent. This is the first time I can remember where the actual talent competition is way more fierce than the market competition.
Jack Altman
What's funny about that is I've heard of people getting upset with their investors because they know this company has nothing to do with it, but they were interested in that candidate. And one of your partners sold that candidate on one of their programs. It's a very real thing. You often don't even know. They'll say, "Oh, I'm talking to a healthcare company," and you're like, "Okay." What's tough, but I guess is also a blessing overall, is that there are a lot more good ideas than there are talented people to work on those ideas. One of the hardest things right now is clustering talent densely enough behind a good idea.
Martin Casado
This happens when there are these large infrastructure buildouts. This happened with the cloud too. There are moments in time where to build the system, you have to have experience with the system at scale. This happened with the internet. This happened with the big cloud data centers. This is the case with AI. It's one thing to go to school and know AI and be a good researcher. It's another thing to have actually trained a very large model. Maybe there are 30 teams that have ever done it. That's part of where you see these mega acqui-hire acquisition things. Certain experiences are just worth a huge amount.
Jack Altman
A hundred percent. The market always normalizes these things. This is all ancient history now, but the exact same thing happened in the internet. I remember once there was basically one guy that ever wrote a BGP stack, which is a way routers talk on the internet. He was the one guy that could make it work. He just got these crazy offers at the time, and he'd jump between all the router companies. There were very few teams that could do this. We've always seen this episodically in the industry. We're just seeing the new version. The businesses are working and they're doing great, so we're seeing it on steroids.
Martin Casado
If you're a FANG-sized company, what's it worth to have the one or two or three people who really know how to do something huge?
Jack Altman
I also feel like tech always figures out a way around regulations and markets. In the late 90s, you could IPO a company for very little with not a lot of market traction. Remember the whole SPAC craze?
And then now we've got these weird acqui-hire things. I think the reality is in hot markets, people know that there's a lot of value to be had. Nobody knows exactly where it is. So there are all sorts of things the markets try to do to get access either to the talent or the companies or whatever it is. We're kind of seeing our version of that now, whether it's like "I'll hire an individual" or "we're going to acqui-hire." But again, I feel like this is all normal in the sense that we've seen it in different shades in the past.
Martin Casado
Yeah, totally. It's like an evolutionary response.
Jack Altman
What are the markets right now in AI that you feel most confident are totally working? What are the ones where you feel like they're on the horizon and should be working very soon? And then what are the ones, if any, that you maybe have low confidence will work, period?
Martin Casado
So the diffusion markets are all working. Any area where you bring the marginal cost of creating something—a piece of content—to zero is clearly working. You're creating an image, creating music, creating speech. We don't think about these markets as much because we're all so focused on the frontier labs. But it's cheaper to build these models because they're smaller. And people need content. The economics are so simple. Imagine you're an artist and you're going to draw a picture of me. How long would that take you? A while. Three hours, and it costs you $400. But if I have a model do it, it's a hundredth of a penny. So you've got four orders of magnitude difference in economics. That's why we've seen those types of companies—I think like 11 Labs or whatever—do very well. So that's clearly working. This is content creation where the marginal cost of creation goes to zero.
I actually think the whole loneliness and companionship stuff is definitely working. It's just a very fragmented market. The unit economics are fine. I'm not sure from an investor standpoint how you think about it, but it's a use case that will be solvent and will do fine. Code seems to be working incredibly well.
Jack Altman
Yes.
Martin Casado
You see this in Cursor and the whole thing. The areas that I don't know—I mean, they're working, but I don't know how the economics actually pencil out—are the enterprise use cases at this point that are kind of a bit more agentic-y, autobate-y.
Jack Altman
These are the ones you're putting in the middle bucket? This is what you're saying is kind of working, but not 100% sure yet?
Martin Casado
No, no. So the middle bucket was actually like the friend, the emotional, the character that AI is. There's a long tail of companies that are basically emotional support or friends or entertainment. It's probably also a big component of the usage of the main models. That's clearly working in the sense that people are willing to pay for it, the engagement's great. From an investor's standpoint, it tends to be kind of long-tailed and fragmented. And then that enterprise agentic workflow type stuff—that was the fourth one I mentioned. That's like chatbots. Clearly it's working and there are companies doing it, but if you look at the companies, there's a lot of bespoke work going on. It's a different type of economic model than the content creation one where it's just a model. So those ones we're still trying to understand.
Jack Altman
Yeah. One way to think about that is how confident are you that highly skilled work will get replaced in, let's say, legal, finance, accounting, tax—those kinds of areas?
Martin Casado
I think the way I view it is actually very simple. If the use case is the model creating content and that content is whatever—it could be language, it could be image—that clearly works. But if the model is automating something a human being would do, and we conflate these two things all the time, that's totally different. If I'm like, "Model, do this thing instead of me," that's not content creation. It somehow has to mimic exactly what I do. That area still needs a lot of work. They clearly can do some work of a human, but not as exact, and they need a lot of guidance. I think that's an area that still holds a lot of promise, but the economic case isn't as obvious. "Make me a picture" versus "Go browse the web for me." It's that second one—automating what humans do—where we're still like, we've got lots of investments, we're very excited about it, we think there's a great future there, but the economic case isn't nearly as good as "Make Me a Picture."
Jack Altman
Totally. Let's double-click on code for a second. Obviously, you know a lot about it through Cursor, you were a technical CTO. Where do you think we are right now? I just posted one with Guillermo, who obviously knows a lot too.
Martin Casado
Yeah, he's great, he's amazing.
Jack Altman
This is both the future and it's also at the moment—it's not obvious that in today's incarnation it's necessarily producing quite as much value for engineers as even they themselves experience. But clearly it's going to get there. So I'm curious.
Martin Casado
AI in general has this problem, which is so dazzling. People conflate "this is dazzling" with "this is useful." That's for everything, not just code. You're so impressed—these things are magic. And then somehow that dopamine hit... What's really funny is, I posted this on X—a study that was saying people experienced their own programming as plus 20%, and the observed results are minus 20% or whatever. There were a bunch of reply threads where somebody was like, "No, no, no, this is crazy. I've been using it and I'm so productive." And then the reply to that is, "That's what the study is saying." I'm sure that's true for some people, but there is this thing. There is an endorphin hit. These things are absolutely magic. But I think it makes it very hard to think clearly about the actual utility.
Now, I think you can say a few things. There's a lot of things it does very well that programmers don't like to do that are pretty routine. Documentation. It's great at writing documentation. There's a lot of boilerplate stuff it knows. The thing I use it for the most is—writing code isn't just writing code. Writing code is understanding the frameworks, knowing how to deploy it, knowing how to run the tool chain. There's no first principle way of knowing that type of stuff. There's not some core computer science fundamentals on deploying to Netlify. That doesn't exist. It has all of that knowledge, which is clearly very useful. The writing code itself—I think it's still clearly very early days. If you constrain how you use it, it can be very effective. If you don't, it may not be as effective. Studies like this that are purely observational are hard to read into because you can use it for anything. And because they're so magical, I do think people tend to use them for stuff that they may not be so good at, just because the experience feels good.
My guess—like any new technology—will develop best practices. I feel very strongly we're going to get a 10x in productivity, but it'll take a while to get there. I remember another epoch of development productivity: when the IDE came out, when high-level languages came out, when OOP came out. We'd get so enamored with the tool set. Object-oriented programming—I'm going to do everything in it. And it turns out these were great advancements in computer science. They helped the best practices, they helped architecture and engineering. But at the time they came out, they were just so cool that they took a lot of our focus. I think we're seeing that now. This drop in productivity is not necessarily because code requires you to drop productivity. I think it's just so cool that I'm going to try it with everything.
Jack Altman
It does seem like it's on a path to—in the same way that people have been saying this about self-driving cars, and it seems like it's going to become true—it does seem like we'll get to a place where you can...
Martin Casado
It's so clear. This is incredibly obvious. You can literally just scope it to a subset of things that are obvious. It's really good at writing text. It's really good at writing documentation. It's really good at dealing with a bunch of long-tail framework stuff that you don't know. It's good at teaching you things. Clearly, there are things that it's obviously good at. I just think we get so enamored with it that maybe we start using it for stuff that it's not so good at or not so equipped to deal with. But this is very clearly going to change software.
And sorry, I don't mean to ramble on. I just have to say, I have been in software for a long time, since the late 90s. We disrupted everything. We disrupted the back office. We disrupted hotels. We disrupted everything. This is the first time I say that we're probably getting legitimately disrupted as a discipline. What it means to be a software engineer is changing pretty fundamentally. I think it's because of AI. So it's kind of fun to actually be the disrupted for a change.
Jack Altman
That's awesome. Another topic I wanted to get your thoughts on was why is open source so important to you? I saw you were really excited about Open AI as an open source model. I know this is thematically and spiritually important. Why do you think it is such a critical part of the way that this plays out?
Martin Casado
I think open source is historically one of the best mechanisms that shows a healthy ecosystem. What normally happens is somebody does something closed source, it turns out to create a market, and then somebody releases open source and it stops a monopoly from forming and enables everybody else. It keeps the people that are closed source continuing to be innovators and allows everybody else to come in. So it's been very, very healthy.
The thing that really worried me last year was that in the past, academia, VCs, and startups were all very pro-open source because they understood it was a very important part of a healthy competitive ecosystem. But last year, the people that should be championing open source—like VCs, startup founders, academia—were decrying how dangerous it was in relationship to AI. The implications of this to me are huge. The national security implications are pretty straightforward: if somebody else does the open source, it proliferates, and that's not good for U.S. interests. But for the industry, it's terrible. This is how you actually create monopolies. If you're not allowed to create something that enables everybody else... It was very dramatic when the Node-Cose was like "open source is bad," and Founders Fund was like "open source is bad," and academics were like "open source is bad." So I was much more interested in trying to reset the discourse than any specific open source release.
Jack Altman
I guess a lot of that probably came down to how potentially dangerous the technology was. If you thought it was extremely dangerous, then there is an argument for massive containment. What was the steel man in your mind at the moment when closed source was such a strongly argued position? If you had to argue why people were saying that, what would it have been?
Martin Casado
I really think it's the legacy of Bostrom. Bostrom wrote the book *Superintelligence* in 2014. "Terminator is waking up, we got to contain it." But this was before all of these things. Bostrom's book was a thought experiment—a platonic ideal of AI. And then somehow that created a very interesting intellectual journey on the perils of AI. But then GPT-2 lands, and these two things got totally conflated. There's also a lot of incentive for people to be doomery, I think. It gets a lot of clicks.
Jack Altman
But not historically. It's the weird thing historically. Let's take the internet as an example. I was there during the early days of the web. We had a lot of examples of how it actually changed the dialogue when it came to risks. We were running critical infrastructure on it. We had totally new types of attacks like the Morris worm, which had actually taken down computer systems. So we were at this space where, okay, it makes you more vulnerable if you use it, and you have new attacks that you can attack with it. And yet academics were like, "This is great." The technologists were like, "This is great." So you had this very even-handed debate. What was so weird about AI was it wasn't even-handed. I'm all for both sides of the debate. I'm not pro-innovation at all costs. But that's not what was happening.
Martin Casado
Maybe you're right. Maybe the doomers got more clicks. But I think it's something more than that. I think there was an existing intellectual legacy that came from Bostrom that had some very influential people. Elon was killed by that. Eric Schmidt, Moskowitz. They had very legit concerns, but they were kind of already primed. There was already that ready. So when it happened, I just think they were already ahead of the game. It took a while for the rest to catch up. Now when I listen to the discussion, it just feels more even-handed, which thank God I don't have to be on Twitter talking shit about this like I did for so long. I feel like the right voices are in the room.
Jack Altman
For sure. And obviously, neither of us is saying that there aren't real risks.
Martin Casado
No, there are totally real risks. It was just a totally lopsided debate. It's so crazy when VCs are talking against open source. I mean, no academics were talking in defense. But now we have a bunch of academics in defense. I think the right people got mobilized, but it really took some rallying to get the right folks back into the conversation.
Jack Altman
Yeah, absolutely. I want to ask you just a couple more questions about the structure of the firm and your own work outside of the specifics of what you're investing in. One that I thought was a really interesting point from Mark when I talked to him on the podcast was basically around how you drive the right overall aggression of the firm. He said something to the effect of, when you're in a market moment like this, the right answer is to just encourage people to do more. So many partnerships are trying to get other people to a "no," and that is the common function. He was basically describing something where it was like, how do we get people to a "yes"? How do we take advantage of this? I'm just curious how you feel and your experience in the last couple of years in AI land.
Martin Casado
One of the reasons Mark is such a great leader is his intuition on the temperament of people is almost perfect, and he will drive the right behavior relative to that. If he thinks people are being too conservative, of course he'll drive them to be more aggressive. There's definitely a more aggressive stance. The reality about AI is there's been a lot of money that's already been lost in absolutely record time. Just because the upsides are great doesn't mean the downsides aren't. There's been a lot of money lost. As a firm, we tend to be fairly disciplined and do a lot of market analysis. He thinks we're coming into a moment, and he's very good about pushing the team. I think he's absolutely right to do that because there is already a foundation of discipline that is not going to be eroded or compromised by doing that.
On the other hand, there are individuals who are like, "Don't shoot from the hip, do everything." And then he actually tempers his messaging a lot with those people. I don't know what was in his head when he was saying that, but my observation of Mark is he's very good at pushing when people need to push, but he understands the situations where that's probably not the most appropriate thing. This is a core issue of leadership: you need to provide the right kind of macro shift to get the right outcome without being too overbearing.
Jack Altman
Yeah. I think some of that conversation was in relation to fund sizes and what's possible. And I think some of it was articulating just that these companies can be enormous. This opportunity is quite oversized relative to what's going on.
Martin Casado
But again, it's a per-person thing. If you take him at his word, it would be an infinite fund deployed infinitely. So this is an intellectual landmark, which is set and is 100% the right one to do. But what's important—he's so good at that—is it's in relationship to the mindset of the people he's talking to. He knows that. And what flag post you have to put out to get people to the right place. I have seen him very subtly, depending on who the audience is, move that flag post to different places.
Jack Altman
Where do you normally experience yourself there? Do you feel you're often needing to get pulled into more aggression or into more conservative?
Martin Casado
I think I'm a 7 out of 10 for aggressiveness. I would say my team is a 5 or a 6 out of 10. There are people on my team that are 10 out of 10s. There are people on my team that are 3 out of 10s. If you did a normal distribution, I'd say we're six and a half to seven. So for us, he pushes pretty hard, but I know he knows that he's getting a step forward as opposed to... Being a leader of people is always in relation to pulling minds. It's a rare thing to be able to do multiple versions of that at the same time with lots of different people.
Jack Altman
Exactly. It's hard to appreciate from the outside because you have to actually see the conversations. It's something that he's just phenomenal at. The very nuanced different takes where he knows where people are and kind of nudges them in the right direction.
To the extent that we are in somewhat of a gold rushy—in the good sense—moment overall, does it change your perspective at all about what you need to see to want to make an investment? A specific version of this question: in a somewhat stabilized time, most people would agree that you should only back extremely special founders. Is there ever a version in these kinds of moments where a good founder in a great market with exceptional traction—some configuration like that—where you say, actually, that works here and that can produce something really big?
Martin Casado
Do you know how we think about investments? Because I think it answers this question. The way that we think about investments—and the reason is the only way you can scale because you can actually distribute this kind of algorithm to a team—is the only sin is picking the wrong company in a certain space. Because of that conflict thing? Because you're conflicted out of the winner. Investing in a space that doesn't work is fine. There's no way you can actually predict whether a space is going to work or not. That's like weather prediction. But in a given space, if you know all of the companies, you do the work, you can most likely at least tilt it. You can do the work to determine if you think one is better than the rest. That's something you can actually put effort into.
So the way we view the world is: first you have to identify legit spaces. We think that founders are smarter than VCs. So I don't care if VCs think it's interesting. If there are five founders in a space and they're good founders—they're betting their families, their fortunes, their time—it's probably a real space. Then we do the work to understand the space and all the teams. And then we make a pick within that. That's really how we think about it. That's true within AI as anything else.
The thing that's harder is—and I've evolved so much as an investor—I used to think, "Oh, we'll get good deals. Price matters, outcome matters, TAM matters." And more and more, especially with AI, that's what you have to throw away. The market is the market. And that's what matters the most.
Jack Altman
You're saying it doesn't matter at all?
Martin Casado
I'm saying we don't know what the TAM is because it's growing so fast. Nobody knows valuation. I think this is contrary to common belief. In these times where you don't know the TAM and things are moving quickly, you definitely want to pick the best team. You definitely want to pick the best team. I don't think you should overthink the space. Asking questions about TAM or valuation or value makes a lot less sense because that's actually what's uncertain. You just need to be in the best one. You have to be in the best ones. And the market will just produce what it produces. The market is the market.
Listen, either you believe that the stuff is expanding very quickly and markets are efficient, or not. Having been through the dot-com boom and bust, the reality is the market was actually pretty smart. If you put the bets in the right companies, that would have been generational wealth. Same thing with cloud, same thing with mobile. The goal is finding the right companies. If there's one change, I would say you need to throw away too many thoughts about market sizes and TAM.
Jack Altman
Is the rational sort of behavior then to invest as early as possible, but only when you know that you've got the right one? Does that ever push you to being like, "I'd rather wait around or two here"?
Martin Casado
Oh yeah, all the time. All the time. These are all very rough heuristics. We get it wrong all of the time. I've made so many mistakes. You're just trying to beat the market. You're just trying to have... But yeah, very often our discussions are like, "Do we actually know who the winner is?" And often we wait for that reason. Basically the earliest that you feel confident you can pick it.
When we do seeds, normally it's the person that did the thing in the big company and now is doing the thing and is the world expert. And the thing is a very technical thing. Somebody else isn't just going to wake up and decide to do it who's a good founder, because the pool of people that do it are like five. And this one's the best of the five. That's kind of how we think about early investments. Most everything else is the result of a lot of market work. And then this is our best guess: best approach, best team, best market. This comes from market. And market is totally right. If you think you can outsmart the market, I think it's very tough. So just being in the best deals.
Jack Altman
As a final topic, I want to hear your perspective on the board relationship and board roles in general. As a prompt on this, I feel like you've done something which I find very impressive: it seems like you're able to manage successfully many more board seats than a lot of people. I often hear common wisdom that it could be 10 or 12 or 15. But it seems like you found a way to do more than that and be very effective with those founders. So I want to hear your perspective on what that relationship is, what you think is the limiting factor here, if any, and how it all plays out.
Martin Casado
I do think that a lot of the common wisdom on boards came from that earlier era in VC when people would literally choose VC for a life choice or whatever. If you come from pretty serious operating like you do and like I do, we've just got a lot of hours in the day to throw at it. But VC has also evolved—we've got better platforms that really help with these things. Between the hours in the day... I mean, how much does that board work take? You're a board member, you know.
Can I just take a step back? Let's talk about what boards mean. I always ask when I invest in a founder, "What is a board for?" What do you think they say normally?
Jack Altman
I'm actually curious what the most common answer is. I could imagine. What do you think it would be?
Martin Casado
I would think they would probably say something like governance and approvals.
Jack Altman
No, that's what they should say. That's what you and I would say. They say to provide guidance, to help with hiring. I'm like, "No, that's not a board." There's a lot of this belief that a board member is somehow helping with company building. It's actually hard for a board member to be the best friend of a founder for those types of things because you're a fiduciary and you do governance. The actual board work itself from the fiduciary governance standpoint is just not a lot. So often implicit in this question is non-board stuff: "How can you be helpful to a company?" That can take a lot of time. I do think having a big platform to lean on and being totally available helps a lot. But it's not the board work—it is the other stuff.
I would say to anybody listening who is a VC: you can take as many board seats as you want. The actual board work itself is not the issue. The question is, can you still be available to founders and add value whether you're on the board or not? A lot of the companies I spend the most time with are not even ones I'm on the board of.
That's what's funny to me. People talk about this with me. There are some board seats I have where the founder is asking quite a lot less than some seats where there's no board seat. They're sort of decoupled.
Martin Casado
Exactly. So I feel like we need to be very clear. A board is to keep everybody out of jail and to do the right thing for the shareholders. The actual work requirements of that are relatively little. That's actually not the hard thing, but we use it as a proxy for the hard thing. The hard thing is, how do you add a lot of value? For that, I have the pleasure of working with the best team I've ever worked with in my life. They're amazing. We've got a phenomenal platform. And I get to leverage all of that. It's not a board thing—it's the "help the company" type thing. I just think this is the new era of VC. You help these companies with more than just one person who shows up off the golf field every Thursday.
Jack Altman
Well, that's a great place to leave it. Martin, thanks so much for making time for this. I really enjoyed it.
Martin Casado
It's a pleasure. That was great.
Music fades in
Host
Thanks for listening to the a16z Podcast. If you enjoyed the episode, let us know by leaving a review at ratethispodcast.com slash a16z. We've got more great conversations coming your way. See you next time.
Disclaimer
This information is for educational purposes only and is not a recommendation to buy, hold, or sell any investment or financial product. This podcast has been produced by a third party and may include paid promotional advertisements, other company references, and individuals unaffiliated with a16z. Such advertisements, companies, and individuals are not endorsed by AH Capital Management LLC, a16z, or any of its affiliates. Information is from sources deemed reliable on the date of publication, but a16z does not guarantee its accuracy.