Why We Are in a Bubble & Now is Frothier Than 2021

Thanks, Harry, for having me, and likewise. I’ve been a long time listener and fan, and I’ve heard the best things about you from a bunch of friends who have been on, so I’m excited for this conversation.

Yeah. It’s a great question. So I grew up in Houston, Texas. Was born and raised there all through high school and then moved out to California for school. There’s a lot in my childhood that shaped me. Think maybe two things are probably most formative. I wasn’t built for sports, but I was really good at debate and really competitive in policy debate in school. I think what that taught me is just a love for competition, and whether it’s on the venture business when we’re competing to work with the best founders or the companies we partner with, I think you have to be mega competitive to be good in our industry.

And then the second is, I also did a lot of biomedical research and work around designing early cancer detection techniques that we published on when I was in school, and that taught me just the power of small teams. And with the small research team, were able to produce some pretty cool work. And I think that’s been quite formative to now what I do, is invest in teams that are really, really small, and then help them go on to build really formative companies.

I mostly agree, but I’ll tell you there are exceptions. Right? Like my partner, Asheem, the data would suggest is a truly generationally great investor, has been doing this for more than twenty years, has had many IPOs under his belt, and still works seven days a week. And, you know, if he learns with a founder of fundraising that, you know, he thinks is high quality, he’ll drop everything he’s doing and go meet that person tonight. So I think some people are just wired that way.

If And you’re wired that way, I think it doesn’t matter if you’re 20 years old, 30 years old, 40 years old, 50 years old, but I think if you’re not wired that way, you’re dead, and it takes a while to notice, given the lagging nature of our business, but this is a super, super competitive game.

The short answer is yes. We are in an exuberant bubble. I think we may not even fully appreciate in how big of a bubble we’re in. And what’s happening, I could argue, is even crazier than what was happening in peak 2021, twenty twenty two zero interest rate period. It’s not unusual for us now, again, in kind of pure AI to see seed rounds for companies that are just getting started, being priced in the many tens of millions of dollars, even a $100,000,000 plus post money ranges.

And then companies that have a little bit of revenue growth, you know, raising at a 100 times, even 200 times their revenue. That’s with a backdrop where when you look at the top public names, I think the most valuable public company today for on multiples basis is CrowdStrike, and CrowdStrike is trading at, I think, 20 times forward revenue. Right? And by the way, there’s, like, maybe only five companies trading at north of 15. So you have this odd thing that’s happened where in ’21, we were investing at a 100 times revenue, but you had public names trading in the 50 to 80 x range.

Now you’re investing at a 100 to 200 times revenue when public names are trading at the best public names are trading at 15 to 20 times. That dislocation does not make sense to me unless you believe that these companies fundamentally have much more persistent growth than prior generation of software companies. And like, look, it was just reported, I think the information ran a piece last Friday that OpenAI is at 3,400,000,000 of ARR, and it’s doubled in the last several months. That’s impressive. But I would question how many companies other than that are going to be able to persist that rate of growth at those scales.

And if you don’t have the growth persistence, it’s very hard to see any of this investing make sense.

Absolutely. I’d say that’s a really big distortion that’s happening, Harry, to your point. And the other big distortion that’s happening, which again is is a positive, but I think it also cuts negatively, is some of these companies are growing explosively. When you look at the revenue growth, it is amazing. Yet, at the same time, the long term retention on these companies is very unclear. And I don’t want to pick on a specific company, but take your favorite AI prosumer application that’s growing really fast and just go run the Google trend graph on it.

For many of them, you’ll see an amazing peak, and then also it really quickly comes down. And so, you’re investing in revenue today and you’re putting some multiple on that revenue, but I think you have to ask yourself, have these products really proven persistent user value? And are you really buying revenue that’s kind of going to recur with high net dollar retention, all the things we love? I’m not sure. And so I think that’s also creating a bit of a distortion. Do

It’s a good question. The way we reconcile this is we take two lenses. One is we look at the data. And as my partner Reid always says, when you see the data that that’s explosive, the question you have to ask yourself is not why invest, but why not invest? You need to come in with an orientation of something seems to be really working, we should be in this business, and then convince yourself why you shouldn’t be. So we are certainly reactive to that. But the other lens we take, which is an equally important lens that I think the venture industry has sort of forgotten in the last year is a very fundamental lens, which is what are the market dynamics?

Does the product really have product market fit? What does retention look like? What is the defensibility of this business? And if we don’t like the answers to those questions, it doesn’t matter if the company’s gone 0 to 20,000,000 in six months, we’re not going to invest. We feel completely comfortable with that decision. Again, it’s early, but we’re seeing some of the first generation AI apps that two years ago, everybody was talking about writing assistance and the rest that had unbelievable growth metrics, you know, now come down quite a bit and really sober.

Hasn’t that always been the challenge about SaaS investing? At least in the eight years I’ve been doing it, there’s always been multiple competitors doing very similar things. Could argue now there’s more excitement. Maybe there’s seven competitors instead of two or three, but I feel like that’s always been a question we’ve had to ask ourselves as investors. And I think for us, come back to like the basics, which is which one has the best founder and the best management team? Who has a point of view on what’s going to drive real product depth and workflow and stickiness and defensibility?

Where is the distribution really unique and self compounding? But candidly, Harry, those are the same questions we ask ourselves on, you know, non AI SaaS companies.

I don’t think it is. And I think this is like one of the biggest misnomers in the venture zeitgeist where everybody’s talking about, like, hey. Are these just rapper companies? What makes these things defensible? Is value going to accrue to them? I don’t think any SaaS software company is rocket science in terms of the underlying technology. I saw a tweet from Brian, the founder of HubSpot, maybe yesterday or today, where he said like, hey, people used to say HubSpot was a wrapper on a database. Right?

And just some workflow on top of a database. And I think we’re gonna look back on this discourse a few years from now, and it’s gonna feel very similar. I come back to, like, the SaaS companies that have really become market defining, build for a specific end user, very deep and valuable workflow that becomes very sticky and critical to that person’s job and are able to have significant pricing power. And then they get to distribution before others who have the distribution, namely incumbents are able to copy their innovation.

And the ones that have done that really well, HubSpot being very high on the list, Figma and our portfolio being high on the list have gone on to be iconic businesses. By the way, SaaS is littered with lots of point solutions that didn’t do that, had more superficial value. Maybe they got bought a lot during the COVID period, and now they’ve completely flatlined, and there’s no growth or there’s decelerating growth, including names you and I both know well. And so, you know, we’ll see that happen in AI as well.

So we have to start by acknowledging OpenAI is ruthless, and the quality of their execution is just incredible. And we should expect them to continue executing and shipping amazing products. Harry, I’ll tell you my mental model for this. It’s not quite consumer versus enterprise, but the way I think about it is there are some applications that I use the word, and this word is overloaded, but I think it was very foundational, like very foundational primitives and workflows on top of AI. And you talked about, like, the calculator or maps on the iPhone.

You could say those are foundational capabilities for the iPhone. Content generation and writing, editing is a foundational capability. ChatGPT is really good at it. I think coding is a foundational capability. And I think if you just take kind of pure code generation, I would bet very strongly that OpenAI is going to compete ruthlessly on that. Now, do I think building a copilot for lawyers or for physicians or building software development tooling that’s not the cogeneration itself, but maybe it’s the incident response and SRE workflow. Maybe it’s debugging tools.

Do I think those are gonna be kind of core foundational things that OpenAI is going to need to own? I don’t. And so kind of our lens is we’re not closed for business, but we’ve really got to believe to back a team that’s competing on something that we’d call quote unquote foundational, just like we’d really need to believe if someone wants to go compete with the Apple calculator or maps. But we think there’s like immense, immense opportunity in really focused applications on top. And that doesn’t mean focused applications are small.

I think there’s going to be many, many large companies, public companies built that will be focused applications on top of these large model capabilities. But that’s where we’re focused.

I’m unsure of exactly what the opportunity is going to look like. And I think anything anyone on your show says, you should put high error bars on. Because what the last several years of AI have taught us is it’s remarkably hard to predict the future and how this technology plays out. And even the people building it are unsure. And that’s why we talk about emerging capabilities and the rest. Now with that said, our mental model is there may be one or two players who are able to persist a lead and actually compete selling models as a service.

Right? API clouds, etcetera. OpenAI is obviously a leader there. Anthropic has a scaling business there. Even that you could debate. Like, you and I could have a healthy debate on three years from now, is there a real separation between those models and the next size? And if that separation is small, can those models really command margin? And so I think we at Greylock have taken the point of view of like we’re less excited about that as a space. On the other hand, we do think there were some applications where you need to own the model because the model needs to be tied to the application, whether it’s a personal agent as an example you referenced inflection.

And for those, we actually do believe money can be made and will be made at the foundation model layer. So what excites me about an OpenAI in terms of its opportunity to create enterprise value is its first party products. ChatGPT is an outstanding product. I think they’re gonna have more outstanding products. And those products are gonna require a really deep tie in between the model and the application. And I think those can have enduring value. Think if you’re just building like an API service, I think there’s a lot of value this year and next year.

I’m skeptical how it plays out.

Personal agents, possibly horizontal enterprise agents. I think code generation is one where you might want really tight tie in. There are companies taking both approaches, and we’ll see what ends up winning. But those are three that I think require really tight tie in. Whereas you mentioned customer service AI, that’s an area where I don’t think you need tie in, where the startup should be really focused on everything except for the model.

You know, I do and I don’t. I think a lot of it depends on how your product drives value. I think we’re gonna see hybrid pricing models where you’re gonna layer on to seats the type of work that you’re doing for the end user, and you’re gonna monetize that separately. And that will allow these companies to continue growing inside accounts, even if seats are not expanding.

I couldn’t disagree more. I sincerely believe it is the best time in a long time to be building SaaS companies and investing in SaaS. And I’ll tell you why. If you think about what have been the largest outcomes in SaaS, most of them are deep systems of record for important horizontal functions. We talked about Salesforce, we talked about Workday, which started at Greylock, ServiceNow, HubSpot, another great example. Now, why has there not been a new deep horizontal system of record SaaS company built over the last five or six years?

I can’t think of one. There have been vertical ones built. There have been features built around the major systems like a Gong built around Salesforce, but no one’s been able to go after the core. I would argue you can only go after the core when either the data model profoundly changes, the delivery model profoundly changes, or the interface changes. What’s the data model? So Salesforce has an opinion on how you should run sales at your company. There’s a customer object. There’s an account object. There’s a relationship between the customer and the account.

There are different dimensions around that that your reps are filling out. And they’ve picked a model. We can debate if the model is good or bad. The reason why it’s important is because everyone around them, whether it’s the users, the sales managers, the channel partners who are implementing Salesforce, or the tools like Gong and the rest that have been built on top have adapted to that data model. And so that data model, that schema, if you will, has become the standard for how people think about CRM.

What’s the delivery model? Delivery model is how the software is actually delivered and consumed and priced. Right? I would think about the shift from on prem and Oracle Siebel to cloud based, SaaS based. And now we may see a new shift in delivery, which I would argue is mainly going to be on pricing. It’s not just seat based, but it’s seat plus work or work only. That’s pretty disruptive.

The interface. So Harry, you probably have a lot of sales reps in the companies that you work with who complain about Salesforce. They hate using it. They hate the UI. It’s super clunky. They’re constrained by the interface, yet they’ve learned how to use it. If you hired me as a new sales rep tomorrow, you’re going to train me on Saam. Here’s how you do your work inside Salesforce. With generative AI, the entire interface could change. There may no longer be an interface. I may have an agent that’s working alongside me as the sales rep helping me do my job, and it’s navigating all the underlying systems on my behalf.

And by the way, when it doesn’t know something, it’ll come back to me and it’s not just going be a chatbot, it might be a UI, but it’ll be a generated UI. It’ll be very dynamic. And I would argue in ten years, sales reps won’t even know what the Salesforce UI looks like because the interface to the underlying system of record has changed. So

Yeah. So now to connect it back to why is now a great time to be investing in SaaS? If you buy my point that the largest outcomes in SaaS come from when you’re able to go after these horizontal big application companies. Right? Again, think Salesforce, think ServiceNow, think Workday, think SAP. You need disruption on these three buckets for the opportunity to be real. Otherwise, it’s just incremental. And I’ll give you an example of what I mean to make it more concrete. There are many, like, mobile CRM companies.

Like, hey, I’ll put, you know, CRM on the iPhone. Great. That’s not that different. It’s just another client for interacting with the CRM. Salesforce now has a mobile app, you and I don’t talk about any of the mobile CRM companies anymore. Let’s take the last eight years that I’ve been investing. Those three things have not been true, which is why I would argue it’s been impossible to go after those large, you know, platform companies. I now think they are true.

I now think you don’t need the data model that Salesforce suggests anymore, because your AI agent can go and just suck up your inbox, suck up all your Gong call recordings, and on the fly materialize the views of data it needs to help you run your sales team. So when you’re doing a pipeline forecast, you don’t need to go into Salesforce. You can just ask an AI application to generate your pipeline forecast for you. By the way, it’s gonna be a lot more accurate because it’s not gonna be based on what your sales reps put into Salesforce.

It’s gonna be based on the actual raw text of all the interactions with your customers. That’s on data model. On delivery model, pricing’s gonna change to our earlier conversation. The way you think about seats is not going to be the same, and sure Salesforce over time will adopt that, but there’s going to be a window where they won’t adopt that because it’s going to cannibalize their business. And startups can come in with really disruptive pricing. And then number three, and most importantly, I think the interface is going to change.

I don’t think we’re gonna be spending our time doing pivot tables in Salesforce and figuring out where to put in dimensions. We’re gonna be talking to an agent. That agent’s gonna be consuming information from us, and it’s gonna be navigating systems for us. And those three things mean that you and I could go start a very different CRM company, and I think actually build the next Salesforce, but something that feels radically different. The same is true across all of these functions. That’s why I’m excited about going and investing.

I think it’s too early to say. A year ago, everyone got really excited. All these public companies were telling analysts, hey, people are gonna pay a lot more. I can’t remember exactly what week it was, but there was that week, three weeks ago, I think Salesforce reported earnings and all of software had a bit of a correction.

I think people like saw like, hey, actually it’s not going to have the impact we thought at least this quickly. But the bet I would take is if you take a slightly longer term view, I do think you will be able to increase pricing power. But the way you’re going to do that is by really replacing elements of work. And if we take the sales rep example, actually making the sales rep twice as productive. It’s not just like a copilot inside Salesforce. I think I’m less likely to pay more for that.

But if Salesforce would come to me and say, hey, Saam, you don’t need a BDR anymore because I have this AI BDR. And they can go out and email Harry and get Harry on the phone and book a sales qualified meeting with Harry for you, and you just show up and that meeting is booked on your calendar. Well, suddenly, that’s really valuable. And so I I could see myself paying a lot more as a sales rep for a tool that could replace that element of work for me.

I just think it’s too early and we haven’t yet seen again, you and I are seeing it in the seed stage founders that are pitching us, but in terms of, like, companies at real revenue scale really delivering replacement to parts of work. And when that happens, we’ll see pricing power improve. Listen. You mentioned kind of the founders that

Let’s talk about pricing first, and then is it rational? And I’ll start by saying the thing we all know, which is there’s a wide range, but let’s talk about kind of what is the bread and butter. So in Seed, if you are a high quality team coming out of a company where, you know, you’ve been on a growth clip, you’ve seen how a great business gets built, and you’re building a new company in an area that is secular, right? Maybe you’re building an AI BDR, maybe you’re building a copilot for lawyers.

I think seed pricing is between 20 to 40 post. For Series A, based on what we’re seeing, again, that cohort of companies, if you have a little bit of product market fit, again, we’re seeing pricing between 80 and maybe 100 and 50 and 200 post on the high end.

There’s variance, but no, I’d say in many cases, I’m seeing teams where the company was incorporated last week raising in that range. Is that rational? I think one of the distortions of the twenty twenty to twenty twenty three period is we all forgot about the power law, and we all forgot that very few companies matter, but the companies that matter end up being much larger than we think. And so with that lens, I actually do think it’s rational. And I get really upset when I hear people say, I love this company at 20 post, but at 30, it’s too expensive.

And the reason for that is like, if you run the scenario math and you ask yourself, Harry, what is the outcome for this company where at 30 I’m disappointed, but at 20 I’m thrilled? The only possible outcomes where it would matter are really intermediate ones, right? Okay. The company sells for a 150, and net of dilution, you make a five x versus a three x. That’s irrelevant. It doesn’t matter. Maybe it matters if you’re really and I’m not talking about a $150,000,000 fund. I’m talking about a really small fund.

But for most funds, those outcomes just don’t end up mattering. All that matters is are you backing the team that has a chance of building an iconic enduring company? And those teams are really scarce. So when you find one of those teams, I really don’t believe you should be passing on price in those ranges. Because whether you do it at $20.25, thirty, thirty five, 40, if that company goes on to get to a 100,000,000 plus of ARR with high growth, you’re gonna be very happy as an investor.

And if it doesn’t, and it has an intermediate or mediocre or bad outcome, you’re gonna be unhappy. And maybe you’re slightly less unhappy if you paid 20 than 30. That’s not the thing to optimize for.

And this is where it probably comes down to, like, the specifics of the situation, and it’s maybe harder for us to reason about generally, because I would argue that in many cases you are being paid for the risk. I’ll give you an example. Like we backed a team coming out of NetApp where they had previously built and scaled a business and they sold us an app for $500,000,000, and they’re building a new cybersecurity company called Upland that’s off to the races. And we led a really large seed round.

In fact, I think the seed round, the round size was $26,000,000 Now you could say, hey, Saam, that’s crazy. You’re not getting paid for the risk. But I take the counter and say, I’m getting in business with a team that has built a business to a 100,000,000 of ARR before, understands how to build products, sell products, scale management teams, think about how little risk there is on so many dimensions around that project. And so I’d much rather do that than do a $2,000,000 seed round where I’m backing a team where I don’t know if they actually know how to build a product, sell to the enterprise, recruit a VP of sales.

There’s so many layers of risk. And so I agree with you, Harry, generally speaking, but I would say we just have to look at the specifics of each situation. And I would argue there are many situations where the expensive seed round is a lot less risky than people think.

So in that case, the valuation was not that high, and I agree with you. Like, I think you have to have a ceiling on valuations you’re willing to pay.

Exactly. That’s an excellent point. And that’s why we maintain two truths in our head to use your parlance. One is for the right situations, the rules don’t matter. And then the other is at a portfolio level, the rules do matter because of every investment in your portfolio is an exception, then there’s no actual portfolio construction rules that you’re enforcing. And so for us, you know, we invest, you know, we’re investing in our seventeenth fund. It’s a billion dollar pool of capital. We are more ownership sensitive than price sensitive within a range.

And so the lens we take on our kind of core early stage positions is how many of them are there where we have adequate ownership. And we think of adequate ownership as 20 to 25% plus. Right. Do you still get 20 to 25% plus? It varies Harry, and we’re not as religious about that as we once were. In many of our companies we own in those ranges or more. Now, the reason why is because we’re willing to give founders more capital early when there isn’t proof because we have belief in them.

And so we are willing to do things other people may not do.

I think it all depends on the founder. It just all depends on the founder. And the more I do this, the more I come back to like everything is about the founder. I can tell you, Harry, there are situations in our portfolio where we’ve done large seed rounds and where the velocity and desperation and paranoia is unlike anything you’ve seen because the founder is just amazing. And there are situations where we’ve done a million dollar round, and there isn’t that orientation. You are right to say that when there’s more capital, perhaps there can be more of a temptation to overspend, not be as scrappy the rest.

But I think the best founders, they inherently have the right orientation on those dimensions. And so the capital just becomes an accelerant to them. It allows them to think bigger, build more of a complete product, target larger customers from day one, and that makes a big difference.

Harry, we make very few investments at Greylock. Each partner might make one to two investments a year, and many of them start as very small checks. The last two investments I made over the last twelve months, one was a $6,500,000 check and the other was a 5,000,000 check. But our constraint is not the capital. Our constraint is our time. Because when we make these investments, we sign up to be accountable in service of the founder forever. Like it’s not an option for us. And so we actually, on the $6,500,000 check, I spent ninety days getting to know the founders in offices before we wrote that investment where I was working with them every single day.

Is that still possible? With deal compression, with heat in markets. It is if you build the founder relationships early. In this case, we got to know these people when they were still full time employees elsewhere and we’re helping them before the company even existed. Actually, I think something like 80% of our investments, they’re of that flavor where we met the person before the company existed. And that’s not to say we don’t sometimes make decisions in forty eight hours, sometimes we have to, but that’s our orientation.

And so that 6,500,000 might grow into $50,000,000 over time. And if you look kind of over our more recent funds, whether it’s a Figma, Rubrik, an Abnormal Security, a Discord, many of these companies we’ve put real capital to work over time, and we’ve earned the right to do that with the founders, but they all started as really small checks. My question

I think it’s total bullshit. I And I think the best found Do you some. I wasn’t using that. Yeah. And of course, everyone sells their own products. So at risk of sounding like that’s what I’m doing. There’s different ways to cut this, Harry, but I’d say I’ve never seen a situation in our portfolio where it has mattered. And actually, it’s the flip, where the seeds we’ve done, many cases we lead the Series A’s, and in many cases we help the founders land tremendous Series A’s. And I, you know, I know some other funds run these stats, they publish them on Twitter and all the rest.

I don’t have the exact data, but I can tell you that our seed companies almost always raise Series A’s, and the conversion rate to A is is much higher. That

Maybe I misunderstand the signaling argument, but my view of the signaling argument is like, hey, if I raise my seed from Greylock and Greylock doesn’t leave my Series A, is it less likely that another investor will leave my Series A?

Harry, I’ll I’ll give you an example. Like, I led the Series A in a company called Cresta. Right? I may have mentioned it earlier, contact center AI companies, Sequoia led the B, company’s doing great. Andreessen had done the seed. I remember when we looked at the company at the A, I was like, Hey, why isn’t Andreessen doing the A? They probably passed. I mean, I don’t know. I still do this date don’t know, but I assume they passed. Doesn’t matter. We evaluate the company on the merits of the company.

We love the founders. We love what they’re doing. And, you know, we did the A. And I’m just picking mean, that’s the first example that jumped into my mind. There’s so many ones where we did the A, where someone else did the seed. We just look at it as independently, especially with some of these global platforms that are doing fifty, sixty, 70 seed deals a year. I think it’s hard for them to even keep track of what companies are investors in. And so I’m not going to let that get in the way of backing a great founder.

Yeah. So a couple of things. One, I should say, if we own 20% of a company at Seed, like we will view that as kind of being in our core ownership bucket. And so it’s less likely we will necessarily have to lead the next round, whereas if we own 10%, it’s a different equation. You know, Harry, like on every round, we sincerely ask what is the right thing for the company. And our view is most people in venture capital are not helpful. They’re not additive. We don’t think founders need incremental brand when they already have the Greylock brand behind them.

And so we challenge founders, and we say, Build the list of people who if we get them on the board, they will actually change and have impact on this business. It’s a really small list. Every time I do this exercise with the founder gearing up for a fundraise, I’m disappointed by how small the list is. I mean, you mentioned our shared friend, Pat, earlier. He’s on that list, but there’s not a lot of names on that list. What are the common names on that list? You know, an enterprise where I play on the early stage side, I have a lot of respect for Pat.

I love Eric at Benchmark. I’m on a board with Ravi at Lightspeed, who I think is excellent. I’m on a board with Ted at Kleiner, who’s excellent in security. Like, there are names, but it’s not 50 names. It might be five to 10. And I think the names vary a lot based on the domain that you’re in. I can’t talk about this company yet, but there’s an exact one where we led the seed. We built that list. We would have loved to lead the Series A.

The founder went, pitched the three names on the list. All three of them wanted to invest. He picked the one he liked the best, and we invested as well. Do we own a little bit less than we would have if we had led the a? Of course. Is that the right thing for the company? Of course. Because that person’s gonna be additive to the business. And if that makes the company bigger, like, we’ll all do better. And so we’re not in the business of like, wanna own 40% of companies and do every round.

We’re more than happy to win great businesses, but we first always ask ourselves the question, is there someone out there that could actually be additive? And if so, let’s go get that person. But if we’re just gonna go get capital from someone who’s gonna show up to board meetings once a quarter and not do anything, like, I’d rather give the capital because number one, we can do it at a better price since we already have an existing position in the company. And number two, let’s keep the company tightly held.

Why not? It’s better for the founder.

We don’t have a super formulaic perspective. Like we, a portfolio level, think about it. We reserve a lot generally because we invest in things early. We’re fortunate that at least to date, a good portion of them have had the chance to go long, which means even just the pro rata dollars really add up. And so I said, it’s not uncommon for us to start with a $5,000,000 position and it yields a $707,500,000,000 dollar position. By the way, the other thing on reserves, which nobody talks about anymore, is like reserves aren’t just for when things are going great.

It’s also for when like companies hit an air pocket. And many of the great companies we’ve been a part of have hit those air pockets. And we’ve had to step up. And we have gladly done that and led inside rounds, not out of offense, but out of defense. And we won’t have the capital to be able to do that. So I’d say we run a relatively high percent of our fund in reserves so that we have ample opportunity to play offense and defense for our companies.

Yeah, this reminds me of what we were just talking about when an investor says they like something at 20 and not at 30. So it drives me crazy when I hear, you know, investors write these blog posts, your checklist for the Series A, Number one, a million dollars of ARR. Why does that drive me crazy? Because for you to make money on a Series A, and let’s just take SaaS or Enterprise Software to simplify, you’ve got to get into a business that gets into the hundreds of millions of ARR.

That’s the only way you can become a public company. And so if your goal is to get into the hundreds of millions of ARR, and your main thing that you’re looking at is, is this a company at a million or 2,000,000 of ARR? Harry, how many companies get to a million of ARR that don’t get to 10, that don’t get to 50, that don’t get to a 100? The vast, vast majority. And so my argument is like, I don’t get why that’s the primary thing that people look at.

I really believe it’s misleading because you might look at that and be like, oh, this thing’s at 2,000,000 of ARR, it’s great. When it turns out that the market’s super capped and it’s gonna grow to 20,000,000 and then massively decelerate and the business is gonna be worthless. Or turns out, you don’t really believe it’s a superstar founder. And conversely, we we didn’t invest in this company at the Series A. We fortunately invested later in the growth rounds. We take Wizz. I’d argue Wizz is one of the most important privately held companies today.

Wizz, they raised their Series A at a $500,000,000 valuation, had zero in ARR. But what did it have? It had an amazing iconic founder, and it was going after the cloud security market at a time of dramatic transition. And look, 500 is a big price. Most of them don’t need to be that high, but I would much rather take that shape of bet and bet ahead on the ARR than be happy that I’m, you know, getting into a business with a couple million of ARR, but where I’m in a stunted market where I don’t believe I’m backing an iconic founder.

And so it’s not that I don’t look at ARR. I look at it. It’s an important proxy for product market fit, but I always start with who’s the founder and what’s the market, and then I go to ARR as as a secondary consideration.

I agree with you, but I would say that great money is made being contrarian and being right. Right. And there are multiple ways to do that. I think the common way people do that, congrats to you on those companies, like is they find companies or markets that others aren’t all over, and they invest in them. And then a year later, it’s clear that things really working and they benefit from cheap follow on capital. But another way you can be contrarian is right, is to take a really competitive situation and say, hey, I’m willing to pay twice the price anybody else is willing to pay.

Because I actually am such a believer in this thing that I’m willing to price it at twice the price you are. And again, I come back to The Wiz Series A, right? That’s a good example. I think at the time, the price was astronomical. I’m sure many people passed due to the price. Credit to the people who funded it. They were right.

I think for Greylock writ large, and then I’ll give you a couple concrete examples for myself personally. We are willing to invest behind great people and great markets when the traction data is not there, often when others would pass. For instance, we were fortunate at Greylock to initiate a company called Abnormal Security, which recently announced they crossed a 100,000,000 of ARR growing north of a 100%. You know, it’s one of the fastest growing private companies today. We did multiple rounds in that company before it was clear that there was repeatable product market fit.

But we had really deep conviction in the market and in the founder’s Evan and Sanjay. You know, we mentioned Upwind and the large seed round. I’m sure many people looked at that and were like, wow, these guys are smoking something, dollars 26,000,000 seed It’s like, no. Amiram’s amazing. He’s he’s gonna be an iconic founder. That company just started sales. It’s one of our fastest growing new companies. That’s the form of risk we’re willing to take. And I think any investor has to be willing to take some type of risk that the market writ large is not willing to take.

I think the way you just articulated that is spot on, and I’m gonna steal that because it’s very clarifying for me as I think about how we look at Seas versus Seeds. You’re mostly right. There are two exceptions. One is markets have gravity, and I don’t think even the best founder can overcome a market that has the wrong dynamics when it’s too late. What do I mean by too late? If you start a SMB SaaS company that sells to other startups and you start flying the plane and the plane gets to 50,000,000 of ARR, but the overall market is $100,000,000 it’s really hard at that point to reengineer the plane and to go after some other segment.

And so as these companies get more baked and valuations go up, we pay even more attention to, is this really one of those markets that matters? And not just matters, but has the right dynamics to support a new company.

I think it’s a little bit of a philosophical debate. I mean, yes, I think the best do, there are many examples, but I think you make the odds just lower by picking the marketing correctly. And maybe the best teams are able to pivot, but there’s so many things that can go wrong as part of that process. It’s like, hey, why not just get it right from the beginning?

This question is the bane of my existence. Like it’s one we debate every day at Greylock. Really

It’s really hard, Harry. I think the only things I can offer as learnings I’ve had, and I’ve made so many mistakes, so many mistakes on this exact question. It’s painful. Is one, you’ve got to be really honest with yourself. Like, is this truly an iconic founder? Right? Is this, you know, the next George Kurtz at CrowdStrike or near Zucca Palo Alto Networks? And that’s a really high bar. So one that. And then two, I think we are more likely to do it if we think the general zip code is good.

So like if you’re in a good zip code, but the street you’ve picked to build your house on is a bad street, that we’re more willing to get behind because we’re like, Hey, Harry’s smart. He’s going to figure that out. He’s going to shift streets before he starts construction. But if you’re in the totally wrong zip code, it’s hard for us. Does that differ by stage? Will you do it at Seed? We’re more likely to do it at Seed. But even at Seed, that’s been where we’ve made some of our biggest mistakes.

I’d say at Seed, our biggest mistakes where we have not pursued the investment has been where we’ve met someone truly iconic and we did not like the idea. And by the way, Harry, often we were right. Often two years later, the founder was working on something different, but we should have been in business with them.

Yeah. It’s not true again. I see the same things you do, and I I just don’t understand the data people are looking at because I see a totally different battlefield. Do

Exactly. You nailed it. And it comes back to the power law, right? Which is very few companies matter, but the ones that matter become much bigger than you think. If you’re a company that’s building something secular, maybe it’s AI enabled, maybe it’s not by the way, you have great growth, great market dynamics. This is gonna be controversial. I think the Series B market may be even frothier than it was in ’21.

That’s what I thought too, but I think the thing maybe we both missed is those players would leave the market, but there’s a new set of players. And those players are the early stage platforms that have raised these massive growth, global growth, etcetera, funds, incredible amounts of capital and dry powder at these platforms. These platforms have all hired a lot of new partners who need to come out of the gate swinging. And because those platforms don’t have public market teams, they don’t have to face the harsh reality of where public multiples are the way if you’re at a Code two or a D1 you’re thinking about every day.

By the way, that can also be an advantage because you could argue it allows you to be longer term, blah, blah. And those players are competing for a smaller set of companies. Because in ’21, so many SaaS companies were growing from like one to four to 12. Yeah. All those companies got funded at big prices, but there were a lot of them. Now there’s fewer. But the ones that have it, you overlay that with all of this demand, the pricing is exuberant. Mean, I really wonder if the Series B asset class as a whole is positioned to make money this vintage.

Wow. I mean, that’s

bold statement. Again, Harry, come back to unless you’re betting that public market multiples are going to re expand or that the growth durability of these assets is a lot stronger than what we’ve seen in the past, the vintage writ large is not going to perform. It does not mean that there won’t be outstanding Series B investments made. There of course will be. The vintage and the basket overall, I don’t think will perform well.

And it’s incredible. It speaks to who he is as an investor, that he’s able to do that.

Pat’s obviously one of the all time great growth investors. And so to your point, I think it’s interesting that that’s what he’s doing when many others are playing the, quote, unquote, game on the field and paying up for these Bs and Cs. Do you agree with the play the game on the field mantra? I don’t. I think playing the game on the field would have cost you dearly at many different periods in time, both in the generation that you and I had been investing in and the generations that predate us.

So I think you’ve got to be really cognizant. And if you’re not playing the game on the field, you better be right. But I think you have to have independent thinking.

There are many, right? The two that like first came to mind. The first is Glean. We at Greylock have known Arvind Jeyne, co founder and CEO of Glean for a long time because we were the Series B investors of Rubrik, where he was one of the co founders. And so I’ve known him for a long time, have thought super highly of him. When he started Glean, I took the meeting on the think the Series B, the round that General Catalyst ended up doing. At the time and still now, but at the time, you know, there was an enterprise search company.

Enterprise search as a market was littered with so many people who had tried and had been unsuccessful and who hadn’t been able to crack enough end user value to get recurring user love. And you know, the business was early. And so we made a really bad decision, which is we didn’t try to win the right to invest. And today I’d say Glean is on its path to being an iconic company. In my view, it’s one of the most important AI application companies. And I think anyone who’s not using it should immediately start using it if you’re a large enterprise.

And so if I reflect back on like, what was the learning? I think there’s two learnings there for me. One is Arvin is an iconic founder. He had a long tenured career at Google, started Rubrik, which has gone on to, you know, just went public recently, become the defining company in the next generation of ransomware protection and data backup. And when you have the opportunity to work with someone like that, you don’t overthink the dynamics around the market because he for sure understands it better than you do.

That’s number one. And then number two, I think the thing we also failed to see, which I think was hard to see at the time, but we failed to see it because this was a few years before ChatGPT, was that the rise of generative AI would fundamentally change the way search in the enterprise worked and what would be possible. And Arvind and his team were the best positioned to take advantage of that and build truly transformative experiences. So that’s one. And then the second is a company called Codeium.

It’s a company in the code generation space. We’ve known the founder of Varun for many years. In fact, was a Greylock fellow when he was still a student in college. You spend thirty minutes with this guy and it is very clear to you that he is of the making of these very, very, very special entrepreneurs. But when he pitched us on his seed round, he was building a completely different company idea. At the time it was model inference, which ironically now turns out to be a great idea, but several years ago, you know, there wasn’t a lot of demand for, and so we made a mistake of not trying to compete to invest in his round.

Did you change how you invested on the back of these? Yeah. So now at the seed, when we see someone like a Vruin or an Arvin, our orientation is towards yes.

I think all three are important. I don’t want to give you a non answer, but I think one of the hard things about venture capital is you have to be good at all three, especially if you’re playing at a seed in Series A firm like I am. I don’t think you have to be good at servicing. That’s a good point. You’re right. There are many great examples of investors who don’t do the servicing, and there are many founders who don’t need that. But I think the hardest and most important is sourcing.

I believe like you are only as good as what you source. It is incredibly difficult to stay relevant, to understand where the new pockets of entrepreneurs are going to be and to see them before anybody else. And we have a nine person investment team at Greylock and we’re competing with people with 50 investors. We have to be on top of our game, and we care about sourcing tremendously. That would be my answer for both what I think is hardest, and I think what’s most important. Because if you find the right founders, you could argue you don’t need to do any servicing.

Can I

I think most young investors make a big mistake, which is they think too much about themselves and not enough about the firm. I’ll give them the benefit of the doubt, maybe it’s not their fault. And what I mean by that is a young investor shows up at a firm and classically the way they think they can have impact on the firm and progress themselves in their career is by sourcing great investments. And so they start sourcing investments and they, you know, come to their partners and their colleagues and they say, Hey, you know, this is a great company.

That’s a great company. All of these things are great companies. Because in their minds, they just wanna make investments happen. And I’ve seen many people, including people who formerly worked at Greylock and at other firms be oriented in that way. And I think those people are doing themselves a wild disservice.

I think you’re right. That does happen quite a bit. I hope we don’t ever hire anyone at Greylock who’s wired that way, but you are right that that does happen.

I think at Greylock, like, the lens we would take is why is this person going to have something to offer an entrepreneur? Like, why would any entrepreneur at some point wanna work with this person? And if you don’t have a really good answer for that, I think it’s really tough to be successful, certainly at Greylock. Do you think

I think it’s both, Harry. I think your point’s fair, but I think it’s both. By the way, I don’t just mean it because of servicing. I just think it is the case that entrepreneurs fundamentally, in my mind, make decisions to work with venture capitalists, and they have the power, and they choose who they wanna work with. By the way, that might be they have six term sheets. It might be that you’re trying to preempt a Series A in a company that doesn’t need to raise. And the founder, of course, is excited about the capital, but also has to be excited about you.

And again, Harry, doesn’t mean that you’re gonna come on the board and drive a ton of value. That’s one flavor of it. There are different reasons why a founder might wanna with you. You might have a great brand. You might have great distribution. Like there are great reasons, but there has to be a reason. I fundamentally believe that. When we add people to the team, we think about telling them, hey, what’s your reason going to be? And Harry, often that will also connect to sourcing because your reason might be, I’m going to go become an expert in fintech.

And I’m not going to be focused on like, what’s the deal I bring in this week, but I’m going to take the bet that over the next ten years fintech is going to be important. And the next year, I’m going go map out Stripe and Square and Klarna and all these great businesses. I’m going to meet the product managers, the directors of engineering. I’m going to know everybody in the next generation of fintech. And a year from now, that’s going to result in me meeting Sally coming out of Stripe six months before every other venture firm.

And that’s going to enable me to win or help our firm win that opportunity. That’s the type of thing that I think a thoughtful young investor who really wants to build a lane for themselves that a firm should be focused on. And Harry, actually think we’re saying two sides of the same coin because you become known for something. And that’s something both helps your sourcing, I would argue, and it helps over time your ability to win the best opportunities.

Sounds awesome.

I think investors can help companies in the pre product market fit stage. I think there’s a conception that finding product market fit is magic, and I disagree. I think if you’re an investor, you can be a thought partner to the founder on how to segment customers, how to orient on an ICP. You can sit in customer meetings and give real feedback, and I’ve seen it have real impact.

I love Pat, and there are many amazing people that you and I both know who I learn from. So it’s hard to pick one name, but I’ll go with Elad. I think Elad Gill is just an outstanding thinker and ambassador. You know, we’ve worked with him closely over the years. I have the good fortune of working with him at Braintrust, which is a company in the AI developer platform space. And his ability to do everything from, you know, help initiate new companies like Braintrust to lead Series A’s to do terrific growth rounds.

He’s a thinker that can extend across all stages in a single person, and I’m just very impressed by him.

I’ve also learned that mistake painfully on on companies he’s been involved with.

So there are many, Harry, but the one I would mention actually talking about Braintrust is the founder of Braintrust, my friend Ankur Guel. So Ankur is someone who I’ve known for a long time. He was the first VP of engineering at a company called SingleStore, then went on to start Impera, which Figma acquired in our portfolio, ran the AI team at Figma. But when I met up with him last year when he was getting Braintrust started, and he painted a vision of how people would actually build AI applications.

And everyone’s debating where’s the value going to accrue the tooling layer, the model layer, the app layer. And he’s like, Saam, I was at Figma. I know the team at Notion. I know the team at Zapier. Here’s what actual developers care about in their pain points, and here’s how I’m going go build a solution. There was such clarity in the way he spoke about an emerging market. I walked out of that first meeting, I texted my team. I was like, We are immediately investing in this man.

I’m glad we did. Now all of those companies are customers of his. How big was that first round? We wrote a $5,000,000 investment.

Broadly, I think incubations don’t work for a number of reasons. I think most firms take too much of the cap table. There’s negative selection bias where they don’t get the best founders to want to work with them. They can’t actually help. And so I broadly think incubations don’t work. That said, some have worked out outstandingly. And you know, Greylock, Palo Alto Networks and Workday are two largest historical outcomes.

My partner, Sheem and I were fortunate to help incubate a company called Abnormal Security, mentioned earlier, that, you know, is on its path to be a company of that And if I think about those businesses, they had amazing founders, they picked really large markets, and then they worked really collaboratively with their venture partners around recruiting and customer development to build the right team and initial set of customers out of the gate. And others like the folks at Cutter Hill have done that really well. But broadly speaking, I don’t think incubations work.

Twelve months ago, I was of the mindset that the large foundation model companies would crush all of the smaller focused models. So if you were building a model for audio generation or, you know, voice synthesis, yeah, you might have an advantage today, but how is it not going to be the case that OpenAI two years from now is going to have this way better? And so we didn’t invest in any of those. And I still hold the view that just on the raw, like generate a voice, the underlying large models will get better.

But the thing I have changed my mind on is if you can start with a better focused model and then very quickly move up the stack into the application layer, it’s actually a very good strategy. What is the best example? Eleven Labs, which unfortunately we’re not investors in, but it’s the best example.

Mean, you just go look at the product they’ve built around their underlying model. It’s amazing. And if tomorrow OpenAI actually releases, you know, the research preview they’ve shown that competes, yeah, sure. Might compete on the model layer, but I don’t think it’ll matter for 11 Labs because I think they built really fantastic workflow around it.

I have immense respect for many, but the first thing that comes to my mind is Gilly, who I think has been on your show, Gilly Ranaut, and what he’s done at Cyber Starts and the companies he’s been a part of when helping initiate and get started, it’s outstanding. I’m lucky to work with him at Upland, and we at Greylock work with him on several companies. I think he’s a phenomenally smart investor and an incredibly impactful board member.

It’s amazing.

I knew this when I started. I wish I knew how important it was, which is the importance of building in large markets and that most markets just don’t matter. You talked about Nikesh from Palo Alto. I spent a lot of time in cyber, I think Carefully Apollo or CrowdStrike, both of which are now, you know, ballpark $100,000,000,000 market cap companies. They’re operating in phenomenally large markets. And so I now am very oriented. Like, if you’re a new founder and you’re gonna spend the next decade doing something, let’s pick something where there’s no ceiling and where the growth can keep compounding.

Because why not go build a $50,000,000,000 company, not just a $5,000,000,000 company? And so I’m incredibly oriented on market size.

I think this is a misnomer. Yes. I think of markets as having concentric circles. The great things start with a really narrow concentric circle. Palo Alto started as an add on to a firewall. There’s intentionality from the founder and there’s a sequencing of how that’s going to build into bigger and bigger concentric circles. And they’re swimming in, just to go back to the zip code parlance, in zip codes that are really fertile.

I think people don’t ask enough how you actually work with founders. And I know that sounds so basic. Like, I’m stunned by how few founders do references on VCs when they decide who to go with on a term sheet.

no,

make Perry. I mean, like, when, know, you have a term sheet out on a founder and the founder’s evaluating whether he should work with you or three other great firms. Like how often do founders actually pick up the phone and call the CEOs you work with? A lot do, but a lot don’t. Is that not a sign of quality? It’s not correlated. Like I have backed great founders who did 12 references on me and I’ve backed great founders who did zero references on me. And so for what it’s worth in my dataset, it hasn’t been correlated.

But I really believe people should be asking, like, what is the nature of how you work with companies? And what would be different about your company without this person’s involvement?

This was a lot of fun, Harry. I love the conversation.