Sandesh Patnam of Premji Invest joins Nick to discuss The Downstream Effects of SpaceX, OpenAI, and Anthropic Soaking Up $3T in the Public Market, Who Is Netscape and Who Is Google in the AI Era, and the Impact of Private Credit Redemption Requests on PE and VC. In this episode we cover:
- Themes and Dynamics of the Current AI Shift
- Investment Opportunities and Challenges
- Thesis and Focus at Premji Invest
- Advice for AI Companies and Public Markets
- Impact of Mega IPOs on Private Markets
- Challenges in Private Equity and Private Credit
- Listening as a Secret Weapon
Guest Links:
The host of The Full Ratchet is Nick Moran of New Stack Ventures, a venture capital firm committed to investing in founders outside of the Bay Area.
We’re proud to partner with Ramp, the modern finance automation platform. Book a demo and get $150—no strings attached.
Want to keep up to date with The Full Ratchet? Follow us on social. You can learn more about New Stack Ventures by visiting our LinkedIn and Twitter.
0:17
Sandesh Patnam joins us today from Menlo Park. He is Managing Partner at Premji Invest, an evergreen investment firm with over $17 billion in AUM, investing primarily at Series A and Series B. Sandesh has invested in companies including Navan, Hippocratic AI, EvenUp, Baseten, Cohesity, and Stripe. Sandesh, welcome to the show!
0:42
Nick, thank you for having me. Pleasure to talk to you today.
0:45
Yeah, it’s great to have you. Can you start, maybe with your, your two-minute background and your path to venture?
0:51
Sure. The venture path, you know, never really thought it was going to be an investor. Now, 26 years ago, 26 years of investing really started life as a microprocessor systems architect, I was doing my master’s at Stanford while I was working. This is now the first bubble. Got attracted to all things startup. Did the prototypical thing, which is to stop my master’s and leave Silicon Graphics. Joined a startup, interacted with a lot of venture funds at that time, raised capital, exited the business, and during the course of that, I had requests from many VCs to do diligence on a bunch of their semiconductor investments back in the day, and got a preview into what I was, was quite exciting. One is talking to many entrepreneurs, getting deep in the tech stack, understanding product, and when my company got sold, I got the offer from one of the, one of the investors that I’d been interacting with to join venture as a place for me to go hang out for a few years, see what other people are thinking about, and 26 years later, you know, I’m still investing, it’s been obviously an absolutely phenomenal ride, but excited to continue doing this, which is, I think, the best job in the world in many ways. 20-six years, so, so nothing’s changed since he started, right? Oh my goodness, you know, history never repeats itself, it rhymes, and I hate to use that, because I feel like an old commission at the office, right? I go, well, back in the day, you know, and people are like, “Oh, here he goes again, you know, but no, this is great learnings, you know. But yeah, no, I think it’s exciting times, right? So, you know, having lived through sort of the first post 96 telecom act change, the telecom sort of craziness, and the internet bubble, he hate to use the word bubble, but you know lots of fun things that happened there, and some great learnings that came out of it, right. So, and as an investor, I’ve had the, you know, the fortunate sort of view, which is I did product, and you know, both in a big company and as a startup, I did about seven years in early, mid to early stage venture, so very early, as in series A back in the day, did public markets for seven years, and so before I started Prime to Invest here in the US, so it’s been a third in product, a third in early stage venture, a third in public markets across all formats, so Prime to Invest was sort of the perfect amalgamy of early, mid growth late public all coming together in a seamless fashion, so it’s been fun to exercise that muscle in that fashion. Yeah,
3:28
Sandesh, what?
3:29
What
3:30
are the themes that rhyme in the most visceral way between what we’re seeing now with the shift to AI and maybe some of the previous waves that you’ve lived and invested through,
3:44
so I think you know, I think no point in history, even if you go back to sort of the mid 90s, when we first went through it, has the diffusion impact of a technological shift across various industries is going to be as profound as this specific change, and to you have to go back beyond the 90s to some of the very big technological revolutions that I’m talking about, industrial revolution, and the likes to see the impact that it’s going to have on the humankind. So, from that perspective, I think it’s bigger and more massive and more impactful than I’ve ever seen in any of my prior cycles, right? So that’s the first sort of the big, big overarching thing. The second is capital intensity, you know, like if you go back even in the 90s, and even if you adjust for inflation and all these other things, I think capital intensity has is very, very, very, very high related to all other cycles, and then it’s not about capital intensity in its aggregate, it’s capital intensity in its duration, which is to say I got to deploy this capital here and now over a two five year, seven year compressed cycle, which I think makes the math that much more staggering. A third is more from a venture perspective, and this is sort of a tailor caution rather than anything else. The tail of caution. Question is, if you go back, and I can, you go back to sort of 9697 you take the Netscape moment, right, Netscape, and you look at the IPO of Netscape, and look at the three rounds of financing, and this is sort of ballpark, I’m not trying to be specific and accurate, Series A for Netscape was in the 14, $15 million range, Series B was roughly about 40. Series C was about 80 million. IPO is in the one $50 million range. And if you do the math from A to the IPO, it’s roughly a 10x right? So that’s the 10x model that a lot of venture folks kind of talk about, and the frequency of the round, which is between A and B, between B and C, and the IPO, roughly speaking, two issues, two to two, and then you think about the maturity, the company, when the company went public, roughly about seven or eight years, and maybe it’s slightly longer. I don’t, not going to be perfectly accurate about the exact timeframe, but roughly in that timeframe, and then you move from the post Netscape to sort of 99 so say 96 to 99 and you see an interesting phenomena, one is the series V of the prior round, you know, generation becomes the A, the C becomes the B. You have the staggering effect, right? So, the earlier the rounds, evaluations start picking up. What’s most disconcerting is the frequency of the rounds, right? So, the time between rounds starts shrinking, and you go from two years to year half, year half, and so now the company that’s potentially going public is less and less mature, and an absolute peak of the bubble. You saw companies raising capital within the year at absolutely insane valuations, and going public from inception to an IPO in three years. Now, look, that was absolute insanity, not that we’re going to repeat that. The thing that rises to me in this era is something sort of similar, which is to say that you know the frequency of the rounds, which means that has sufficient enterprise value being created for the next valuation, that I think is the part where I feel like investor hype and desire to be in a few names overwhelms fundamentals in a very, very dramatic way, and so if there is sort of for this disclosure thing that goes off in my head, it is the frequency of the round, not even the absolute valuation I can give away, get away from like thinking about very large markets and things of that nature, which let’s talk about discipline in the public markets at a later point, but the frequency of the rounds is telltale sign of something, and and so that to me is a sign of caution, but you know, look, every every single cycle is going to be slightly different, and this one has its own manifestations, and I feel like there’s going to be a very large dispersion here. I mean, in aggregate, it’ll be a great investing error, but it’ll be in, you know, kind of think of it on the private side, not equal to max seven, maybe the mag 25 something like that, right? So the mag 25 is going to generate a bulk of the outcomes, and then you have, you know, very – I’d say massive capital deprecation for the remaining investments, and the number of them are staggering, right? So that, I think, is probably going to be, you know, who knows, this is a prediction, but you know, I think that that may be the end result of this whole cycle of investing,
8:03
so with the dynamics you just outlined, and some of the risks that we’re facing in compressed cycles, and maybe less enterprise value when going public. What’s the sort of the the investment opportunity right on the other side of it? Like, where do you find the best arb or opportunity or upside understanding the dynamics that we’re faced with.
8:29
Yeah, look, I think power dynamic dynamics in this era is significantly more strained, right? I think in venture the biggest mistake you can possibly make is the uncapped mistake, right, is missing out on the next topic sharing, and I think that forces people to make a bunch of, you know, investments at times where the obvious is, you know, thing is to step back and see or turn the next card, and I think the nature of that makes it incredibly hard. The second, I think, is the revenue growth at the speed with which these companies are growing, you know, the think of it as maybe it’s low calories, right? So this is the question that we always struggle with, and so therefore, is that a durable revenue mode when it comes so easily, right? If you think about true enterprise, you know that’s one way to sort of, you know, test it where there’s a lot of work that goes in, and once you’re in there, you’re in there for a few years or longer, so you can actually monetize, you know, in some significant ways, you know. Those things are TBD, you know. I’m not sure what that, what that’s going to look like in this era. And we can go into AI-specific things over time, and tell you how we were thinking about it, and how it’s evolved over time. And then there is sort of this notion of, you know, intersecting the market, right, which is to say, look, I think there are some projects that are incredibly capital efficient, capital intensive, rather, and this is lowered down the stack, so I’m not trying to play in the model layer, that is, what is the next iteration of the model going to do to this set. The investments I’ve made, you know, we don’t like to play that game, which means that I want to be in the enabling side of the house, right? So, which always points down to like hardware components, inference chipsets, network silicon, things that nature, and this I feel like there is significant art here, like to answer your question more specifically, and the reason I say that is, you know Silicon Valley, and the investor base, it always has sort of a core area of expertise, right? So, and they all enter venture because there is a certain, you know, end trend that’s in play. So now let’s go back and think about it for a second, right? Either the last sort of 10 years has been sort of this era of software, and before that it’s probably an era of mobile, and before that, maybe a little bit of the cloud. All of these eras were, you know, layers to the stack that were above the silicon layer, or about the hardware layer. And so the primary expertise of the investors that are in this market comes from that expertise. They do not have, you know, deep semiconductor background, deep networking background, deep optical background. You have to go back to the 90s to find GPS and venture investors that have truly understand what that looks like, and so where the arb comes for me is that, you know, having grown up in that era as an engineer and as a product person, you know, there are certain telltale signs of why these and how these companies will manifest themselves into success, and to me that is either massively mispriced, which means that you know for the right opportunity, you know I feel like we are going to appropriately risk adjust and size our checks better, and also find the type of entrepreneur that knows how to build these companies, and not, not to name names, when you see like a dropout of some big IV institution starting a chip company in inference, and then you have a pylon effect of like 20 investors writing $50 million checks, not knowing what that means, you know, I’m not doing that investment, okay, yeah, this act requires lots of gray hair, and gosh, have you done a tape out? Do you know what you know a GDS file looks like? Do you know how you know the TSMC fab operations look like? And none of these things are known, right? So I think there is a this ARB, and the fact that a lot of people don’t understand how to build these hardwares companies. Why? And for the writer of entrepreneur, we feel like that is a really differentiated place that we can really lean in on. Awesome,
12:27
so I kind of skipped over this, but just give us the quick summary on the thesis at the firm, and kind of your focus.
12:36
Yeah, so look, I think we’re relatively thematic, so and but at the high level, because we are an evergreen crossover neck, you know, we are not looking at similar exit cycles. Duration is like a really big thing that, you know, plays in our place to our strength. We think of our primary responsibility is to hand back to the endowment a set of companies that our increased enterprise value in a, you know, you know, sort of responsible way for a long period of time, which means that we’re looking for really durable assets, and so during our diligence process, if we come up, we come away with the impression that the company is getting built for potential M and A, or the thinking is M and A, or the product footprint is going to lead to M and A? Is not a good foot, not a good fit for us. We are looking for companies and founders and entrepreneurs that really want to go the distance, because it sort of aligns with our mission in many ways. That said, I think AI-specific themes, right? So the big axiom at the end of the day is, look, multi capabilities are going to continue to, you know, get better, and token economics, the cost of intelligence is going to continue to drop. So those are the two axioms that we absolutely, absolutely believe in. And so there’s a utility point for every task, every type of, you know, you know, skill where the where the money capabilities mix the cost, the token economics in such a way that it’s efficient for us to go down that path, and we feel like we want to play that theme out broadly speaking, right. So that’s a high-level two axioms that we absolutely believe in on the hardware side, which is what I was alluding to earlier. There’s another axiom that’s absolutely true, that the need for a second supply chain beyond the Nvidia supply chain is very real now. Nvidia could be incredibly successful, and it will be, but there is a need for an alternative supply chain, because these are huge capex investments, and in some ways you can’t have single sourcing. Right? Let’s assume Nvidia has no performance issues at all, and they continue to execute, which they can. The most important axiom is that there’s going to be an alternative supply chain, so with that in mind, if you truly understand not just inference, right, if you understand why they bought Grok, for instance, you know, if you understand the networking layer, what the implications of NB Link is in both scale up and scale out, which is their, you know, underlying protocol, if you truly understand CPO, where you know you think about, you know, co. Packaged optics near, you know, near silicon or MPO, for that matter, built for, you know, the prior generation of companies were built for a completely different type of workload, and now we’re trying to build for a workload that’s very CPU to CPU, which has been very compute-intensive. I think we can make very, very specific investments along that realm, and here is where either the differentiation of the art comes, which is to say, like, if you’re building a software company, for instance, you can get 50 customers, 20 customers, you still have revenue, these are companies that need to get the big design wins, and you have to intersect the market where you feel like that specific insertion of technology, whether it’s, you know, some specific thing about CPO or lasers, compared to when the transition from electrical to optical happens at what you know node when you think about you know model capabilities like there’s all these things that have you have to you have to get right if you miss the window the carrying costs of these companies because there’s no revenue is very very very high right so you better be sure about so many of these things and I think, for us thematically speaking, because we understand that better, we’re going to lean in further and further on the hardware, and then we are looking at thematically what I would feel like is full stack companies, and the reason I say that is because if you think about any big platform shifts, you know, I think that the best manifestation of controlling your own destiny is to control every layer of the stack. And now let’s simplistically say there’s three zero specific layers: there’s the model layer, everything that comes in between the model layer and the work product. I used to use the word application, let’s say the work product that’s delivered as a result of it. You know, so all three layers, and I think that gives you a very sort of compelling value proposition, because what it does is that if there’s a new feature or something that you need to introduce, you can make a determination where I have to tweak something in the middle, the middleware, or do I need to change the model? This doesn’t have to be your own model, you know, it could be sort of a mixture of experts, it could be a core model, it could be a bunch of different models. Then it gives you a full stack of ability, and if you’re going to go down the path, you have to be sort of vertical. You can’t do all, you can’t be all things to all people. And so I’d say the theme that we’ve leaned into quite a bit is this vertical expertise. Now you can comment it in two different directions, and I’ve seen it’s TBD as to which one is better, one where you are building on a model team right from the very get go, making sure that the full application is built from the ground up, including the work that’s getting delivered, and then you have the opposite model, which is kind of like the cursor approach, which is to say, let me own the interaction, the interface layer, I may not own the model, but once I have love for the product, and there is stickiness associated with it. Perhaps I go back and build model capability, so both can be true, but for durable companies, the full stack approach is a must have. So, you can approach it in both directions. So, if you were to ask me this question, I don’t know, a year ago, I would have probably been a lot more skeptical of companies that didn’t start with the model layer. I would have said, man, you know, I’m not so sure, because this is this notion of a thin wrapper, and that, you know, as the model capabilities, the underlying core models keep increasing, what is the value that you bring to the table, and it’s unclear, maybe my mind is a little not as sure today, which is to say, if I own that interaction, if they own the interface, and if I really have customer love and it’s really sticky, maybe I have earned the right to vertically integrate down and introduce my own models, which changes the economics of my business and
18:30
control now and destiny. What distinguishes a domain where the model must be built in house versus one where you can leverage an existing model?
18:40
So I think it has, I feel like it’s more complicated enterprise use cases, more vertically oriented, where compliance and regulations are important. I mean, Hippocratic is almost a, you know, perfect example, you know, in from that perspective, right? A mixture of experts, maybe a core model, maybe it’s open source, maybe it’s closed source, right, that sort of does the primary thinking, and they have a family of models, 2530 models that work in tandem to deal with the application. I’ll give you an example. Do you have an interaction, you know, one of the, you know, the skill or the agent that’s speaking with a, with a patient, and doing the interaction, you know, maybe the main model hallucinates for a second, says by the way, like you can take four ad wells, right, and then realizes that for this specific patient the for ad will comment was probably bad, and so you have a model that’s running simultaneously that’s prompting the main model that says, hey, that four is not right, it’s actually two, right, and then you realize within the span of like 30 seconds after call this interaction with nurse or this whoever the medical provider is goes back to the conversation as goes, hey, Nick, did I mention four adults? I just checked with the doctor, it actually has to be two now. Okay, so make sure just confirm to me that you’re only going to take two adults right now. That is amazing, but you can do that only within with sort of this governance model that sits at the. Back, and now this is just an example of a case where a mixture of experts approach is super critical, and I think in a case where you know governance, compliance, depth, ensuring that there’s no hallucination, you know, that’s a place where I’d say, you know, a model capability is a must have. So, I’d say, you know, this mixture of experts and mod capabilities must have are specific to those types of industries.
20:22
Yeah, the
20:22
Frontier Labs’ original pitch to investors was that Frontier capabilities would be proprietary long enough to build durable businesses on top of them. Open Claw has sort of collapsed that timeline to months. If you were advising Open AI or Anthropic as a board member right now on capital allocation and product strategy. What would you advise?
20:44
It’s a tough one, right? I mean, this is the think of it as the perfect harness, and you’re seeing this movement between consumer adoption leading to enterprise adoption, you know, quite the opposite of what we saw in some of the other eras, and in terms of product chops as well. But what I would say is something like this, which is, you know, SMB SAS is dead. Okay, so what OpenCloth told you is for the more simpler use cases across any dimension, you know, I think the easy use case one can do easily, and if I were to say, look at, you know, maybe AWS is a perfect example of analogy that I can draw here, you know, at the peak of their product velocity, you know, new product introduction time frame, they would introduce, they would take any open source project that’s getting traction and they would introduce the first version of something that can give you the base capability of making sure that open source technology can be adopted, and it kind of took the legs out of a lot of companies that didn’t add value on top of it, but then there was real depth in some other things, right? So, if you look at Confluent, it’s a very good example, still a very, you know, good company, you know, you know, deep expertise in Kafka, but you know, open source Kafka is available, and Amazon gave you an open source Kafka version, right, but comfort was still very, very, very good. So, I’d say understanding where you can capture value, I’d say it’s the right thing for these big model companies to do, because if you think that there is commoditization of tokens, which I think will happen, because the base token may not be differentiated, I have to increase margin, and the only way for me to increase margin is to figure out the specific places where my token is more valuable, and to go do that, I will have to go do this. So, I would say absolutely a must have a lean in. I think you have to draw the line somewhere, because Amazon has tried and shown that if you want to go deep enterprise, it’s not possible to do that, right? You know, you can take the base layer of compute and whatever they offered at that era, and then build a layer, high margin layer, that’s very, very makes the platform very sticky, but I’m going to leave a bunch of innovation for deep enterprise to still stay with companies that have that expertise. I think a very similar, you know, thing is going to happen here. So, my advice, if I were on the port, would be to lean in and figure out the layer, though, because I think you have this tendency to go too deep, too far, and then you realize that you know you’re not going to be all things to all people, especially when it comes to enterprise, but SMB, for sure, I mean, you even, yeah,
23:17
I mean, you even mentioned before, like the Netscape example, in the context of rounds and going public, yeah, but you know, Netscape was the darling, and it was part of the infrastructure with the browser, and it did not win, right? It ultimately was displaced, and it turned out application layer, whether we like that word or not, like, you know, the Googles were the winners. How does that rhyme or not rhyme with what we’re seeing today? Like, is there a future where the the LLMs that are so celebrated and have so much money being plowed into them end up losing the economics?
24:00
Look again, I have not to sort of overuse an analogy, but I would say I think people do learn from their mistakes, or to learn from history a bit, and these are not like companies that are, you know, we’re not traversing this for the first time, and so they know that in obsolescence risk like this exists because they’ve just seen it, it hasn’t been like two decades ago, it was, you know, 30 years ago, so there’s people still having experienced that in the room today, trying to ensure that doesn’t happen. So, my view is that, you know, having left that era, knowing, you know, it’s also how companies are getting built right now, like, because capital is plenty, and you’ve been given the right to go up and down the stack. So, I’d say, I’ll give you an example, and I’m kind of digressing a little bit here, but we were spending time with a company that has some absolutely insane technology spin out of Stanford around how to make, you know, more efficient to use of compute. Okay, and it is, you know, very deep in the weeds, you know. Thinking of synchronizing clocks that have skeep. Okay, so let’s not get rid of the too much of the detail, but think of it that way. If you were to build a company in the Netscape era, you had to say go build that software, right, and then go sell it to every neocloud and every operator, every enterprise, and that’s your mission, and that’s all you’re going to do, right? And this sort of insane focus, which is allow for you to build a maybe a very profitable smaller business, but if you were to say, “Look, I think this technology is very, very interesting, but go all the way up in the layer of the stack and offer this as a solution, and maybe you build your own inference cloud, because you have efficient user compute that nobody else can do today, I’m giving you the right to go build that business right, and so to me, I think the big takeaway is that you take away potential areas of disruption because you’re being given almost a hall pass from investors who are willing to fund you to go create all of the layers of defensibility around you, right. So to go back to the Netscape moment, I would say, if Netscape were to say I’m going to do search, right, and I have given me given the right to do search, and I’m going to be the right to do more applications, you know, make the true web server, perhaps they would have been one of the biggest companies on the planet, but they seeded that because from a product perspective they were fighting the browser war, specifically from this platform, and Microsoft came in, obviously, you know, took sort of the, you know, the cut them off for the knees in many ways, because that was the distribution that they did own, right? In Kuya, you’d have said, “Why don’t you make Netscape the operating system? You would have given, given the right to go build an operating system to compete against Microsoft, you would have been, you know, people allow you to do that these days, right. And so I think that’s the, that’s how it sort of draw, draw a parallel to the prior era, is because I think companies are more full stack these days, at least the winning companies
26:48
is a part of this like a focus issue, you know, if you had spoken to Mark Andreessen way back in the day, you know, about, well, you should do a search engine as well, and that should be the homepage for Netscape, and, oh, by the way, you should build the operating system, it was probably hard enough just battling the browser wars and trying to win, right? And I mean, hindsight is always 2020 but we could predict that in the future the LLMs are battling out for, you know, the agentic layer and you know everything else, and we may be missing what’s next, but it might be too hard for them to take on too much, even with the war chest they have in the talent.
27:30
Yeah, I mean, that’s a very astute point. I think that both is both, I mean, look, I think it’s, you don’t know how things are going to play out perfectly, but I can definitely see that, and they feel like we’re still very early here, right? So we have to sort of see what happens. I’d say we’re still in maybe the first half of the first inning in many ways, and so as you think about diffusion, you know this specific technology across a lot of big industries. You know today we are underwriting to, you know, these two companies doing a lot of all of that, and I’m not sure that’s going to be the case. Look, we’re still investing with the exact notion that he just said, which is to say that at some point we know that, you know, there’s probably going to be places where we can be quite disruptive and still own a very defensible piece of, you know, whatever value add we bring to the end customer. So I do agree that there’s the need for focus. The the the only thing I will say is that because in the past the cycles were more, were longer, and the reason that the cycles were longer, not maybe even from an adoption perspective, because the time it took to build a product was high, and that I think would create some of these management priority issues, the, you know, try fast, fail fast, the cost of new things have dropped so dramatically. I can test a thesis very quickly, because I can get a product manager who has a vision to go build a specific product, testicle set of customers. These are like the these are the reasons why I feel like if you are building a certain moat and you have some network effect and you have scale and size, because the cost of producing something new is so low. I can go try a bunch of these things, which kind of puts the dynamic of yes, there needs to be focus at the same time, and not giving up in many ways of the opportunity that needs to be had. At some layer of focus is a must have. So the point I was making earlier, which is to say, please draw a line right of the of the near places that we’re going to attack, which is to say prosumer, and sort of the, you know, the low end of SMB. Yeah, absolutely, go do that, but don’t go after like him, but of it go very, very deep and build something out in, you know, specific vertical, like whether it’s, you know, biology or defense, or like, you know, pick any, any, any sort of big sector materials, you know, that I think will be a fool’s errand in my mind, that I think that would create like massive distractions.
29:48
So, you’re unique, you’re a crossover investor, you’ve invested in venture and privates for many years, you’ve invested in the publics, and I’d love to take some time to talk. Out, what we’re looking forward to in the coming 12 to 24 months with some of these potential mega IPOs, right? And in particular, maybe second order effects that people aren’t anticipating. A lot of people, there’s a lot of excitement. Oh, liquidity frees up in the ecosystem, and you know, retail investors get access to SpaceX, etc. but there are other effects that are going to happen when some of these multi trillion dollar companies, you know, suck up a lot of that retail money in the public, so in terms of capital flows, talent movement, investor attention, you know, How should smaller businesses position themselves, and give us a sense for what is on your mind and what is most present for you as we move into kind of a new dynamic of IPOs.
31:00
Maybe I’ll answer the question in two steps. Next, so one is just talk about the IPO itself, and then, and then the second part is, you know, what impact it may have on the private side, right? And so the IPO itself, in many ways, you know, public markets, a lot of the funds that eventually bind to these IPOs, they don’t really sit on the sidelines with a lot of cash. Mutual funds have a requirement to be in the market all the time, by definition of who they are. I mean, there are a few private funds that can keep some cash, but you get penalized for holding cash, so everybody’s deployed in some ways, right? And so, historically, if you think about the size of the IPOs that we’re talking about, let’s just pick the three SpaceX anthropic Open AI, and the expected rates. I do not know any better here in terms of what they’re looking for, but just ballpark, based on what I’ve read in the press, it speaks for roughly in my mind 80 to 24 months of potential IPO volume in any given year, in any given era. Okay, of when, when I, when IPOs are a flavor, right, of demonic Apple. These two cup, these two, as these three companies may end up raising now. Quote, unquote, there is a feeling that it opens up the IPO market, but I don’t feel that that is the case. In fact, it could have the opposite effect, which is to say, if I was a portfolio manager, and I’m going to say I know SpaceX is coming, or I know OpenAI is coming in, no anthropics coming. Will I be open to buying another IPO that comes through the door? Absolutely not. I have to keep that capital, you know, reserved for these situations, and I may have to even sell my existing positions to get an allocation in these things. So, even if I think about where would I sell, right, then you know people think, hey, software stocks, look at the valuation. I don’t know, maybe I would have to sell a bunch of my software positions still, you know, to buy my appearance in Spasik and Open AI, so it has an impact on, you know, the notion of, you know, whether these things open up the markets. It may have the opposite impact, which sucks up so much of the dollars available that it makes it really difficult, you know, for other companies to go public, so that’s, you know, point one. The second point I’d say is because, you know, if you think about, you know, purely from the from the perspective of, you know, how people view these businesses, right? So, like, you know, in these cases, largely owned by still a price discovery of not many to one, but maybe a few to one, right. And so, how do you think about token economics? You know, is it a gross profit business? There’s been a rule of thumb, historically, at least in the public markets, for a long, long period of time. Is that if you are, say, a 65% ish gross margin business, plus, right, then you can be considered a product company. Anything south of that is considered maybe services plus plus those companies always get measured on a gross profit multiple as opposed to a revenue multiple, at least in the early years of growth. So, I don’t know what the exact economics of how this business is going to get reported, but it’ll be an interesting dilemma to see how public topic markets forget about the hype for a few minutes, we’ll start viewing, you know, how to value these businesses now. That leads me to the second part, kind of directly today. When I talk to entrepreneurs, they go, “Well, if anthropic is value valued at, you know, three 80 billion, why can’t I be valued at 3 billion? Right? You have this conversation at all times with various entrepreneurs. Now, they’re not wrong to think that. And then you have investors say that founder is killer, you know, they won this other thing that is great. These are all absolutely fantastic qualitative terms. Public markets don’t give a crap about these things. Okay, at the end of the day, and so translating, you know, private market hype on like many individuals, things that, by the way, I heard Anthropic is doing this. I heard, you know, Open AI has done this. You know, there’s all these things.
34:46
Sandesh, don’t you feel like there’s a corresponding hype with retail investors out there? I mean,
34:53
there is
34:54
everyone uninitiated that I know is asking me, how can I get access to Anthropic? Nick, can you get me you. Yeah,
35:02
premium is priced in, but you know, like, okay, so let me first of all, I’ll come back to this point. I just read, yes, my only point would be that I feel like, yeah, as a result of these companies going public, I feel like it may bring more rational behavior on the private markets, because now I can say, hey, what are your token economies? What is your growth rate? Public markets view these things this way, and so therefore I can better value your business today, and I think some of this absolute, like qualitative hype that I was alluding to disappears. Okay, so that’s one point of view. The second I’d say is look, unlike in prior eras, where companies that created a significant amount of market cap didn’t raise the same level of capital, they’re relatively capital efficient. Look at Facebook, look at Google, they turned cash deposit very early in their life, and they never turned back, right? So, these were increasingly cash, which also meant that they, not a lot of money went in, that’s not looking for a return, this era, unlike other eras, even if these companies go public, there have been so many people who have been investing billions and billions of dollars that all need to seek a return, so the overhang, you know, so I would say, as a retail investor, there’s no reason to run and buy, you know, you got to sort of let these set of investors, actually, there’s going to be many, many times of dislocation, because many people who build billions of billions of dollars, they all need to see to return, they all have to sell at some point, even if they love these businesses, purely because they’re fiducia responsibly for the funds that they sort of manage, which means that there’s going to be more dislocations, and this is going to get overhyped right at the beginning, valuation fundamentals are going to be out the door, so there’s no reason to chase, right? And so, if these are actually building the enduring businesses that they are, you’ll have plenty of time to buy to the public markets openly, and so patience is the name of the game in some ways. So, you know, you never know how these things play out, and for a while I might look like I was absolutely wrong, because OpenAI in public is up 100% for two years, which could very well happen, and he goes, I got a man, I missed out on all these things, but you know, at some point you got to be sort of fundamentals and sort of not feel like you have this massive FOMO, and you’ve got to chase this, and you got to have it all your own. We
37:15
are humans,
37:17
I’m encouraged of your portfolio, free money, I
37:21
am encouraged at the prospect of pricing normalization, though the public’s like introducing some discipline around pricing, because, yeah, we just had Steve Blank on, I don’t know, a week and a half ago, and we were talking about the reality distortion field of founders, but I feel like we have one with VCs right now, where if you know you’re a 10 million ARR company, and you’re growing 200% per year, but you’re considered more SaaS with AI, you know, enabled by AI, then you’re valued at some, you know, low cram down kind of multiple right now, whereas if the growth rate is much higher, but you’re only doing one or 2 million of ARR in, you’re considered AI native, like all the rules are completely out the window on valuation and pricing. So, it will be nice to see these economics kind of find their, their, find their center. I
38:18
didn’t
38:19
think so. I think it’s out of time, like I also feel like, you know, think of it this way, right? Like, there’s lots and lots of VCs and investors out there, they’ve made certain promises to their LP, one of which, unsaid, is that I’m going to be an every winner because of the size of capital I’ve raised, and perhaps the way I positioned the phone, and so what ends up happening a lot of times is that you can’t afford to miss, and so therefore you are in everything you know, or you have a play in every category, or every subcategory, or every sub, whatever, and you know, I, we haven’t talked about our fund more specifically, you know, we are, you know, we run very concentrated portfolios, it’s one balance sheet back end, we do have fun cycles, just like Google does with its balance sheet investing. Every time,
39:04
how many holdings?
39:06
So, you know, our funds, you know, the way we operate is that we agree with the endowment, you know, it’s sort of an annual deployment pace, like a better word. So, in the case of US growth, we say between 300 to $500 million a year, and early, we also have an early fund between 50 to 100 so we try to deploy between 400 to $600 million a year. So every three years we have a fund, roughly 1.2 ish billion dollars in growth, and about three to $400 million in early. So that’s been how we’ve operated this fund for the last 12 years, right? But I could be half the size in the same time period. If I went back in the end of three years, I want to deploy at 500 million, so be it. I just move into the next fund cycle in the calendar year, right? I don’t have an issue, which means that I don’t have to chase, I don’t have to be in every one of these things. I’m looking at enduring businesses, and if I can’t make a decision because we have full stack, I’ll come back to it at a. At a point where underwriting is a lot easier, right? I don’t have to feel like I gotta, you know, click on every button that you know shows up. The second bit is, I see this with the team all the time, you know, we don’t want to do relative investing, because if I have a very large team, they can say, okay, you got to do, you know, storage, you got to do infrastructure, you do foundation models, you do security, and then maybe they’re bad neighborhoods, and you’re buying me or giving me the option to invest in the best home in a bad neighborhood. I don’t have to be in every neighborhood, so I tell my team, look, we don’t need to be in those neighborhoods, I don’t need to buy the best home in a bad neighborhood, we can just be in the best homes in the best neighborhoods, you know, to the extent that you know BDI is the beholder, but you know, to the extent that we can make our decision on what those look like, let’s just do those right. So, I think to your point about valuation, I think there’s a lot of capital chasing, and so there’s a reason for us to be invested in every category. The amount of capital chasing few good great entrepreneurs results in absolutely bunker economics, and so, so that’s one element that’s creating, you know, these, you know, high valuation rounds without really any fundamentals supporting them. And the second is, you know, and this is a very AI-specific response, like I would say something like this, which is to say, you know, not all, you know, people have heard about German’s paradox, and they go, well, Germans paradox, you know, token cost kids going up, mod give his, and these, there’s going to be explosion tokens, but I don’t know, let’s ask ourselves some questions, and I don’t know the answer, the question, I’m just posing the question for your audience, and for you and I to discuss this, but not all tokens are the same, okay, so let’s, let’s say I have in unlimited healthcare tokens, now I can see unlimited or very, very numerous healthcare use cases today, because there’s a, there’s a big sort of gap between what value and care it can provide than what is available today, doctors, nursing, whatever you want to call it. Let us pick law, okay? There’s a lot of hype around law, and let me tell you that I have unlimited law tokens, unlimited law tokens, okay? What does that mean? Are we going to have more litigation? Are we going to have more people doing more law? I mean, there is a cap in terms of how many law tokens get consumed by human beings at some level, right? So, this – I mean, not to say that there’s going to be an explosion of like more token usage around law, but I’m saying you can’t compare a healthcare token and a law token the same way, because the size of those two markets are very different, and the margin profile of those two markets are very different,
42:27
but aren’t the tokens fundamentally fairly commoditized? Like, they’re,
42:31
I mean, the base layer is, but if you think about some of these application companies that are building on top of those base layers, potentially, quote unquote, adding their value, you’d assume that there is some premium associated with those tokens with that manifestation, right. So I would say the other part that I think investors are not fully thinking through is this notion that, you know, are all tokens going to have the same type of return profile in economics, you know, specific to these specific industries, whether it’s law or whether it’s, you know, healthcare and things like that. So I think both these elements create some distortion, and pricing is a little bit off, and for me, I just feel like if I can’t make a good decision, not a great decision, a good decision with the data I have at the time, I, you know, I can wait.
43:13
It’s a good luxury to have Sandesh. I’d love to talk about private equity, just a bit, and private credit, you know, before we wrap, so both are kind of in a tough spot at the moment. The SaaS pocalypse has collapsed software multiples. We’ve seen one of Carlyle’s private credit funds just got hit with a 15.7% redemption request, more than triple the limit, and of course, Blue Owl and BlackRock, everyone’s been talking about those the past few weeks, so are we looking at a significant contraction in private credit, and what do you think that means for PE and VC,
43:53
PE, more specifically, you know, VC, probably less so, right? For sure, I mean, the first telltale, I don’t know if you saw this, there was an article that went around about Medallia owned by private equity being handed back to the creditors, we’re equity, oh
44:06
yeah, it just, yes, yeah, just saw that this morning,
44:10
that’s Canary in the coal mines, beginning, right? I think, here’s the, here’s the staggering math, so let’s just hypothetically think about a company that got acquired for 10 billion, right, and this is sort of any peak, for instance, right, and LTV was maybe 65 maybe 60% So, let’s just say 60% so $6 billion of debt and $4 billion of equity. The $6 billion of debt with a five or seven year term, and because you know people thought that these cash flows were significantly durable, they were considered AAA credit and AAA credit got in the most favorable rating you could possibly get, right, which means private equity could really lever up. Now, four or five years have gone by. What has private equity done? They have essentially hollowed out R and D. They have, in some ways, maybe invested a bit in part, perhaps a tad bit. Right, business profile is probably growing 10 to 15% which is what I’ve seen in a lot of these companies, maybe 10% And then you have 30 to 40% maybe 35% free cash flow margin. Now you’ve paid back some of the debt, you’ve taken six to five. Now, what would you pay for a business that’s maybe say doing a billion dollars in revenue with 35% fake actual margin, and going 10% today, because I don’t know what terminal value looks like anymore, right? And so, if you use public market comps as a proxy for that, I may value this business at about four or 5 billion. Now, if I have to refinance, right, I have to go back to market, and I’ll go, wait, I got $5 billion and refinancing, maybe there is refinancing. Let’s just assume that private credit is there to do refinancing. If they’re there, they’re going to say, “Fine, business is not great as what you did. I may be able to underwrite a 50% LTV. I can be a two and a half billion dollar loan, and you have to put in an equity cure, another two and a half billion dollars into this business.
45:55
Whew,
45:56
now the math is staggering, and this is what’s going to affect private equity. Now they can talk their book, and they can say how AI is beneficiary to them. They can do these deals with Open AI, but if I was a really great AI engineer, would I go to a private equity-backed company? Absolutely not. So, there is no very big discrepancy here. So, I feel like private equity and private credit is just sitting on its one, is they
46:20
can unravel slowly, or is it all of a sudden we’re going to see a big, I
46:25
think,
46:25
there is
46:26
a bunch of paper that’s coming due for some of the excesses that will happen in 2020 2019 so that’s like really reaching term, right? So 2026 2027 but you know, look, you know, it’s it’s like the age-old theory, if you want to launch the bank, you have, you know, ability to dictate the terms of the bank, so private credit, they don’t want to call on this thing, because they may have a delegate in their hands, which means that they don’t know how to operate these businesses,
46:51
they don’t want the keys,
46:52
they want the keys, so what are they going to do, they have to grin and figure out terms, and you know, potentially extend, and my only point is, it’s going to further deprecate these businesses, not the other way around, because the ability to invest is really hampered, and they don’t know what they don’t know, and attracting the right talent and building these companies is incredibly hard. So, I think the private equity side of the house, especially for things that the bond enterprise software, is going to be very challenged in our mind, and
47:19
for them,
47:19
you think the private credit, the creditors have more leverage than the PE guys, or vice versa.
47:26
Today, I think that you know you’re on top of the stack, so therefore you definitely have more leverage, and you have a paper that
47:31
they also don’t want the keys, and I don’t know,
47:34
yeah, but you know they can, can you do get paid on it, right? So you can go back and you can say, well, I’m gonna increase your, you know, basis points by this much stuff, I want to steal a $500 equity infusion, a little bit of a cure, you know. You can do a bunch of these things, and because equity, others that can get wiped out, credit can get still get paid for a long period of time, as long as these cash will still exist, right. So I think that the moment they probably have a little bit more leverage than the private equity guys, but private credit, you know, that’s a different dynamic, a bit, because now you’ve raised retail capital and retail capital is not well informed, and then retail capital says, well, I want to redeem, and so that that issue can be very, very painful. So, look, I don’t know whether this happens overnight, but you’re going to see, I’m sure a few more funds here or there that will have some redemption problems. Yeah,
48:25
it’s, it’s, it’s an interesting time, because
48:29
a few
48:29
years ago, poster, I remember meeting with my, my credit fund buddies and my PE fund buddies, and they were pretty smug, pretty, pretty happy. Yeah, and it’s changed a lot, there’s a lot of panic and worry and narrative spin for our LPs and all that.
48:48
Yeah, no, we’re seeing some very nice, wonderful decks hit the market recently.
48:55
Sandesh, if we could feature anyone here on the show, who do you think we should interview, and what topic would you like to hear them speak about,
49:01
oh gosh, I mean, the podcast world is littered with doing everything, I can, like, you know, nothing jumps out, like, in my mind, but I like to go back to historical figures, lack of a better word, and I’d say, you know, for me, Carl Sagan is somebody that I follow and read and understand and love his thinking, and how he thinks about stuff, but gosh, if we can go back in time and bring him back to the current era, I’d love to see what he thinks about what’s happening, you know, here and now.
49:30
Give us a book, article, or a video that you would recommend to listeners.
49:37
Let’s see, you know, I think that you know, the I forget the name of the title now, the Darling Kruger effect is kind of the one that I always like to sort of go back to, and this actually alludes to some of the decisions I made earlier, right? In the less you know, the feel like you are more convicted on these things, right? And to me, I give this book to every one of my, you know, you know, investment. Fashions in the firm, and the reason I say that is simply that you know sometimes you know this overthinking of things can be, you know, negatively impactful, and there’s a fine line between what you know and what you don’t know, and how you try to make a good decision as a result of it, and it’s a, you know, it’s something that I like to follow as much as I can.
50:18
Sandesh, do you have any habits or behaviors that are a secret weapon,
50:24
I’d say. Listening, I’d say listening is not listening, listening. I mean, I’m just saying you got to truly, truly listen, and sometimes it takes quieting of the mind, because I think your impulse responses, as people are saying something is to start forming your question or calcifying a potential response in agreement or in disagreement way too early, and I find myself many times going back and saying I just, this is not the time to have an opinion when I’m hearing something out, and I think it allows me to sort of be a little bit more patient, and in some films, only make a decision when you have to make a decision in some ways, because you know you have more knowledge over time. I go through what I call a sinusoidal wave of excitement and disillusionment across an idea, a person, people, and I think you know listening and being a little bit patient before you have to make a decision allows you, you sort of smooth out that effect, and I’d say, if there is a bar, it’s like listening, and really intently listening without passing judgment. It’s incredibly hard, trust me, I’ve tried
51:29
it, and
51:30
I’m listening to a pitch, and I’m going, “I don’t think I’m going to do this. I’m listening, I think it’s awesome, I got to do this, and then everything sort of, you know, gels from that spot, you know. So you got to like truly step back and listen. Sometimes
51:42
it’s good. And then finally, here, Sandesh, what’s the best way for listeners to connect with you and follow along with Brenji?
51:49
You know, just on email, I mean, I think our website and our blogs are.. we’ve been very under the radar, and we finally have a website. Some people joke that we discovered the internet, you know, two years ago
52:00
through heaven, definitely on email, yeah. Well, sir, I appreciate it. Thank you so much for the time. I wish we had more to go deeper on some of these topics, but this was great education for me, and I’m sure the listeners are going to love it. So, thank you,
52:13
appreciate it, Nick. Thanks for having me.
52:20
All right, that’ll wrap up today’s interview. If you enjoyed the episode or a previous one, let the guest know about it. Share your thoughts on social, or shoot them an email. Let them know what particularly resonated with you. I can’t tell you how much I appreciate that some of the smartest folks in venture are willing to take the time and share their insights with us. If you feel the same, a compliment goes a long way. Okay, that’s a wrap for today. Until next time, remember to over prepare, choose carefully, and invest confidently. Thanks so much for listening.
