Village Global Podcast - What It Actually Takes to Get a Venture Capitalist to Say Yes _ Max Kilberg _Village Global_ _ Michael Hochberg _Periplo

0:00Max Kilbergbut I like to remind them that VCs are stupid. VCs are scared and VCs are reductive. I think investors don't really like to invest in things they don't understand. If you can confuse an investor or you scare an investor, that's a very easy way to get a quick no. We're seeing things that five years ago I would have said violated the laws of physics. If you had said, oh yeah, we're gonna see startup companies that are doubling their revenue.

0:30Max Kilbergevery two months for 18 months. I'd have said no way. Can't be done. Hello. I'm Max Koburg, one of the partners on the team at Village Global. And today on our podcast, we'll be speaking with our network investor, Mike Hochberg, on the state of the venture industry. Max, great to see you. It's a pleasure to finally record an interaction between the two of us. We've had so many conversations over the last couple of years. Absolutely. So you've settled here in San Francisco. Yes. You're getting plugged into the the San Francisco VCs and Geist. What are you seeing that's going on that's not obvious from the outside? Well, I would say a lot of what's going on is very obvious from the outside, since VCs are spending a lot of time broadcasting everything that they see on social media to everybody.

1:25Max KilbergProbably the most obvious thing is just the bifurcation that's happening in Venture right now amongst the mega funds versus everyone else. And so Thrive and GC and Andreessen and now Sequoia and Benchmark are raising larger and larger. Quantums of capital, they're sitting on those fees with the goal of...

1:46Max Kilbergpushing a billion dollars plus into a couple of breakout winners. While their early-stage portfolios are really seen as ways to track those companies and gain access to those mega opportunities to push money in.

2:01Max KilbergAnd then I think at the earlier stage, which is the stage that we operate in at Village and the job that I do, it's becoming way more of or just as artisanal as it was. It's really about individual investors, their individual networks, their taste in picking founders and ideas before they become relevant, and ultimately, hopefully, front-running a lot of the...

2:26Max Kilbergbubble is probably the negative term, but the quantum of capital that's being poured into later stage opportunities. I would say that amongst traditional venture investing, which I would categorize as pre-CEED, CEED, Series A and B, you can think of beyond words as growth investing. A non-obvious trend that I think I've been seeing has been that Everything has really compressed into the idea of seed investing. Even at the B, Series B investors are doing traditional seed investing. Not much is being learned at the B, as opposed to the A, as opposed to the seed. Put another way, if Anthropic can add $5 billion of revenue a quarter, does it really matter if you're...

3:12Max KilbergSeries A company is doing zero million in revenue five million ten million twenty million It ultimately comes down to the team the vision and the belief that you can graduate into a very very fast-moving Hyperscaling company. So that's that's super interesting. There's like five things done pack. Yeah, what one thing historically was that the way you manage a fund that needs to deploy billion dollar chunks, you know

3:41Michael Hochbergdoing one on 10 and trying to scale from a $10 billion valuation to $100 billion valuation and then beyond. The people you hire, the way you evaluate companies, every single thing about the culture of the firm is completely different from the way that you would manage a firm that does one to $10 million checks.

4:09Michael HochbergHow are these big firms bridging that gap? Because if you're doing something where you're deploying a billion dollars on 2 and 20, and then you're deploying 100 million dollar checks on 2 and 20, the person who's deploying the billion dollar checks speaks a little louder in the meetings than the people who deploy the smaller checks.

4:33Max KilbergI mean, I think that's why there's been such an exodus of young people out of these multi-stage firms. Most of them are going into operating. That's where a lot of the alpha is early in folks' careers. But it's certainly correct that if you're at a multi-stage fund, a lot of the attention goes to the five deals a year that matter to that multi-stage fund and the people doing them. And most of those deals, if not all, those deals are growth deals. And so the people running those funds.

5:02Max Kilbergcertainly are the investors writing the biggest checks. It can be hard to really do anything, gain the political alignment needed to get a deal done. If it's just too small to matter, part of what I really like about being at a seed stage fund is being a seed stage investor and getting to do the work that is important to the organization. It's the common saying of you don't want to be doing.

5:28Max Kilbergcomputer science or software engineering at JP Morgan, you don't want to be doing finance strategic banking at Google. You want to be in the part of the organization that is the money making part of the organization. One of the peculiarities of this is, as an entrepreneur, when you formulate an idea, oftentimes you want to formulate something that is as low capital intensity as possible.

5:58Michael HochbergYou want something where, in many cases, the ideal thing is something where you don't have to raise an immense amount of money in order for it to be very successful because you can choose to bootstrap it. Because that gives you a lot of independence and a lot of freedom of action. It gives you, in effect, strategic autonomy. One of the things I noticed, I guess this is maybe 10, 15 years ago, is that a lot of venture firms were really only interested in deals that could absorb huge amounts of capital at multiple stages because they basically saw much of their alpha as being, okay, I'm gonna come into seed stage or A round for millions or tens of millions, but I need to be able to deploy giant chunks of capital later at low risk and still capture.

6:57Max Kilberga lot of upside. Sure. Is that becoming a dominant narrative now? I think it is the dominant narrative of our time, especially when the hottest companies have become so capital intensive that they can actually take on these massive raises.

7:12Max KilbergIt makes our job as seed investors slightly difficult sometimes. In our fund strategy, when we model out our fund returns, we make approximately 50% of our fund returns on our follow-on decision-making, right? Participating in future rounds of our breakout companies. But now, when we're writing...

7:34Max Kilbergwe write $1 to $5 million checks, 10% ownership target. When we're writing that $1 at $10 million investment in the inception stage round of a company, we're seeing a lot of multi-stages come in because they've scaled AUM with their equivalent of 1 at 10, which to them is 10 on 100. And then we need to make a very quick decision on whether this company is actually breaking out or whether a multi-stage fund is just taking a seed option check on that company because they to be able to put in a billion dollars later when that company, if it really takes off. Interesting. Part of what's interesting from an entrepreneurship perspective is that there used to be lots of opportunities where VCs would get very excited about something where the expectation, if it was successful, was you'd come in

8:35Michael Hochbergfor a few million dollars on say a 20 million dollar valuation something like that and The upside of it was a few billion dollar exit, right? So you're talking about a few hundred X Right, it seems like a lot of people nowadays are focused on the very small number of places that are going to exit in the $100 billion plus range. The 100x that they're looking for that pays off the fund and does the 5x on the fund is going from

9:19Max KilbergEvaluation in the hundreds of millions to evaluation in the tens to hundreds of billions Is that yeah, is that is that really because I think there's so few ideas that can do that I think it's a product of these funds raising larger and larger and larger Quantums of capital that they then need to you know promise to return and so it becomes a game of Deca corn plus hunting for these multi-stage funds That said, there's no one way to make money and venture. For seed investors like Village, we can look at our portfolio and say, hey, $110 billion outcome returns the fund and more for us. That said, we can also make our money on low unicorn outcomes as well. And so there's a couple of conflicting tensions here, I think one of which is

10:14Max KilbergEveryone wants to make money in venture in the way that or believes that the right way to make money in venture is sort of just the first way that they ended up making money in venture, right? So a lot of folks who were very successful at maybe doing incubations early on in their venture career, you know just focus on that a lot of folks who are You know great growth investors want to just write the biggest checks in the room and command all the attention But as these fun sizes are getting bigger and bigger and bigger as they are The outcomes need to get bigger and bigger as well. I think the good news for entrepreneurs is it's really never been easier to raise a $10 to $20 million seed round and really have a right to win a nascent market, to be able to hire whoever you want, to be able to scale an engineering team, go to market team, pitch the best customers in the industry. I think the real winner of a lot of this is actually the founders.

11:13Max KilbergOne of the things that's been really interesting to me is that the shapes of the motes have been changing. I'm a no motes exist believer, actually. Oh, that's interesting. Why do you say that? Because AI makes it so easy to build so quickly that every new week it seems that a new model is better than the last that a new product is launched. Ultimately, I think the moat becomes how fast you can move.

11:43Max KilbergIt becomes the people on your team, which I think is actually a fair mode. It becomes about the talent you can aggregate and the momentum that you can achieve with that talent in terms of go-to-market output. You don't think there's scale modes or network effect modes or frequency modes or brand modes. You don't think any of that exists anymore?

12:04Max KilbergI think that there are certain moats around brand, for sure. And I think that there are certain moats around scale as it disincentivizes other people from trying to do the same thing you were doing. I think entrepreneurs really like new markets and being the first mover because you have a higher likelihood to succeed if you look at the historical data. But that said, I think traditional moats around customer lock-in or data.

12:31Max Kilbergflywheel effects are really, really decreasing over time. I think network effects still certainly exist. Obviously, consumer social platforms haven't been the hottest thing to fund recently, right? And that's, I think, where they exist most. What about training modes? As you get more people onto your AI platform to do whatever, you learn to be better at whatever it is that you're doing. And I think of that as a flywheel mode.

13:00Michael HochbergRight? The more people are interacting with your AI system, the more data you get on what is good versus what is bad, and your product starts to improve over time. Do you see those modes going away? Really depends upon the domain. I think that brings us into something else that we were going to talk about around.

13:24Max Kilberginteresting ideas, and one of them is sort of LLM determinism issues, right? And so if that domain is verifiable and easily replicable, coding, I think, is the best example of that. Yes, your data mode is very, very powerful. That said, if your task is slightly differs between industry or business to business and customer to end customer. I think that mode is less than it was. That said, I do think the best mode is moving fast. What about formal IP? Are you seeing people care less and less about things like patents? I recently saw, I think it was Palmer Lucky speak about this on one of his podcasts about the importance of patents. I think that in

14:13Max KilbergEarly stage at least in the market that that we've been investing in it doesn't come up a whole lot I think as VCs we really underwrite the likelihood that there will be fast followers and there will be copycats of any sort of proprietary system that that you're building I think for later stage investors it probably differs in the calculus. I mean the way I look at patents is I think they're only valuable for people who are very sophisticated about how they write them because

14:43Michael HochbergYou have to be thinking about, are you going to be able to detect that someone is violating it? You have to be thinking about, how much are you actually enabling your competitors? Because, of course, enablement is a test for patents. If you're not teaching a competitor to do the thing, then you're going to have a lot of trouble enforcing the patent. And then you also have to think really hard about, where am I going to enforce this?

15:12Michael HochbergAnd how much is it going to cost? Because patent enforcement is extremely expensive. And so there are plenty of places in the world that matter where you just can't do it, starting with China. All propaganda to the contrary. And so my sense of it is what I always tell entrepreneurs is, yeah, you want to file some patents if there's something that you think meets all the criteria. Yeah, file a couple of patents.

15:40Michael HochbergBut don't spend a lot of time and money on it because the reality is you're not going to be able to use those patents for anything other than performative purposes for at least five years. You're not going to get into patent litigation until you're very successful. And then it's useful, potentially maybe. But it's mostly not a great use of time except in very specific edge cases.

16:09Max KilbergBut that's a very good point. One of the most valuable pieces of advice that you could give to an entrepreneur earlier in their journey is just around focusing on the things that matter and are important. We all have 24 hours in the day. Yeah. My old CEO used to say that we can do anything, but we can't do everything. And so we have to pick the very specific one or two things that we're going to do that give us a right to win. And everything else falls by the wayside. Absolutely. And I think that's another very essential point to early stage company building. That's zero to one, one to 10 stage, if you call it, which is playing games that you actually do have a right to win.

16:55Max KilbergAnd I think there's nothing more than we love as investors when we meet a founder in a pitch meeting. And they tell us that there's really only one to two things that they're going to use our money to go accomplish. But then the unlock that accomplishing those things brings about in terms of the quantum of capital or the customer demand or the momentum that they're building in terms of their go-to-market is almost

17:25Michael HochbergMind blowing sometimes. The typical question that people love to ask VCs is, what do you think is hot? What are you interested in? I'm not going to ask that question because it's just not a very interesting question. The question that I do want to ask is, one of the things that entrepreneurs want out of a fundraising process is they want determinism. Because they want to be able to get on with building their company.

17:54Michael HochbergAnd I'm curious if you have a founder who is optimizing for determinism, both in time and in effort and in money raised. Defined determinism in a fundraising context. They want the fundraise to be successful. As everyone does, yeah. They want to have a systematic process that they run.

18:20Michael Hochbergin order to make sure that their fund raise is most likely to be successful on a timeline. What would you tell them to do? Well, I actually spend a lot of time with...

18:32Max Kilbergportfolio company founders that we've backed, helping them think about that next round. At Village, we spend a lot of time with our founders at the pre-seed stage, at the seed stage, and then our goal is to really stay close to them, but help them effectively leave the nest and fly, basically. I think there's a couple things that I find myself saying to founders over and over and over again. The first of which is that As a founder, you're really raising money from an individual person, more than you're raising money from a firm. Back when I was a founder before, I became an investor. I don't know if I fully grasped what the day-to-day, week-to-week, month-to-month, year-to-year job of an investor, a venture investor, really is, which is that you are making a very few amount of bets a year, and so they really matter.

19:27Max KilbergAnd I would encourage founders to put themselves in investor's shoes. If you could only make three investments a year, how would you spend your time? Would you be investing in the founder who you don't know who sent you a cold email? No, you probably would be spending time with folks who you think are incredibly intelligent ahead of their fundraise. You would be thinking very deeply over whether founders and ideas are exceptional, whether they have validity, but also momentum. And as we talked about, DecaCorn plus potential. And so I find that when founders really understand that they're raising money from

20:10Max Kilbergfrom a individual person who only has a couple of shots on goal a year, they really up-level their game. Because you ultimately need to be the best meeting of someone's month, of someone's quarter, of someone's year. And I didn't realize this when I was a founder, just how many founders are out there pitching the same VCs as you.

20:33Max KilbergThe second thing is that, when I work with founders on their storytelling and things like that, I like to remind them three things. And it sounds a little crude, but I like to remind them that VCs are stupid, VCs are scared, and VCs are reductive. And what I mean by that is not in the literal sense, but most investors are dumber or stupider than on the subject of the company than the founders that are pitching them. I think founders really need to meet investors where they are. I think investors don't really like to invest in things they don't understand. And so if you can confuse an investor or you scare an investor, that's a very easy way to get a quick no. And then I think the third realization is that investors are reductive creatures. We're always trying to map

21:19Max Kilbergwhatever new information or pattern match, whatever new information we're getting to something else that is known or is understood. And so educating founders about how to properly allow investors to be productive in the way that...

21:34Max Kilberghelps them sell their vision most effectively is really, really powerful. You almost want to think when you're meeting a new investor about what is the one to two sentences that you want that investor to leave the conversation with and take back to their IC or investment committee to be able to effectively sell your vision to the rest of the partnership. One of the things that I've seen many first-time entrepreneurs get wrong is that they'll have a meeting. It'll be a great meeting.

22:04Michael Hochberggood conversation, interesting discussion, friendly VC, the VC says, oh, this is exciting, this is interesting, but no term sheet emerges. And one of the things that a lot of young entrepreneurs don't appreciate is that the way that VCs say yes is they issue a term sheet.

22:28Michael HochbergAnything short of that is... A no or a maybe. It's a no or maybe in the future or its relationship preservation because they might want to do something in the future and they don't necessarily want to give a bunch of negative feedback to someone who they've just met. But it's a no. The shades of no are important to be able to understand because a lot of young entrepreneurs, they'll have And I've had this experience with many people. So, 20 meetings with VCs that are all very positive in tone and very friendly. And then they'll be scratching their head, why didn't I get a term sheet? Like, everyone's so positive. Where's my term sheet? Do you see people running into the same sort of dynamic? Yeah, all the time. And I think the cruel reality is that

23:23Max KilbergThey are usually not the number one most important deal that the investor they pitched is working on. They were not the best meeting of the year, right? And if you can only do so many deals a year you're going to you know drive to the most important ones and put all your energy into them regardless of how the levels of good that the other ones were. And so that's where I think entrepreneurs have sometimes a misunderstanding of the process. I think it also depends for founders on what type of round you're trying to raise. I think that at the early stages of company building, there are individual investors that founders really want to work with.

24:08Max KilbergWhen entrepreneurs come to me and tell me who those people are, I really advise them. I say, you need to start building those relationships fairly early. You need to be a known entity if there are only a handful of people who you could really see yourself working with. If you're trying to raise a very large round to go after a very ambitious vision, I often advise founders to speak to a ton of people, to start to build momentum, to actually go get a term sheet that you don't want to take.

24:35Max Kilbergso that you walk in when you have the meetings with the investors that you want to take their money with an Exploding offer to provide a little bit of time pressure and competitiveness you don't think that doing that scares people away because like if you're having if you have a term sheet from someone and You're taking first meetings with everyone else You know you I've had the experience that some of the other VCs who you're taking those first meetings with They'll shrug and they'll say, we're not going to move on that timeline. It takes us... It's a risk that you run. I think the art of actually structuring a fund raise is to be in process at the same time with a lot of different funds, both that you are very excited about as a founder and maybe some that you're less excited about, and then ultimately be able to push momentum forward on your round through one person gaining conviction and then the domino falling and another person gaining conviction.

25:31Max KilbergAnd when you see these massive rounds on Twitter, LinkedIn, startups raising $100 million Series A, you're like, how did that happen? It's usually because they went out to raise $40 to $50 and had two to three people fall in love with them. Individual investors at these individual firms fell in love with these founders and bid each other up. Yeah, 100% agreed. I mean, if you want to do a round quickly,

26:00Michael HochbergOne of the dynamics that I've seen again and again is entrepreneurs will fall into this trap of, oh, I talked to this one VC and they were friendly and maybe they're going to preempt and I'm not going to have to go do the full activity. And there's no time pressure. And so because there's no time pressure and there's no bidding war, they can drag their feet for three months, six months, ask for more information, ask for more information.

26:29Max KilbergUnless, as an entrepreneur, unless you can set up a situation where everyone's coming to the table at once, nobody feels any pressure. Well, I think that's the ideal situation, especially at Series A. I think Series A right now is sort of the bloodbath of the venture ecosystem for founders. It's really the haves and the have-nots.

26:51Max KilbergA lot of exciting, mostly infrastructure companies raising as much money as they want. Everyone else, I think, is struggling a little bit. And whether you agree on whether that should be the case, it sort of is at this point. When do you mean when you say infrastructure? I would say that anything involving chips, inference, neoclouds, developer tooling that's built off of those, chips and neoclouds.

27:20Max KilbergAnything below the application layer, I think, is very exciting to a majority of VC land these days. And everyone's a little bit nervous about horizontal application layer platforms, especially newer ones, given how far-reaching the ambitions of the foundation model apps are. It's funny because I have a peculiar view of this.

27:51Michael HochbergJust because you can build a piece of software yourself Doesn't mean it's a good idea and I look at a lot of software that businesses use and It's always been the case that you could build it yourself You know if you're if you wanted to build your own CRM system You could hire a team and build your own CRM system and it would probably be cheaper than what you were paying Oracle and you could do it And it would probably be better because using Oracle is a nightmare, right? Unless you're Fortune 100, you probably don't need 99% of what's in Oracle. And it's just implementation burden. Something simple and straightforward is better than going and using Salesforce, for instance. Because implementing that stuff in maintaining it is just a nightmare. But yet,

28:50Michael HochbergPeople tend to be conservative and to use the proven solution because they don't want to get fired You know, they don't want to rule something themselves. And so even if the costs to Implement something comes to zero, right? We get two more generations of foundation model and you know, I can sit down with fathom and and or I'm sorry with fable and and and generate multi-million line code base for you know tens of thousands of dollars of tokens and it'll all be right and it'll be well-structured and it'll be maintainable and it'll be you know a thing of art and beauty yeah right okay great I can do that but if I build some custom thing someone has to maintain it someone has to pay attention to it it could be a security problem someone has to check it right and

29:47Michael HochbergEven if all of that is highly automated, some human has to take responsibility for making the decision to do it in-house, as opposed to just paying for something that someone is selling you, where there's a company behind it, where someone is making sure that it's right. I'm not sure even if the feature modes fall, even if you can build whatever feature you want relatively quickly.

30:15Max KilbergI think we're going to end up in a situation where the branding boats really matter because people will still not want to get fired for building something defective themselves. Well, I think that incentive alignment question is something that is pretty poignant and still needs to be figured out. There's certainly going to be a lot less humans in...

30:37Max Kilbergyou know, AI native, agentic organizations. But the humans that remain will still have lots of different moral hazard considerations to make. But there is a future of the world, you know, where you point Fable at Salesforce, at your CRM system of record. It copies the entire thing once. Boom, you don't need Salesforce anymore.

31:00Max Kilbergquery and interact with your CRM or your Salesforce with a prompt directly in Slack or on your phone with that agent. And so the question is like, do we really need the dashboards anymore? Once it's working and once you know what you want it to do, sure. And I could see that. But the big challenge is specifying what you want, right? Being able to say, OK.

31:31Michael HochbergI mean, once you have a system that's up and running and you're reasonably happy with it and you have improvements you want saying, I want a copy of this that's local, that's a compact and straightforward thing to say, I want a copy of this thing that I will then own. And if that thing that you're copying doesn't have a lot of magic behind it, right, if it's just a database or something, yeah, of course. If it's something that's got a sophisticated trained model behind it, that's hard to replicate and that requires a lot of data? Maybe not, right? But coming up with the specification of being able to say, here's what I want when you don't already have it, that's going to be a real challenge. Absolutely. And the other thing is that I do think that in a lot of these cases, the fear that commercial SaaS or

32:31Michael Hochbergcommercial AI as a service that that's going to disappear and that everything's going to get sucked into the foundation models. I think that you're going to have lots of companies where they have their own models team. Of course. And where that's the mode. I think it's actually going to segment into two different kinds of problems. One is going to be things where there is a clear, good enough threshold.

33:01Michael Hochbergwhere there is some version of perfection, say where there's a mathematical limit on how good something can be or where, let's say you're doing speech to text as an example, where at some point it's just perfect. It's good enough. Where the only way to make it better is to add more functionality, more interaction, more smarts, whatever. And you see this with a lot of traditional things in hardware, where it's like, you run into a point where you just have a trade space. You can make the transistor smaller, but you give something up. You're running into the limits of the physics. I think you see this a lot of the time in physical systems. In cases where you get to the point where there's a first place that has just solved the problem.

33:58Michael Hochbergwith a specialized model. At that point, the competition is just, okay, the problem is now solved. How do we make it cheaper or faster? But there's not a quality competition. That's something where it's going to commoditize. You can make good money in a commodity market. You have to be better, faster, cheaper. Faster, cheaper. Better.

34:23Michael Hochbergis the other side. There's going to be other things where it's a competitive market where, okay, the place that has the AI that is generating higher quality output wins because everyone goes for the higher quality. For those, it's the sort of infinite spending model where money will pour into whatever is going to give you a better result.

34:53Max Kilbergbecause the whole market sloshes over to the model that's best. And when you ask why AI infrastructure is so hot amongst VCs, it's really because venture capital is an asset class set up to pour as much money as possible into the fastest growing, highest upside opportunities. And right now, it's that. When people build hardware, if you're building a new chip, Basically, you have to call the ball on what chip the world is going to need two to three years from now. Do you think that we're still going to be in hyper growth mode for, say, token demand three, four years from now? I think so. I think the CapEx build out is happening three, four years from now already. Those contracts have already been purchased.

35:55Max KilbergI think what happened with Kimmy this week and the Gevan's paradox around GPUs being at capacity, even when the market thought they weren't, was particularly surprising for some and obvious for many in the meantime. I think a lot of it depends upon the scaling laws of LLMs. Everyone got really excited when LLMs got really good at coding, that these scaling laws were non-existent. They were just going to scale upwards infinitely.

36:24Max KilbergI think now you see folks start to realize that a lot of the value within enterprises is actually the harness, right? You need the proprietary data. You need someone to build the evals, to structure that data, to build the evals, to do the RL, to build your custom model, to do that specific workflow. You have folks realizing that a lot of white collar.

36:45Max Kilbergjobs are very different from software engineering, where a code base is a machine-readable document for excellence at the job of software engineering. There's no machine-readable document for being a great lawyer, investment banker, venture capitalist, things like that. And so a lot of the instructions for how to do these GDP-producing jobs very well in the digital world, let alone let's leave the physical world aside, because robots coming is also a huge title wave, too.

37:15Max KilbergThe expertise really exists within the humans that currently do those jobs really well. So those insights need to be extracted and, again, structured. And it's not enough to just read their email. And it's not enough to read their email because the internet's already been trained on and these things still aren't at ASI. Now you have also some other folks like...

37:36Max KilbergI guess, you know, Jerry Torak's new Neelab Corrado, who are now saying that there needs to be a different architectural substrate to solve self-learning, right? And no one really knows what that is yet, and whether that'll ramp as quickly as LLMs have. But I think it's become more commonplace to acknowledge that the scaling laws do exist. And maybe it doesn't get anything but marginally better than this.

38:05Max KilbergNo one knew that pre-training transformers was a transformative idea. That's not what OpenAI started with the idea of. It was some intern, I'm forgetting his name, had the first idea to do that. Then all of a sudden, chat GPT popped up, or GPT-3 popped up. It's funny because we're in this regime where there really isn't any unifying theory for any of this. It's all what I think of as pre-science.

38:32Michael Hochbergyou go back to the days before we had chemistry. And people were doing things that were empiricist enterprises. But it wasn't really science the way we think of it now. It was alchemy. It was zoology. It was, all right, I see a critter. I'm going to draw the critter. And then we're all going to get together at the Royal Society and compare our drawings of critters and try and figure out how to make sense of all of this. Same thing with alchemy. Before you had a theoretical framework for chemistry, you just tried stuff. That's kind of where we are in a lot of AI, in the sense that the interesting ideas are not theoretically driven. It's just people trying stuff. It's very smart people trying stuff.

39:24Max KilbergThe pace of progress is astonishing in terms of real economically impactful results So what that says to me is we haven't saturated the easy problems to solve I completely agree and I think the marginal utility of the 98 to 99th percent effectiveness at achieving a problem will continue to make it will continue to financially incentivize people to go after that last mile delivery. I think a lot of what we're talking about in AI right now is really about that last mile delivery. These labs are trained on however many million trillion quadrillion parameters they'll be trained on, yet they're still not effective in that post-AGI promise of doing human work most effectively.

40:24Michael HochbergOne of the things that people forget is that many of the biggest results that we got out of, say, chemistry were before we understood the periodic table, like gunpowder predated modern chemistry, the discovery of gunpowder, the discovery of fire. These were the big, big results, and we've done some good things more recently.

40:52Max KilbergJust being purely empirical and trying stuff often gets you your biggest results. You're a student of history and as an entrepreneur have maybe seen more cycles, not to date you, than I have. I'm curious if this current moment in the AI zeitgeist mirrors to you anything about the dot-com bubble.

41:16Michael HochbergAdvancements and chips even earlier than that it doesn't it doesn't I mean I used to joke when I was an undergrad about all of my so at the time I was doing a chip company and and I used to make fun of some of my friends who were going into These e-commerce things that that were just losing money hand over fist and and the joke I used to make was You're gonna do what you're gonna sell cat food over the internet like Choose a more noble quest. Yeah, I mean, you just did a degree in some fancy science thing from Caltech. And you're going to lose money selling cat food over the internet. Who wants to buy cat food over the internet? Even if you succeed, it's not important. And I was dead wrong. I was dead wrong. Because in reality, once the infrastructure got built out to sell cat food over the internet, now,

42:13Michael HochbergIf you're buying groceries, many people get them delivered. And it's a meaningful thing for a lot of people because it makes their lives easier. I bought dog food over the internet. I never had a cat. But it makes sense. So it was an idea that, at its core, actually made sense. But the infrastructure to do it just didn't exist for several years later. I mean, all of the things that were promised during the internet bubble.

42:43Michael HochbergWell, a lot of the telecom overbuild out, helped cloud adoption a couple of decades later. Yeah, in marginal ways. The big thing is, all of the things that people wanted to do over the internet, they did show up. They just showed up 10 years later. And so, that was a particular form of bubble where people were just getting ahead of the economic impact. This is different.

43:14Michael Hochbergbecause the economic impact is so immediate for a lot of this stuff. It's also easily understood. Yes and no. I mean, I interact with a lot of people who, in effect, are still using Claude as a search engine. Really? Yeah, all the time, including in the tech industry. From the perch from inside Silicon Valley, that view where people understand what you're talking about when you say, I need to build a harness for this. I need to build a skill for this, whatever it is. Most people don't know what you're talking about. And even in the wider tech world, there are a small number of very AI forward organizations. And you see the impact of this, right? In startups that are very AI forward, tiny teams doing amazing things.

44:14Michael HochbergRight, you know teams that are accomplishing things that used to require ten times the people in the same amount of time and much less money It's amazing right, but the number of places in sort of Fortune 500 corporate America where they've managed to figure out how to use these tools to anything like the same effect is very small indeed and That says that there's a huge opportunity here. Because the impact is clear, but we haven't even hit the inflection point where it starts to really have an impact out in the wider economy. I think we're just at the beginning of this. We're seeing things that five years ago, I would have said violated the laws of physics. If you had said, oh yeah, we're going to see startup companies that are

45:15Michael Hochbergdoubling their revenue every two months for 18 months. Sincerely, right? Without gimmicks, without weird circular payments, without playing games, like with real honest-to-god external revenue. I'd have said no way. Can't be done, right? Here we are. Yeah, and here we are. And it's being done, I wouldn't say routinely, but by dozens of places. And that is a secular change. And people now sort of take that for granted. And it gets gamed and all the rest. But it used to be that the only places that could grow their revenue that fast were like, OK, I've invented a viral video game. And maybe it lasts for 12 months as it gets bigger and bigger. I've invented Angry Birds.

46:12Michael HochbergThings where the interface was really easy, but no one was doing that for productivity software, right? Because productivity software, if it was powerful, was hard to learn how to use. So I guess I'm, I know the venture community, people are very skeptical of the application layer these days because they're concerned about software modes and all the rest. I'm much more optimistic about that because I think that you're going to get software that has superpowers, in effect. People will consume more of it. The Jevons Paradox, where the marginal change in demand for small changes in price is huge, you're going to see that across a lot more industries. The embodied robotics thing that you mentioned, the embodied AI thing that you mentioned a moment ago, is an area where

47:08Max KilbergI think almost everyone is underestimating the speed with which that's going to accelerate. Because right now, I don't think the limitation is the hardware primarily. I think the limitation is primarily just the training data. It is. And again, back to non-determinism issues, there's really no way for a robot to know if you pick up your mug this way, or this way, or this way, or this way. And so you need hundreds of thousands of hours

47:38Michael Hochbergof humans picking up that mug. One of the questions I have about all of this, I've written a little bit about this, is, okay, so you have an embodied robot that's pretty smart, right? And that you interact with, like you would interact with a human. And there's a problem, and it kills someone. Or it damages a piece of property. Who owns the liability for that? It is right now a completely unknown question.

48:07Michael HochbergI suspect that from a legal perspective, it is not unknown at all. It's just that the known is dysfunctional, in the sense that whoever owns it is responsible. And then there's a question of intent, and there's a question of who generated the model? Is it the model or the hardware that went wrong? And the liability flows to whoever screwed up.

48:35Michael Hochbergin effect, and it depends on legal, on who wrote what and demodities into which agreements. But it's not clear to me that we have a regulatory environment that's going to work for that. If you have a fully self-driving car that causes a traffic accident, do you know how the liability associated with that has played out?

49:04Michael HochbergI'm not the most knowledgeable on this. I'm not either. It's just really interesting to me because if you have a Tesla... You should call your buddy from Arnold and Porter. We should have a call-in on the pod. I'm sure they could tell me, but it's an interesting question. If you're in self-driving mode and your car runs someone over and you're not driving, are you liable? Is that a criminal act on your part? I would presume the driver is still liable if they're sitting in the driver's seat.

49:35Michael HochbergLet me ask an associated question. I get in a Waymo. I'm not sitting in the driver's seat. The Waymo runs someone over. That's got to be Waymo. Well, I'm taking a nap. Am I liable? Is Waymo liable? What happens if they get hacked and the Waymo runs someone over because they didn't have adequate computer security? I have no idea how this plays out, but that legal morass

50:06Max Kilbergis going to be a big adoption. Well, I think what you're hitting at is that we still have a really long way to go in putting robots inside homes, putting them around children, having them do vague, ambiguous tasks, both in an industrial setting, in a residential setting. But there's a lot of money to be made in that last mile of deployment. See, I actually don't think we have a long way to go. I think the Europeans have a long way to go because the first time that they have an accident,

50:35Michael HochbergThere's going to be a blizzard of regulations out of Brussels that say, you can't do this, you can't do that, you can't do the other thing. In the United States, we have a very different attitude about this, which is we don't really stop things until they turn into a disaster. And so what's going to happen here is this is just going to get deployed, and then the legal framework will follow based on whatever weird exceptional things happen.

51:04Michael HochbergAnd then in places like China, you're going to have a different model where the government just decides that this is going to be deployed. Kind of like they have a cap on the number of people who can be killed in a natural disaster. There's never a natural disaster with more than 50 fatalities in China. We promise. It's going to be the same kind of thing. Everything's going to go really well because the government decided that it will only be reported that it went really well. So I think this is going to get deployed ahead of all of this getting figured out at scale really fast. And then we'll just sort of make the best of it. The superpower of the American economy is figuring it out. Yeah. I mean, one space that I'm watching really closely is the insurance space. Really? Yeah. Because when you have things that front-run the law,

52:00Michael Hochbergthe way it gets worked out is by the insurance industry. The spaces that are not regulated or where there is no law like the maritime side or where there's law, but only sort of. If you want to know what the future looks like, you pay attention to what the insurance company is requiring people to do. Because if I'm a company and I'm deploying embodied robotics into people's homes, I'm going to try and get an insurance policy from Lloyd's on that, where fundamentally, if something goes badly wrong, they're covering it. And then they're going to come back and say, okay, well, what are you doing to make sure that something doesn't go badly wrong? And I think that's where a lot of the action is going to happen. Before there's legislation, before there's litigation, it's going to happen at the insurance layer first.

53:00Max KilbergWell, I mean, we're seeing that already happen, not in the physical world, in the digital world with agents. A lot of neo-insurance companies are starting to offer insurance around agentic products.

53:13Max KilbergAnd a lot of the carriers are struggling to think about underwriting it. That's interesting. Tell me what you mean by that. That's not something I've run into much. Well, how are you going to ensure if you are a organization that has a customer support facing chatbot, and maybe you're dealing with sensitive data, bank data, transaction data, things like that, what if something goes wrong? Who is ensuring that that agent, that LLM, is not doing anything poor with that customer's proprietary. Well, when you say ensuring, do you mean creating technical guardrails? Or do you mean literally legal indemnification? Legal indemnification. Interesting. That's going to be a whole new category of insurance that needs to be figured out. And from the couple contacts at the carriers, they have no idea how to underwrite it.

54:04Michael HochbergThat's super interesting. And I think that'll probably happen first before the physical world insurance gets handled. Yeah. What it says is that all of the things that we need as humans in terms of legal and contractual infrastructure, you're going to need at the agent declare as well, which is super interesting.

54:29Max KilbergA lot of what we're trying to do, I guess implicitly, and maybe explicitly in some ways too, is get these things to think and reason and behave like the human brain. And like human communities as well. Like human communities, which is maybe an unrealistic expectation to put on them. So are these the kinds of things that you guys are looking at specifically at Village? Is part of your alpha that you're looking into these particular kinds of activities right now? I would say we're a generalist firm. We can invest across industries, geographies, things like that.

55:09Max KilbergWe work with a network of luminary LPs who are founder CEOs of the previous generation of transformative, magnificent companies, folks who are more sector expertise like yourself. We have geographic expertise tracking different ecosystems for us. Then I would say the individual investors on our team, of course, have our own little swim lanes of what they like and dislike and understand and don't understand and things like that. I could speak to myself.

55:38Max KilbergPersonally, I've never had a real job. I went from early stage founder to early stage VC. So I would say my knowledge of the intricacies around end markets and buying patterns is maybe less than some of my other colleagues. But what I do really understand is deeply technical founder, very good technical idea, big idea, stick them together, and let's go run at it. So I'd say I'm naturally drawn to maybe things Beneath the application layer at the info layer, but no we invest across the gamut. Do you see stuff coming out of Europe that's investable these days? Yeah quite a bit. Yeah, and Do you find that the overhead slows them down or are you? Equally excited about deploying capital in Europe compared to the US. I would say that

56:32Max KilbergTraditionally, the fastest growing companies have been in the Bay Area. That said, there's a ton of AI and ML talent coming out of European universities. A lot of these founders are exceptionally exceptional, incredibly entrepreneurial and driven, and they want to start companies. Most often, they actually form...

56:53Max Kilbergtheir companies in the United States. They form a Delaware C Corp. That is then investable very easily by U.S. venture firms. Whether they choose to move full-time to the United States or operate their companies in the EU is a variable. But no, I think a ton of companies are coming out of Europe. Look at what has happened in Sweden recently. That's interesting. From an investability perspective, forming it as a Delaware C Corp. from the get-go. You don't care as much where they're physically operating, as long as it's a U.S. entity. I think it makes it very easy on our legal team and our conscious, of course, as American investors investing in, physically located in the Bay Area. That's at Right Village Global. Global is in the name. I think we found that a lot of the exciting companies in our most recent fund are...

57:51Michael Hochbergactually either European founders or operating out of Europe? European founders, I see all the time. People who move to the U.S., very common, very, very common. Do you actually have a bunch of portfolio companies where, independent of where the legal entity might be, they're physically operating out of Europe? Absolutely. That's interesting. You're bearish on anyone operating in Europe?

58:22Michael HochbergI guess my starting point is if you're not prepared to move to the Bay Area or New York or somewhere where there's a concentration of talent and capital, my starting point is how serious are you? Is this really your priority? There is a concentration of AI talent in Europe. There is, but it's not in any one place.

58:51Michael HochbergIt's scattered all over, and there's certainly not a concentration of capital. I guess when I talk to founders in Europe, whether it's hardware or AI or any of a number of other things, they're invariably struggling to raise capital. I see that again and again in a way that I don't see here in the US as often. What I always talk to them about is, okay, What's holding you here? And sometimes there's a very good answer, because there are clusters in Europe for specific things in specific geographies. There's a semiconductor cluster around IMEK, for instance. There's a few others. There's an aerospace cluster.

59:42Michael HochbergFrance I'm forgetting the name of the town, but like there's a lasers cluster outside Paris right there there are places where there are geographic clusters with a high concentration at least of talent and equipment and capability if not of capital so You know I don't ask the question in an insincere way I ask what's holding you here because what I see again and again is You'll have a European company that's in, you know, there's one that I'm thinking of right now that I've been, I've chatted with the founders since before they existed where immense EU money poured into it, immense activity. You know, they've raised a big round and they've realized that they have to move to the United States because they just can't do what they're trying to do in Northern Italy. It's a semiconductor thing.

1:00:38Michael Hochbergyou know, the depth of talent pool is fantastic, but the regulatory environment's a pain. It's impossible to fire people. The government money is non-dilutive, but it dilutes your attention, right? It slows you down. And they can't raise the kind of money they need to raise there. And maybe that'll change over time. But yeah, I mean, I always have a seriousness question when it's like...

1:01:08Max KilbergWhen are you moving to the US? If you're really on a hyper-gross trajectory? Because they always run into a ceiling. I think you're probably right in the broad sense. That said, the magical thing about our industry is for every broad point, there's a very specific counterpoint to disprove the rules. There's always the counter examples.

1:01:28Max KilbergAnd I think what we come back to is I have a certain amount of bullets in my gun every year to use in terms of investing in exceptional founders. And I think that the founders that I work with are truly exceptional. They can really build companies anywhere. This is a thing a lot of people don't understand. What stops you from writing 10 checks a year instead of three? Well, we have a broad strategy with...

1:01:56Max Kilbergmany investors on our team. So not everyone can write 10 checks a year. I think that when I think about my own allocation in our fund and deploying that effectively, there's the constant struggle of, do I write fewer big checks or more smaller checks? Because each investor on our team has allocated a quantum of capital. So it's capital? Yes.

1:02:19Max KilbergThat's interesting. It's not attention. It's capital. It's not attention. It's capital. At least that's the way that we structure. I think that it's because we try and make every dollar really, really count. I would say investors pass for a multitude of reasons. Ultimately, I think when I'm meeting a founder, I'm trying to diligence investment.

1:02:43Max KilbergI'm sort of thinking on the lines of, you know, is this person deeply exceptional? What is the evidence of exceptionalism in their past? How does that transfer into what they're building now? Why does that give them a right to win compared to every other super genius or, you know, moderate genius with the same idea? Ideas have sort of become commoditized in early stage venture land these days. Why does that give them a right to win? Like, is the relationship that they'll be building with their customer an everlasting relationship where the customer's derive a lot of actual value versus perceived value, and ultimately how quickly can this get to scale in a reasonable time, since we're investing on fun timelines as well. And so I think for me personally, I've really tried to spend a lot of time with people at the frontier who are maybe not founders, who are in industry, and really get a sense of maybe what is six to 12 months pre-consensus.

1:03:39Max KilbergAnd I think that that's really where our right to win is as early-stage investors. I was playing sort of Deal of the Week, my first couple months in venture, where I was chasing sort of the hot thing and getting beaten by other firms and things like that. And then I think I really needed to internalize and realize that...

1:04:00Max KilbergAs a junior investor, no one knows who you are. You don't really have a right to win. Obviously, the brand of Village helps a lot. But ultimately, you need to play games so you can win. So the games that I can win are with founders that I know very well or that I believe in more than anyone else. Cool. Let's end there. Let's end there. That's fantastic. Thank you so much. Thank you.

1:04:24Ben CasnochaHey, this is Ben Keznoka, co-founder of Village Global. Thanks so much for tuning in to the Village Global podcast, where we go deep on all of the biggest topics in tech. If you enjoyed this conversation, please subscribe to our YouTube channel. You can check us out on Spotify, Apple, wherever you get your podcasts. We'd love to see you for the next one.