This Week in Startups - VCs Would Bet on Open-Source AI Over OpenAI and Anthropic | E2341

Starting point is 00:00:00 All right, everybody, welcome back this week in startup slash this week in VC. We are every week trying to do it this week in VC panel. This week in startups is brought to you by Rippling. Don't settle for AI that's all talk. Head to rippling.AI slash twist and get the only AI built to give you full visibility across your startup and take complex actions across your entire organization. That's RIPPLING.AI slash twist. Sign up for exclusive access today. Odu, the all-in-one business platform.

Starting point is 00:00:27 Your first app is free. Get started today at ODOO.com slash twist. Northwest Registered Agent. Got a new business idea? Northwest Registered Agent helps you bring it to life. Get a free domain, email, phone number, and more. With no purchase required, learn more at www. Northwest Registeredagent.com slash Twist domain.

Starting point is 00:00:45 And my lord, we've got an incredible panel here. I'll call this old school. Jeff Clavier is here. My old friend, founding partner, Uncork Capital. Jeff, how are you doing? I am good. Thanks for having me. How you doing? It's been a while. Gosh, it has been a minute. You and I have known each other for three decades, I think, since you're at 2000s. You started in venture capital. When? In 2004, I mean, 2000 as a traditional VC, 2004, I kind of started the whole seed craziness. Yes, you were there early. I think it was you. Let's see. First round capital came after you in that same zone. sort of started around the same time with Josh. They were raising vintage funds every year. I raised one of the first seed funds in history in 2007 after being like what they called at the time

Starting point is 00:01:41 super angel for three years. Yes. So there was Super Angel, which meant you did more than a couple of checks a year and you might have had LPs. Then we had the, I think we called it microfunds or microvc early. Some people did. For a few months, they called us Super Angel. funds and then Mike Maples came up with the term seed. I think he kind of invented it. And starting, I guess, 2007 or 2008, they called us like seed funds. What size was the first fund? Well, the first fund was $15 million. And you made fun of me on the TechCrunch 40 stage because we announced it together on

Starting point is 00:02:22 September 12th of 2007. Check out the elephant here. I think I'm the elephant in the room He's got receipts, baby. Better watch it. And you were like, who the hell sort of raises a $15 million fund makes no sense,

Starting point is 00:02:37 but you luck to you. I mean, it's not exactly what you said. You were an innovator. What were the sizes of the rounds that seed back then? When we invested, you know, $250K in a million dollar round, it was actually a pretty standard round.

Starting point is 00:02:53 Fitbit in 2008 raised the $2 million seat round. It was massive, massive. And that was at what kind of valuation? Like an 8 million post, a 7 million convertible note? It was 5 Prie. Five Prith. There you go. Right in that zone. Jenny Fieldling is here. Also, a bit in the business for the while. When did you make your first investment? It's a good question. I was a founder and I did a little angel investing. And then I was working on my third company and I ran into the founders of tech stars, David Cohen and Bradfeld. And they said, what do you think about being on the other side of things?

Starting point is 00:03:29 So I thought, oh, maybe I could like incubate my company, learn a little bit on the other side. And I ended up really loving early stage investing and stayed there for seven and a half years running tech stars New York. Yes. And this was in the time period where each of those tech stars had like really amazing leadership. And they ran like in this affiliated network. I think sometime during COVID, they consolidated, put a CEO in charge. It didn't work out. They may have broken up that federation and it got a little dicey.

Starting point is 00:04:05 But tell us about the golden age there of tech stars. I started like 2013-14. And actually, this is so funny. I have my little mug from my accelerator. I had two unicorns in this class. Oh, my. Wait, it has the logos of each one? Yeah, it has the logos of the 10 companies I invested in.

Starting point is 00:04:22 And I still use it. And this was from, I think, like a 2014 or 15 batch. But it was the Golden Age. I mean, Dave could tell you as well. There just weren't a lot of folks doing, I guess, pre-seed. So kind of before the deaf claviers got involved. And, yeah, that was incredible. We did Sedgrid together.

Starting point is 00:04:42 Oh, Segred. That was your big one, too, Dave. That one were very early on. One of them. Yeah. And Spot Hero, which actually just got acquired recently was also Tech Mars. Oh, only 17 years later. You had spot here for

Starting point is 00:04:55 I mean spot here I I haven't heard It a long long time That revenue went down to pretty close to zero during COVID That came back Yeah well I mean it is an interesting point though About the time right So this class was I think a 2014

Starting point is 00:05:10 2015 class and here we are You know and you know Those unicorns have not exited right So here we are 11 years Topic of the day Well this is going to be two I'm going to make two notes here Number one, to my team, I want mugs.

Starting point is 00:05:25 What? Copy Jenny's idea. I want these mugs. This is a splurge. This was literally a splurge. I was like, okay, can we find the budget to make mugs? $600 in mugs. I want to do this for the last three classes because I have one class, hit two unicorns as well recently, which will be a topic here.

Starting point is 00:05:42 But really interesting when you look at that mug of, oh, let me introduce Dave as well, and founder of 500 startups now called 500, Glover. mobile, long-term friend of my... Now the founder of Practical Venture Capital because it isn't. Yeah. And now doing that, yes, as he self-introduced himself, managing a partner, practical venture capital. Also has a great podcast, and I am one of the...

Starting point is 00:06:08 Thank you. What's the name of the pod again? Jason, when are you going to be on our pod, baby? Have you sent an invite? I mean, I invite you to my... I've been to send an invite now. You're done with All-in? You got some time in your hands?

Starting point is 00:06:19 I'm done? Did all-in break up? Oh, thank you. God, it's over. That is, you try being in a rock band. You've been busy. Oh, my. Last week was a crazy week for me, yes.

Starting point is 00:06:36 But we have a full docket here today. I don't want to hijack the show. But Dave, I assume you got to see some of the All In Summit drama. Did you have anything that stuck out to you from the other than time? Did you brought it up? I don't think there's massive drama. But like, you know, the call-in for, you. from Trump? I got to ask you, that seemed staged. That was not dynamic, was it?

Starting point is 00:06:57 Okay. Here's, I would, I'm always a straight shooter here. And let me just start my, hold on, before I do that, I might give the answer. I'm going to start my Plaud pin. If you don't use Plod, they're a sponsor of the program. And you put this little pendant on. It's got a red light. Everybody knows you're recording. Boom. You can wear it on your lapel. You can put it on your wrist. Records everything. And then you are all set. You have your entire, you have your entire day recorded if you want to or just a meeting like I'm doing right now. And I'm addicted to this. We now do our show notes. So pull up the show notes, gentlemen. We take the entire podcast. We put it in there and they have templates, 20 different templates that we use, but there's hundreds of other ones. And you can do a

Starting point is 00:07:34 summary. You can do all of the top insights, the funny moments, the hot takes from this very episode will be on the show notes. So check the show notes in your favorite podcast player. You found a link to applaud.plod.a.a.com slash twist and use the code twist for 10% off. I always applaud. One of my favorite devices. So to answer your question, Dave, briefly, I did not know he was going to do that. I found out subsequently that he asked his assistant to hold his phone and that he had just texted Trump. He was going on stage with Sacks. And that's the-

Starting point is 00:08:06 Jensen. Jensen, then his assistant comes running out. And I think he's putting on a bit. I thought it was a bit. So I said to tell President Trump, J-Cowell says hi. And he's like, oh, Mr. President. And I'm like, okay, so it's totally a bit. He wouldn't do that. And I'm like, put him on speaker if it's really President Trump. And he goes, how do I do that? And I just picked up. I always have a spare microphone in case there's some problem. I said, here. That was genius. And then it actually was president, as a president, as a Trump. He did have J.D. Vance scheduled. And Elon was scheduled, even though I pretended he called in. But, you know, he's, it's a, you know, independent of what you think about Trump, you know, as an individual, as a president, politician, whatever. He is. He is. He's a, you know, you know, he is. super engaged with the tech industry, which is, you know, a good thing, I think. Or better or for worse,

Starting point is 00:08:55 my friend. Yes, you could, that's a one way to say it. I think it's great that he's P-Doom zero and all you need is a high-acute president. Dave, where's your, let's go a role in one here with your P-Dume. Which is your P-Dume, your prediction for doom. Are you 10% chance of the world? No, I'm everybody, a seven billion people dying. Yeah, 90% full. We actually talked about this a lot on our podcast. So I think what's going to happen is there's going to be a AI fund, sovereign wealth fund that'll get cut next year sometime the next six to 12 months. Trump wants piece of the action. Democrats want regulatory oversight. Sam and Dario want product liability shield. So, hey, 10%, 20%. I don't know what the number is, but... So you're saying there's a negotiation slash shakedown about to happen.

Starting point is 00:09:44 This is a Dave McClure prediction. Not about to happen. It's been happening. Bernie wants 50%. Sam offered five to 10%. Like we're just trying to figure out with the negotiation. All right. So they give 10% and then they get a section 230 like law that says, hey, if your large language frontier model hallucinates, you can't sue. That's what your prediction is, Dave. Interesting prediction. I like it. Regular listeners already know that if you've got a great idea for a new business, our friends at Northwest registered agent want to help you bring it to life. They're going to be the most amazing partner. ever had, even if you're not ready to form an LLC, you still need to take care of some basics.

Starting point is 00:10:25 So Northwest registered agent is now offering free identity services. That means a free domain name, open source website hosting, a business email, and a phone number and everything else that's going to make your new startup look and feel like a professional company, all with no purchase required. And you know you can rely on Northwest because they've been helping people like you start businesses for nearly 30 years. If you have an idea, you can't get out of your head. Or even if you're already building something amazing, you already got started.

Starting point is 00:10:53 Northwest Registered Agent is the best way to establish your new company. Learn more at Northwest Registeredagent.com slash twist domain. I think you can follow sort of the cigarette companies structure as a very similar model. I think the number would be like between 10 to 20%. Like if you just did, and I actually would think an equity swap would be better than a tax or revenue deal. Okay. Jeff,

Starting point is 00:11:21 what's your thoughts here on, I guess, one, you know, independent of what you think of Trump, just having an engaged president who's this engaged? And then is your P-Doom high here? You've been in this business. You've been looking at machine learning, AI, expert.

Starting point is 00:11:36 What do we call them expert systems? Was that the original term we used in the 70s and 80s? Super intelligence is the new term now. So now we're down to, okay. I will see it. I was calling it Trump intelligence. Trump intelligence is the new term. Trump intelligence, I've just rebranded it.

Starting point is 00:11:51 But Jeff, as an elder statesman here, expert systems. I actually learned about neural networks back in 1987, believe it or not. So, P of Doom is none zero, unfortunately. I don't know if it's like 5%, 10%, more, whatever. There's a sort of drama, there's sort of people trying to, to make themselves important by claiming all those things. At the end of the day, the way to sort of make AI safer is to get the CEOs and their teams accountable for any of their mistakes.

Starting point is 00:12:31 And as Dave said, they're going to try and figure out a product or personal or company sort of liability shield. Whatever they do, they will sort of get it because at the end of the day, they don't spend enough time to figure out how the AI can be safer. And based on how they actually build the AI now, it's less and less possible. And so the issue I see personally for the world is like it's not that they want for this thing to go, you know, so sideways. But because of the way they build them, because the AI built itself, 26% of Anthropic is building, you know, the next generation of Anthropic as of yesterday. It's actually just not clear how they can actually protect us.

Starting point is 00:13:19 And what is your thesis? And Jenny, I'll go to you next for your P-Doom, but you said your P-Dium is not zero. I'm going to say it's probably low single digits or low teens, whatever. You can correct me if you are. I'm an optimistic, so yeah, I'm going to try and hope that this is as low as possible. So- What do you predict could happen? What concerns you if you were going to get more specific? Oh, I think it's basically the AI just going wrong.

Starting point is 00:13:43 like there was an example reported recently where an agent or butt or whatever misidentified the content carried by a Chinese ship and the Pentagon was about to either intercept it or bomb it or whatever, which obviously wouldn't have sort of gone really well if it happened. And thank God, you know, the human analyst sort of stopped it. We don't know whether it was one second before, one minute before or one hour before. But that's the kind of thing where we know AIs just tend to just make up things, right? It happens all the time. And while it's applied to critical infrastructure or military applications, the fact that it

Starting point is 00:14:30 hallucinates may just happen, it's powerful course. We just use it in instances where the effect of an hallucination could be dramatic. So this is on the human using it and human brains, Jenny, this is the consistent. and I have, humans are having their brains atrophy because they trust this technology far too much. And so they have it write their essay like Druck and Miller did recently, which means like, was their brain actually engaged in the act of writing? And when your brain gets, and they've done MRI scans now of people using AI to write essays, using AI to do research with essays, and then doing it acoustically, you know, find your own research.

Starting point is 00:15:13 and then actually write it pen to paper or at hand speed, typing speed, we have to slow down and use your own words. Your brain activity is incredibly low in the first situation, pretty low in the second situation, and extremely high in the third. We're having people's brains atrophy, like in this case where they were looking at the ship,

Starting point is 00:15:34 they're just going to assume the AI is right. That is, without checking it? This is crazy. Where's your P-Doo, Jenny? Where's your P-Den, Jenny? We're playing this week and P did. Let's go. Mine is actually quite low. I think that this will be kind of the golden age of entrepreneurship and that we're going to just find incredible solutions to all of these problems. I mean, I don't know what your inboxes are like, but I'm seeing so many startups that are working on resilience, right? And that is around all the infrastructure that we have. So I'm actually quite optimistic. This is going to, you know, open up lots of opportunities. In terms of the education, I think that's also an interesting, you know, it seems It seems everyone's very nervous about the younger generation. But the way I think of it is, you know, I work with a lot of technical founders.

Starting point is 00:16:19 You know, I help them, you know, build business models and, you know, spreadsheets, things that they don't necessarily know how to do. So I find that these tools are just like total unlocked for people that don't know how to use Excel or they don't know how to do it. And, you know, I agree with you that you need to be thoughtful. You can't just like, you know, relinquish all control to the AI. but I think it like democratizes access to abilities actually that many of us don't have an innate lake. So I'm more optimistic that, you know, people that don't necessarily have inclinations to be math people or science people, you know, can use these skills to catch up. So I think it's more positive. I think that's the right way to look at it. The guide on the side, the ability to fill in,

Starting point is 00:17:06 hey, you know, my grammar might not be perfect all the time. I learned it over a lifetime of, you know, undiagnosed dyslexia. I just did it. Dislexia over all that time. And now I know where to put the coffers. Now I know the difference between then, then, or, you know, two, two and two. Like, it just took a little coaching on the side. And that's where AI actually can fill those things. But you've got to be thoughtful about it. Speaking of being thoughtful, I think we're a little bit too worried about AI killing everybody. I think way more likely is that, to paraphrase the statement, you know, AI is not going to take your job. A.I. is not going to kill us all, but a human enabled with AI might kill us all.

Starting point is 00:17:48 Yes, there you go. That's going. The original statement is Jensen's law, and then the end of humanity is then McClure's law. We've defined it. The end of humanity will be a human with a... I mean, you're not false. I hate to bring this up, but if you were to look at... you know, terroristic attacks and then play the game of like, you know, uh, you know, master's degree terrorist,

Starting point is 00:18:16 which there have been some of those terrorists have had like master's degrees in Germany in, you know, the UK or they've been flipped. Like somebody with a master's degree could do master's degree, uh, level executed AI of insert bad scenarios here. I don't want to actually say them and inspire Mids to go give it a shot.

Starting point is 00:18:36 But this is where monitoring. is such a powerful dynamic, Jeff. Like, they're monitoring, the example you cite, is because Claude is monitoring the usage of their products now. And so this was my reaction to it. If your software is that dangerous and isn't crisp in terms of, you know, hallucinations yet, why don't you know your customer? Why aren't you tracking their usage?

Starting point is 00:19:05 like that would be standard operating procedure for something like a hacking tool, right? Like even if it was a positive hacking tool, you can't go use the Pegasus software from Israel, which, you know, hacks phones or whatever and breaks, you know, all kinds of back doors. You have to have a license. You have to have a social security number. You have to be approved to use it. And there's a log file that they keep of your usage of it. So like, if Claude's that dangerous, like put in some safeguards.

Starting point is 00:19:32 Yeah. But then you have to, like, how do you? monitor something that executes, you know, so many things in parallel and so fast, while you get the AI to monitor itself, but then we've seen incidents where the agents just started hacking, you know, websites and companies and so forth, and all of them did, right? It's not only Claude or Open AI or Gemini or whatever, they all did, right? And how does the agent know that it's wrong and shooting this? When you're an early stage founder and you're just starting out, you're probably going to rely

Starting point is 00:20:04 on a few different tools to run your business. And then you're going to try to duct tape them together. And that might be a fine placeholder until you finish your MVP. But this will get expensive and messy fast. When you're scaling, that's when you need O-D-O-D-O-O-O-D-O-O-D-O-O-D, the all-in-one management software that already has 16 million users across more than 170,000 companies. Odo is bringing everything you need into one platform. That means your CRM, sales, your accounting, manufacturing, inventory, and of course, point of sale, if you have retail.

Starting point is 00:20:41 All of that is right where you need them. And these tools are in constant communication with each other. So you don't have to log out of one system into another and then reconcile the information by using different apps and spreadsheets. When you make a sale, you want that invoice created instantly. And then you want your inventory to get an update. If you're still cobbling together, your back office across five different apps, get started today at odoo.com slash twist. And your first app is free.

Starting point is 00:21:12 That's odio.o.com slash twist. Speaking of SendGrid, I think you would do what SendGrid did, which was, hey, when you first get your MailChimp SendGrid account, you can send X amount. Then after a certain period of time, you can send Y amount. Then if you've built a trusted relation, you go to Z. Now, that doesn't eliminate it, but at least eliminates somebody.

Starting point is 00:21:30 coming in, like a, you know, Russian hackers would go into MailChimp or Sengrid and send, you know, a million spoofing accounts. And they're like, yeah, you know what, first 24 hours you can send 100, then you can send a thousand, you know, whatever. This is kind of interesting. There was a comment by one of the OpenRouter investors that suggested the Stripe acquisition of OpenRouter wasn't about lowest cost routing to open width models. It was about trust relationships and monitoring untrusted use of AI. And that was actually what was going on. Because they want to be good citizens or they see a business opportunity?

Starting point is 00:22:02 Because they see a business opportunity because Stripe is not a payments company. Stripe is a anti-fraud company. And so if you think about like fraud happening in payments, well, you could have fraud or bad actors happening in AI token usage. And OpenRouter would be able to monitor, hey, unauthorized. This is an incredible thesis in theory that answers the question I had like a month ago when this happened, Jenny, of why were they buying this? What are you, what your take on that, Jenny? Yeah, I mean, I think, you know, you were in Europe this summer, and I'm sure you heard that word sovereignty and defense tech a gazillion times. That's what everyone is trying to figure out over there. And I, you know, I feel that, you know, defense tech is really what we're talking about here. We're just not thinking of it in terms of, you know, the typical defense, but we all just have to, you know, get resilient and figure out, you know, how we're going to protect our institutions. And I think, like, cyber is going to be just this massive space. So, That's my takeaway from this and why all these companies are trying to build solutions that are going to protect them and their users. Wow, very interesting concept going on here.

Starting point is 00:23:11 All right. Let's take this first story. Speaking of valuations, we all grew up in an era where, you know, $3 to $10 million pre-seed seed, then $20, $30 million. I mean, my first investment were like a million pre, $2 million pre. Like, that was the standard back in the next. Yeah, but Jeff, that's because you're an asshole. No, that was the market. No, the market for, I mean, the first three or four investments I made,

Starting point is 00:23:35 data stacks, thumbtack, Uber, you know, Com, these were all four or five million dollar valuations. Jason, you were an investor in Uber? I think, let me check. I got to check my distributions. It's somewhere on there. It might be. I think our average pre-money for 500 was below $5 million for the first two funds at least.

Starting point is 00:23:58 I don't know about the third fund, but for most of those. So entry price matters because 80% of startups, 90% of startups at that level fail to return any capital. Here's a crazy company. Instinct is an AI assistant that works through your I message. They ceded it with the VC community. And allegedly, according to reports, they're looking to raise a billion dollars at a $10 billion valuation. It's not open to the public yet. There's a waiting list.

Starting point is 00:24:30 Noah Shin mentioned. He's the 23-year-old founder, who people speak incredibly highly of, says, you know, that they are in talks to raise money. Business Insider reports, and they get things wrong all the time, so take this with a grain of salt, that they now have 100,000 users. According to Harmonic, our partner here, instinct went from founding to 2.4,000. billion dollar valuation in under five months. Oh my lord. That is crazy. The seed, April 2026, $25 million on a 50 million post money valuation. We have to check that. Series A, August 2026, $75,500 million. Series B, $250 million. Benchmark index ventures, $2.5 billion. This is a fast ramp to the panel.

Starting point is 00:25:25 Super fast ramp. And Sequoia and benchmark reportedly circling right now and that they've raised roughly 350 million of the $1 billion. Series A-B happened in the same month? Yeah. Yeah, we may have to check that. I think that data needs to get checked, yeah. I think it was the end of August was the last round. So less than the month.

Starting point is 00:25:44 Oh, no, no. It does say April, August, August, yeah. So I think this latest series B, we're going to have to check this because. Maybe a whole 30 days between rounds. Yeah. I mean, but this is the crazy. of this is people are doing multiple valuation rounds within very short period of time or, in fact, concurrently. They get the investor they want. Then they pop the price up with their name,

Starting point is 00:26:07 Sequoia, Benchmark, line, or whatever the top name you want to use is. And then they pop it up, you know, as part of a two-stage, you can take that the screen goes, two-stage process. So my question, I guess, Jeff, when you're looking at this, entry price matters, obviously a very promising startup. But I do get a clubhouse vibe right now in terms of the playbook here, which this may or may not be as valuable or more valuable than Clubhouse was, but they also had the same VC baiting playbook. So let's talk about that VC baiting playbook, why it works. Other people who did it, I think, you know, Evan, the team, biz, they did this with Twitter as well, seeded it with VCs. Those were part of the big things. So maybe you can talk about the valuation

Starting point is 00:26:55 here, these valuations ramping up so quickly in succession and the fairness and or strategy of doing that. And do you get a little bit of that clubhouse vibe with an unrealistic valuation? I mean, look, I'm not an investment. We have a couple of people who've used instinct and the experience was just not great. And the feedback we got was that it had actually, actually, access to so much information, there was concerned that they would actually sort of screw it up, so people stopped using it. Look, the investors that you showed are amongst the best investors in consumer internet and consumer AI.

Starting point is 00:27:36 So they know what they're doing. And if they think that by paying $2.5 billion, they can still make a return, we're going to wish them luck. I think the big question to me is, so what does instinct look like after Muse launches as the top downloaded app in the app store. And from what I hear, because I don't have time to just test everything, it's just like it's a literally full-time job to test all those things. So I haven't used instinct or muse or any of the other personal assistants.

Starting point is 00:28:06 You just hire an agent to test all the other agents, Jeff. That's how you get that done. I'm old school. I do it myself. I trust myself. And so like what are they thinking where, and that's sort of the issue is the reading of innovation from all the major companies is so high that when you invest in a startup, you don't know whether you're going to be

Starting point is 00:28:26 crushed by, you know, whether it's open AI or anthropic or meta, because there really something that you haven't heard of and boom, suddenly you're dead. And so I don't know what's going to happen to instinct, but if I'm then, I'm really worried about news. Yeah, I'd say like there, like we shouldn't, you know, the audience shouldn't confuse, you know, a fast funding round with like a real business. or a signal around that, right? Just because there's a markup, it doesn't really tell you much about the business, right? So it does tell you about the competitive, you know, tension among a small number of funds, right?

Starting point is 00:29:01 Who believe that agent products are kind of, you know, the next platform shift and they want to get in it. So I think that's, you know, what the message is. And that was kind of your question, Jason, is like. And we haven't seen anything sort of super interesting in, like, consumer for a while. And so I think it's just that when you add, FOMO, fear of missing out, and the fact that for once there is something which seems to be so interesting and the founder seems to be very compelling. They're just pulling a bunch of money in Clubhouse is a great analogy, right?

Starting point is 00:29:31 They got to $1 billion in valuation, then $4 billion in valuation, and then crickets, right? I think this is different. I mean, I think we were all around for like, Hubhouse and Meerkat was another one that was in this category of like super big flash in a pan. But the valuation for instinct, I think people are underwriting a major acquisition. happens to every business at some point. You're nearing the end of the month or maybe it's the quarter and your sales team is at risk of missing their quota. You could waste days pulling together data from across the entire organization or you could just ask Rippling AI. Rippling's already built on your real-time people and business data, obviously. So Rippling AI can pull metrics from both Rippling and your CRM into a meeting

Starting point is 00:30:13 ready dashboard. See exactly how close you are to hitting that quota. Break down your revenue by region or sales rep or more, all in just seconds with Rippling AI. Look, you're running a company, you're busy, you have critical questions that you need immediate answers to, and you can't just pile a ticket and wait a week for your team to get around to sending you a report. And you don't want to sit around waiting for reports. So you need an actionable insight right now. So if you're ready to rule your business, head to rippling.a.i slash twist to get the only AI bill to give you full visibility and let you take complex actions across your startup. Is that a good way to be in investor is to invest at an incredibly high valuation, hoping for, I don't want to say

Starting point is 00:30:55 greater fool, but a desperate acquirer. No, no. In this particular scenario, yes, as long as you believe that instinct is a real product, because the competitive, you know, options there are ones that are already owned by major distribution platforms like Open AI Anthropic or, you know. So Open AI can't, doesn't need to buy this. Meta doesn't need to buy this. GROC doesn't need to buy this.

Starting point is 00:31:18 Anthropic doesn't need to buy this. So we're leaving Apple, Google, Amazon. Microsoft. Microsoft. At what price? And I would say Apple. But at what price? If you're coming in at a $10 billion valuation, I mean, you think they're going to buy it?

Starting point is 00:31:35 You pay as a percentage of your own distribution. If it's good enough. Well, as long as there's one other buyer out there, then you have a competitive option. See, I think, Jenny, you're making a very important point here. Give it to me one more time and explain your thinking behind this. acquisition price and the distance between these two things. I mean, everyone's squinting when the market's like this and says, well, the exit opportunities are just that much bigger.

Starting point is 00:31:57 But we haven't really seen that. So if you come in at a $10 billion. Wait a second. Yes, we have. We've just seen all the examples. Cursor. Right. Cursor had $2 or $3 billion and they got valued at $60.

Starting point is 00:32:09 This has zero and is at $10. Again, you're going to be raising this on the acquirers distribution and ability to monetize. And I think you can see that meta's already got a shit ton of usage. And it's a pretty decent product right out of the gate. Jenny does Muse's success and Zuckerberg's unique ability with his giant ad network to monetize this with advertising and never have to charge for it. What does that say about the business model for instinct here? I mean, to Jeff's point, I would be worried, right?

Starting point is 00:32:46 I mean, threads weren't there like 500 million people that use threads? And that's not, you know, a great product, I would say. Or, you know, a derivative product. Yeah, exactly. So I think, you know, just the ability to reach eyeballs is pretty strong here. So I would be worried about the long term if I was going in at a 10 billion valuation. So, yeah, I think the way I would like this. The other worrying thing that we heard is that they're burning so much cash on compute.

Starting point is 00:33:12 And you wonder why, because the product is fairly simple. They're bringing so much cash on computer that they need to raise, you know, on this next round, almost like a few weeks after closing the previous round, right? And they're staging and they're limiting the number of people who can use it. And so I don't know whether it's an engineering issue or it's a really complex set of complex tax that they're getting the agents to perform. But it's really surprising that they're leading money on fire the way they do. Well, I think the founder came out and said that he just didn't want to charge people. So I guess that leaves advertising, you know, down the road. That's the issue with consumer internet.

Starting point is 00:33:48 It's very hard to get people to pay. I don't think anybody's worried about, like, getting money out of this right now. I mean, WhatsApp was acquired 10 years ago? Is it still, is it monetizing yet? They did actually have, they were advertising a dollar per user and had $600 million at one point, right? They had a really clever use thing there. Here's what I think, you know, when you look at this, a couple of things can be true. And looking in the history, it does rhyme.

Starting point is 00:34:15 This is a smart move by the founder to make hay while the sun shines. If you can take a billion dollars and put it in your bank account, then you will be the consolidator of the people who run out of money and you have, what, three pivots in you, two or three, like major pivots in you, or brand extensions. So you can play the long game, even if you're burning a lot of compute. But number two, when a price gets this high to Jenny's point, If instinct is at 10 billion now, and you have muse in the market, then you're going to push the potential acquiring folks to say, build or buy.

Starting point is 00:34:52 And Twitter offered $4 billion for Clubhouse, I believe was the rumor. They did not take it. Huge mistake in the history of Silicon Valley. One of the biggest mistakes ever. It was an 18-month-old company at the time or 12 months? Friendster acquisition by Yahoo. Okay. Listen, we got really old people on this podcast.

Starting point is 00:35:13 So Dave's got 30-year-old opinions here. He's got the receipts from 30 years ago. Greatest non-acquisitions, like fab.com was a, I remember Shervin or somebody's like, I own 7% of Fab. It's worth $2 billion. We got a $140 million position, if my memory serves me correct back in the day, went to zero, I believe. It was like part of the fast fashion, fast commerce category.

Starting point is 00:35:37 So the issue here is, At 10 billion, you have to be purchased for 20 billion at a minimum. That would then lead Apple's board, Microsoft Satya Nadella. They would just say, well, wait a second. What can we build for that amount of money? Let's offer, you know, people $10 million, you know, salaries, $25 million salaries, and let's just build this ourselves and go along on it if we actually believe in this. It's not, entry price matters, and this is now breaking entry price.

Starting point is 00:36:09 metrics, but I wish them well. Anybody can think of a big miss? I gave Fab, you gave Friendster, and I also gave Clubhouse. I mean, Fab never had an acquisition offer. Google. Google itself. It just had a crazy valuation.

Starting point is 00:36:22 Were you also an investor? Yeah, I was, you know, I was in the seat around, so they went from zero to a billion in like a couple of years, and it was the darling, but the problem is that the unique economics just didn't work, right? They never had. And if you had sold that, what would impact would that have on on Cork? I mean, it would have been great.

Starting point is 00:36:40 Two or three, whatever it was in. It would have been great. But, like, we never got to that position because investors just looked at the numbers and they said, this is just not sustainable. That's the issue when you actually have numbers to look at. In the case of instinct, they don't have any revenue. They barely have any users. They're 100,000. And therefore, it's all about dreams, right?

Starting point is 00:37:03 And if they're raising, like, three weeks after they raised before, you don't really get the retention data either on those 100,000. So I'd say that's an issue as well. Yeah. I'm a skeptic on this one. I think there's real value in the table here. And it's just based on the monetization or potential the acquires. And the acquires are substantial. Like Apple, Amazon, Google, Microsoft are all potential acquires.

Starting point is 00:37:26 And they definitely need the product. The rumor I'm hearing with Google is big frontier model news coming and some other consumer products coming. They've been kind of laying back because they're conservative. they don't want to release broken products, given lawsuits that they get from large governments and just their reputation, they kind of moving slow,

Starting point is 00:37:48 but there could be significant things dropping. And if you'd think about the Mews opportunity, they have it sitting right there in our data, Gmail, calendar, docs. They have it sitting right there. Google might be able to build it themselves. Google might be able to build it themselves or modified Gemini,

Starting point is 00:38:04 but I think Apple and Amazon are definitely, you know, They have to have an agent product and they don't right now. I had an opportunity to invest in fab.com, Jeff, when it was a gay social men's dating site. Oh, okay. I was waiting for that one. I said to Jason, I forgot Jason's last name. I said to him. Jason Goldberg.

Starting point is 00:38:26 Jason Goldberg. I said, Jason, I'm on your gay dating social network. I created an account and a profile. This thing is the best designed social network. That was why it was interesting. That was why it was interesting. This is incredible. And I'm getting a lot of pings here.

Starting point is 00:38:41 And I am gay kryptonite, according to one of my gay friends, dubbed me as gay kryptonite. Gay guys would be like they would lose their gay power if they were in proximity to me. I thought that was the funniest line ever. I said, if I can do this well on a thing. And then I offer him 50K, you know, whatever, $10 million, $20,000. I mean, comes back to me, Shaka. I got a better idea.

Starting point is 00:39:04 And he gives me this idea. And I said, yeah, I don't want to do e-commerce, not my wheel. house, but I wish you well. And I transitioned from gay dating site to fashion design marketplace. Actually, not quite a stretch.

Starting point is 00:39:18 Yeah, not as much as a stretch. They had an amazing taste, that's for sure. And the website was stunning. It's just the, unfortunately, it's the business which wasn't sustainable. It wasn't the concept. They just had to sort of figure out a way

Starting point is 00:39:32 to make this, you know, kind of venture work. Vantes Privet was the French one. Yeah, Yeah, Bonde Privy was the big one, and they were, you know, Jacques Conjon was extremely successful. I don't know where they are now, but they became like a sponsor. And then there were two or three others in the United States that were in this fast fashion flash shells. What were the other flash shell ones?

Starting point is 00:39:53 They all went to a billion. Guilt. Yes, guilt. There were other flash shell ones. Incredible. Jason was always a snappy dresser. No, I wasn't always. My wife sat me down.

Starting point is 00:40:05 No, no, no. Jason Goldberg. I meant, sorry, the fast. dot com. Oh, okay. You're Jason. Not gay kryptonite. My wife, you know what?

Starting point is 00:40:15 My wife did? She said to me, Jenny, we need to get you a look here. You know, you're out in New York, black t-shirt, black jeans, you know, black blazer. Like, it's a little bit repetitive. I said, okay. And she said, you've got to find like a style icon in somebody. I said, yeah, you know, I found my style icon. She said, who's that?

Starting point is 00:40:32 I said, Daniel Craig. She goes, yeah, perfect match. That's it. So then I went and got Tom Ford. and I literally found every single thing he wears in the James Bond. I just bought all of them. And did you also hit the gym? Shaken, not stirred.

Starting point is 00:40:45 Yes. I lost 40 pounds and I gained 10 pounds of muscle. And yeah, I mean, I'm not. Listen, I'm not coming out of the ocean in like Mancini like him with a knife on my side. Yet, yet, yet. But next summer, I may be a centropay with Jeff Claudia. There we go. And we might come out in our dueling, yeah, do not make, I'm just going to warn my super

Starting point is 00:41:06 I don't want to see a video, AI video of me on Daniel Craig's body coming out of the ocean. I think that's great. I actually have a question for you, Jason. You're a big open claw guy. So are you guys going to move to Muse? I mean, isn't... We gave up on OpenClaw. Oh, okay. What happened was we had four people, three or four people in the organization who were technical enough to do it. And then all of a sudden, Claude Co-work, Perplexity Computer. We started, having other options and those were easier to use. They were delivering faster and we went from three people in the organization using them to, you know, I have like 20, 25 people here, you know,

Starting point is 00:41:47 so all of a sudden went to like eight or nine people. So the ease of use took over and now nobody's even thinking about OpenClaw anymore. I have three or four people who are addicted to Grockbot and the restaurant clawed code. I use perplexity computer, uh, I use perplexity. I use perplexity as well, and then now lovable, we're actually building intranet software. So we've built what was the, Jeff, do you remember? What was the VC, like SaaS software? I think Joe Lonsdale created it. Adipar maybe?

Starting point is 00:42:23 Adipar, was that it? There was a bunch of VC software. For fund management. Which is the reporting. Adipar was reporting, and I thought he had another one, which was like deal flow management and all that kind of stuff. my guys in like three weeks, we get 20,000 applications for funding. We have everything organized now, Jenny.

Starting point is 00:42:42 We have AI so we can go into it and say, what are the companies in Defense Tech that we met with, didn't invest in, who subsequently raised a series B, and please put them on a list. I'm going to email them and talk to them about the syndicate and try to make up for the missed investment. We did. Give me the notes of who met with them in the firm. No, I'm seriously doing this. We're doing the misses.

Starting point is 00:43:06 So we missed Zipline 10 years ago. That's a great use case. That's a great use case. We missed Zipline. Keller came on the podcast multiple times. I was absolutely enamored with him. And then I have the syndicate and we invested in their last round. It was a couple billion dollars.

Starting point is 00:43:20 Look for all the emails that I didn't respond to where there was an institutional round that happened after I got the email. Wow. That would just be like the most painful like analysis. Oh my God. Wouldn't it be? So many companies that I go back and like, Oh, yeah. That found your email me and I didn't. All right. Let's go, let's keep moving here. The docket. Open AI and Anthropic both ship new models yesterday. They did it hours apart. But here's the important part, Jeff. We always talk about profitability. Is there a business there vis-a-r discussion on fab.com? We're going to build on that. They both cut prices meaningfully at the same time. Open AI's GPT6 Seoul and Luna came in about 50% cheaper with Seoul at $2.2 per million tokens. Anthropics Claude Open 5.5 matches Fable 5.1, walk off.

Starting point is 00:44:05 and 40% less to run than Opus 5. Neither a big capability jump, but both of these are masterprice cuts. And what are your thoughts here, Jeff, on selling tokens, the token selling business in the age of open source models? Well, it's a race to the bottom. That's sort of the issue. So you have all the funding that they raised that essentially is sponsoring all this usage. right and they basically sort of predicate their future to their ability to raise the next round

Starting point is 00:44:41 so that they can continue sustaining like the fact that they're losing a bunch of money as soon as someone uses their product and unfortunately like they're getting to a point of like the cash needs that they have are such that you you wonder if they're going to be able to raise you know more capital I'm sure they will but it's just insane so it's a race to the bottom um I At some point, you wonder what token price is going to be. And that enables our startups to basically sort of use that compute at a much lower, artificially low kind of price. But we try and figure out, okay, when we invest in something, what is the true cost of compute? Because you can't find yourself sort of upside down.

Starting point is 00:45:26 Because if ever, for some reason, your compute costs are increasing by, you know, two, three, four, five. you may have a product which was actually generating margins and made sense that no longer makes sense. Well, your margins go up temporarily, but your competition goes up as well. So I think that's the problem with this. I mean, it sounds like a great thing, but ultimately your barrier to entry, other barrier to entry goes down as the cost come down as well. What are your thoughts on this business, Dave?

Starting point is 00:45:58 Is it a great business, a good business, a TBD business? here with all this headwinds. One of the routing companies, I won't say which, I don't want to give these plugs for all these people now doing data and then pitching my people constantly producers, you get free ads on the program. So now I just abstracted. One of these web providers, 80% of tokens being tracked

Starting point is 00:46:20 by this one web players, routing companies, open source now, Dave. So you want to be in this business or not? All right, so here's your open versus closed token, token volume, open weights, open models, 78% closed weights, 21%. Dave, tell me what you think about open source and the impact on frontier models. And let me ask it to you in a very direct way, with your own family office money, with your own money, would you put a bet if I had to put a gun to your head, you know, $2 billion dollar frontier model company like Anthropic, let's say,

Starting point is 00:47:01 you know, 20, 30 times revenue, or you could just put the money into an index of Nvidia, AWS, you know, and the infrastructure companies. Where would you put your money, which is a way to say open source, open tokens, infrastructure versus frontier models? Where would you want to put your entire net worth if I put a gun to your head? Probably 50% index on open weight models and then split my closed models, a third, a third, a third, between anthropic opening eye. If you could only pick one side of the bet, I put a gun to your head.

Starting point is 00:47:33 Dude, I don't have to fucking play by your rules. I'm venture capitalists. I can make diversified bets. Way to kill the momentum of the show. Jenny, here. Let's see if Jenner. I go all open source. I go open source.

Starting point is 00:47:44 I go open source too. I go open source. I think they're going to get crushed. Say more, Jenny. Here we go. Well, there's no rules around open source. You can, you know, the existing models can say they want to slow down. They can change their pricing.

Starting point is 00:47:58 The open source guys. can do whatever they want. And they're just going to be, at the end of the day, consumers don't care. They just care about price and quality. And they're so close to catching up that we're going to go all open source. Most applications don't need the latest and greatest features of those latest models. And so when you have, and the problem is that the model has to figure out, well, is this a complex task?

Starting point is 00:48:22 And I need to actually send it to like this model, expensive tokens and so on so forth. So can I send it to a lower cost? easier, you know, kind of routine. I don't know how to do that because I'm sure that it's a genuine issue for them. Like, not that many people need those kind of latest models. They need it to compete against themselves, but most applications are fairly benign. Okay, Dave, you're my cleanup hitter here. I always put you right there in the fourth slot there.

Starting point is 00:48:49 Does my, no, I'm seriously, a home run hitter. You have lots of bangers. So you heard all of our positions, I would go 100% into the hardware stack. I think the frontier models are trouble. because I'm basing it on what I see on the field vis-a-vis my startups are all going open source, do not want to invest in the frontier models anymore. And then I talk to 11 labs. I talk to Lovable.

Starting point is 00:49:12 Those CEOs indicated to me that they are also following suit, Figma as well. They don't trust the frontier models. They think they're going to be competitors. What do you think close this conversation up for me? I think you're right that there's going to be growing revenue coming from open web models. Right now, there's already, you know, open weight models have taken over a majority of token usage. That's been clear since earlier this, I think, second quarter, but definitely. Yeah, 74%ish.

Starting point is 00:49:38 Yeah. But I don't necessarily feel like opening eye and Anthropic are going to slow down and generate revenue. They might have margin issues. They might, you know, in particular, I wonder if Open AI's, you know, bets on, you know, they're spending in the future, which is a lot of debt. They do have to worry about who's going to be able to pay that debt. even if, you know, Papa Jensen is backing them up. But I think, you know, if you're making a bet on open weight,

Starting point is 00:50:04 you're probably just making a bet on Nvidia, right? There's not a single bet that you're able to make right now in a company that's like, hey, let me play the field. So I still think, you know, and actually, actually, I might even wonder if, you know, SpaceX becomes, you know, that platform for open weight models. So maybe your picks and shovels, you have two bets. You could bet on invidia because they're dominating the market,

Starting point is 00:50:26 but you could also bet on SpaceX being, you know, a little bit to step up the ladder from just Jensen's play, if you believe in open weight. But I still think that both OpenEye and Anthropic are going to do fine. They're going to be growing revenue. They're both going to go public for, my guess is Anthropic at least $2, $2.5 trillion and probably opening eye at $1.5 trillion. Whether they keep growing at that rate and whether the growth in the future, I do wonder maybe the growth of market cap slows down.

Starting point is 00:50:56 is interesting, Jenny, because we do have a corollary here, which is SpaceX's IPO, which had a major bump, a pullback, and then it came back. So it's kind of was priced to perfection, I think, would be, you know, what Bill Gurley would have. Only because he pulled a rabbit out of his ass there. Well, you're talking about the cursor acquisition, perhaps. And no, well, yes, right. Services, EWS, where he's got 20, 30, 40 billion dollars in compute. Whatever you want to call that that business did not exist nine months ago. Like that is the most fucking crazy shit is that a 23 year old company does a massive pivot.

Starting point is 00:51:34 And in less than a year, you know, a business that didn't exist becomes the majority of portion of their revenue. Like that would be like insane. It would be as if Amazon in one year added AWS or something, you know? Right. They were just selling books and cables, Jenny. And then all of a sudden just fell in their lap, the AWS.

Starting point is 00:51:54 which, you know, it took them a decade to build up that AWS revenue. But in all seriousness, these valuations are 20 to 30 times top line revenue, Jenny. These things will be priced to perfection. They won't be a very good bargain for public market investors. How many years will it take them to fill in that valuation if they go out at that 20 times top line revenue in your mind? And would you advise family members who might be calling you? Like, should I buy Anthropic and Open AI? Like, what's your instruction to them?

Starting point is 00:52:22 I'm a seed state investor. I'm a pre-seed investor. What do I know about the public market today? So, no, it seems, it seems crazy to me. Got to get in early, right? So I would not advise people to buy right now. Even though I do think, you know, there is potential there, I, you know, I think we're going to see ups and downs.

Starting point is 00:52:39 Jeff, what are your thoughts here? You know, you're in the family office stage of your life as well, having done particularly well over the course of 30 years in venture. So what are your thoughts here in terms of owning the incredible, you know, you're companies with incredible leadership, with incredible products, with incredible growth, but, man, they are pricing certain public stocks, whether it's Palantir, Tesla, SpaceX, OpenAI, and now anthropic. They're pricing them like venture investors would. Yeah, but as public markets. I'm still looking for my entry price in SpaceX. So I haven't bought SpaceX. I want to own SpaceX,

Starting point is 00:53:21 but it's too expensive right now. Right. Like, I don't know. what my buying price was. I was like probably one trillion. I would buy like $2 trillion. Nah. It's probably ridiculous. But like the goal of the family office is not to make bets that are potentially money losers.

Starting point is 00:53:38 The goal of the family office is to not lose whatever we've made so far, right? Preservation. So that's why we have a financial advisor who goes into like these days invests in energy. Like energy is interesting, right? We invest in things. which are a bit more sort of peasant and predictable, because everything I do at Unquark is still sort of pretty insane, because we have to deal with this environment

Starting point is 00:54:04 where every valuation we look at is absolutely insane. Interesting chart from a company that tracks cap tables. I'll leave it at that. I won't tell you which one. No free ads for these guys anymore. Enough. But here's C valuation according to a source. It's a limited dataset, but it is a dataset nonetheless.

Starting point is 00:54:24 Man, 2017, you had seed round valuations. Ah, you know, 50th percentile was 8 million. 50th percentile is now 28. And then you get into the, you know, 75th percentile. We are going from roughly 12 to 54X. And in this elite, 95th percentile going from 25 to 209x, or 8X exactly, sorry. So is it possible, Jeff? in your experience to make money in this environment as a seed stage investor.

Starting point is 00:55:00 And if so, what do you have to do in terms of discipline to not overpay and wind up being 1X as a fund, which is death? The way we have adjusted to this environment is basically continue investing at higher and higher valuations. And we don't want that, but that's what we have to do in order to win those deals. when there is a company that we believe is interesting, that the founders are exceptional, that the product is very differentiated,

Starting point is 00:55:29 and we think there is an opportunity to build an iconic company, then we're going to say, well, we want to own 10% or 12% of that business, and I'm going to write the check that allows us to do that. And the good news is, having been around for so long, we rarely sort of lose deals. The only few times we've lost deals recently is when a multi-stitch fund,

Starting point is 00:55:51 So someone who has, you know, a multi-billion dollar kind of pockets that can invest at seat stage, series A, series B, and so forth, come in and write up, you know, $10 or $15 million seat check at $100 million or more valuation. And at that case, in that case, we just bow out, right? Because as you said, the big issue is, in order for us and our economics to work,

Starting point is 00:56:17 we need to be able to return, you know, ideally the entire fund or half of the fund, our fund is $300 million, so $150 million on that one single investment, which means that if we plow $5 or $6 million into it, we need to make a $30x. And $30 on the valuation, which is already pretty high, means, oh, this is a multibillion-dollar outcome. Like right now, unless you sell something for $10 billion,

Starting point is 00:56:43 you barely sort of make money, which is insane, right? Yeah. At the seed stage. So if you come in at a $24 million post, which I think was the new number for seed, like you're expecting decacorn exits. So that seems like if you do the math, that's pretty shocking. I think we also have to consider time frame here. It's not just multiple, but what time frame does it take to get that multiple?

Starting point is 00:57:05 This is a very, I think, challenging time. And I look at my investments during that SaaS peak Zerp. And we tried to be net sellers during that period. and we had a couple of great exits. We were able to sell in secondary to the TPG, Tiger Cubs, whatever, the people who dip down. But man, if I could replay that, I would have pushed harder to sell more at that unreasonable peak, which means if you were to fast forward that to today, Dave, you would want to do that at this moment in time based on your new business, which is your, as I jokingly said, the last time you were on a strip miner,

Starting point is 00:57:46 you like to pull the strips out of our previous. We buy used cars in good condition. They just happen to be startups or fun. You buy collections of used cars. Let's be clear. You find the rich old guy who's got a barn collection of Ferraris and Corvettes. You go to his widow and you try to get them to give you a really good deal knowing there's like a Ferrari like from 1965 in there.

Starting point is 00:58:09 That's the big winner. And the rest might be, you know, just one X or two X. I don't think we're trying to take money from widows. that's not our general game plan. But I would say, but the point is well taken that there are assets, again, the top 30 names,

Starting point is 00:58:25 it's a seller's market for those assets. If you have positions in Anthropic or any of those companies, even not Anthropics, let's just say, you know, and stripes and databases and all those companies, they can set their price, right?

Starting point is 00:58:39 If you go outside the top 30 names, there's not necessarily a market for the names. And some of those companies are doing a couple hundred million in revenue, growing 40, 50%, profitable. There should be a market price, but there's no data on that company for the average retail investor, and there's not people going to buy those companies. So it's a buyer's market for a lot of the names that are outside the top 30, if you know what you're doing.

Starting point is 00:59:02 How do we get companies 31 through 250 to get on the radar? Because the opportunity is probably in that group. Is it not, Dave? It is, but the challenge again is transparency and liquidity because most of those companies are not publishing their numbers. Should they? Should they create a market? Well, that's a really interesting question. Yes. There's an interesting question there. Theoretically, cost of capital would be lower if they were more transparent. But you have to play that against the cost of being public and that's expensive. And so that's why companies aren't going public until they have half a billion billion dollars in revenue.

Starting point is 00:59:43 I talked to a CEO the other day who, you know, I won't name the company, but they're doing more than half a billion in revenue. And I said, when do you think you would consider going to public? He said, not until they're doing 10 billion in revenue. I was like, wait, did you mean 10 billion in market cap? And he said, no, 10 billion in revenue because, and, you know, I actually think he's not being insane because he probably can get there. But he was basically suggesting it would be another five years, you know, before he goes public. For a company already doing half a billion in revenue profitable, you know. So that's the market conditions we have is all these companies, we talked about this last time,

Starting point is 01:00:19 all these companies would have gone public 10 years ago at 50 to 100 million in revenue. Now they're not even going public until they have at least a half a billion to a billion in revenue. You have all these now private companies, which like you said are great opportunities, but there's no reporting. There's no disclosure in reporting, unless you can get access to their numbers. And so that's the arbitrage. Yeah, there's a lot of bargains out there as long as you know what you're buying. And the problem is that because the market is not ready to get anything below half a billion

Starting point is 01:00:48 in revenue and even that, that question whether they would actually go public today, then you have thousands of companies which are kind of stuck in the middle, as you call it. And unless you're lucky and someone sees the potential and comes and acquires you for like 10x revenue, That's what happened to us with drone deploy, which is a company I invested in 13 years ago, and we had like a great exit, return, half the fund to investors. That's how we make money as venture investors, right? But like for one that we managed to exit at a great multiple, you look at bending spoons, average like 2x revenue multiple that they pay

Starting point is 01:01:30 when they buy something which is making like $400 million revenue. It's a disaster for the investors. were involved. Those companies are getting rid down by 80 to 90% from their last round mark. But do you guys think this is going to change when Anthropic and Open AI go public? Do you think they're going to be the floodgates are going to open? What do you mean for other secondary startup purchases? Well, you take the top two out. The top two represent 50% of the demand. I'm picking a number out of the air here. No, you're right, Jason. It's pretty big. Half the market is five companies.

Starting point is 01:02:03 Remember that SpaceX Anthropic and Open Air Air. will essentially raise in the public market more than all tech companies have raised in the last 45 years. More than $5 trillion. More than $5 trillion. The numbers are just insane. Insane. And what's really interesting. But there's still going to be three to $5 trillion of other assets that are private.

Starting point is 01:02:27 It won't be concentrated in as many companies. There will be now more. But I still don't think we're going to see everybody start buying, you know, company number 140. Like they're going to be looking at Anderil and Stripe and Databricks next. Maybe they'll expand their aperture to like 20 or 30 companies. But again, there's no disclosure of information. The average retail buyer is not going to be looking past, you know, company number 30, maybe number 40. And that's where the buyer's market is if you know what you're doing.

Starting point is 01:02:58 We've started a process here. If you go to Twist 100 to try to address. this. So I'm having my team just make our take on the top 100 companies here. And we're going to keep editing this as a list and come up with some ways for the community to participate in them and try to find the latest prices and have discussions. So if anybody has any ideas, we vibe coded Twist 100. We're also using that for our topics as a way to get the topics going. I love that idea of Jason. We did a very similar. Oh, did you really?

Starting point is 01:03:36 And that's at a practical? We used augments numbers, but yeah, we re-host the top, I think, 40 companies. Got it. Okay, there you go. But it's interesting. It depends on how you do that rating and, you know, could be based on revenue, could be based on growth. Revenue, valuation, total amount raised, likelihood to go public. Demand in the market.

Starting point is 01:03:56 Oh, demand. That's a good one, yeah. Secondary is. I think right now what the key question is, will there be sort of a question. required, what will be sort of the potential multiple. We've seen quite a few early stage acquisitions. And for the old parts on the podcast, that sort of reminds us the early days of Web 2O. Because I mean, Jason, you were acquired early, right, by OL. Weblogs, like 30 million, which at the time was a lot of freaking money.

Starting point is 01:04:28 Back in the day, like 2004 to 2008, 9, we saw a ton of. companies being acquired for like really good multiples, didn't have much revenue, but those acquisitions, the $30 million for user plane or for Weblogs Inc or the $50 million. Delicious Flickr. All those, you know, at that time, because we're investors. Trip down memory lane. Yeah. In the early days at low valuations, we made a bunch of money.

Starting point is 01:04:59 Yeah. 10, 20, 30. Yeah, you didn't come in at 24 posts. Mark Cuban invested 300K for 15% of the company. And he gave us 300K. And under 18 months later, we gave him back $5 million. So he was pretty excited about that. Didn't move the needle for him.

Starting point is 01:05:19 But I went to him. I was like, we just got a $30 million offer. He said, take it. He said, how much money did you have in the bank? I was like negative $30,000. I just bought my wife for $20,000 engagement ring. And I've been floating the sales guys salary for, Weflux income on my corporate card.

Starting point is 01:05:35 And he's like, what is the question again? I'm like, here we sell? He's like, what's our revenue? I was like $125,000. And we have like two other contracts coming in for another 125K. And so they paid 300 times revenue. And there were all kinds of headlines at the time. AOL is the stupidest company in the world for buying 95 loss from JCal.

Starting point is 01:06:00 Take the money, buddy. And I was like, you know what? I'm negative 30. said to my wife, I'm taking this money. It was me, Brian Alton, and, you know, Peter Rojas had a piece. And we, we splashed some cash around for the rest of the group just as like a mitzvah kind of situation. We had some great deals. Slideshare, Mashery, and Mint.com were all great, you know, only $100 million, $200 million exits.

Starting point is 01:06:22 But the point is that we're starting to see those kinds of early acquisitions again in an era. The valuations are under the same because the entry price is higher. we don't always sort of make the same money. But at least we've seen more acquisites. Like we've sold about $1.5 billion worth of stuff in the last three, four months, which hasn't happened for a while. Three, four months. Yeah.

Starting point is 01:06:45 A billion dollars in the last three months. And it's been a pause for five years. Yeah. No wonder Jeff is starting a new wine company. It has nothing to do with it. Oh, yeah. Right. Sure, sure it doesn't.

Starting point is 01:06:57 Sure it doesn't. Jeff is now in his 50s. he's got a certain amount of time left and he says, hey, third act, I want to be able to have my own my own side quest and that's wine and I think it's fantastic. Jeff, tell us a little bit about it.

Starting point is 01:07:12 Cheers, bitch. Well, it's not exactly that but basically after 21 years running Uncork I've asked my awesome co-managing partner Andy McLaughlin to take over the firm so I'm no longer running Uncork. He is.

Starting point is 01:07:29 I'm now the founding partner, which means nothing, which means I can do whatever the fuck I want. And that means I continue investing in deep tech, which is awesome. I'm heading an absolute blast. But my wife and I decided to leave California, which we did five months ago. And we moved to a small town called Wala Wada, Washington, which is an incredible place, lots of wineries, awesome people, they don't care about tech. and we've opened a small winery called Clow Cladier after one name.

Starting point is 01:08:02 I didn't name the firm, you know, us, but we named the winery after us. And we just put our first couple of wines on the market. And they're absolutely fucking delicious. I'm super proud of what we've done. So great. Which will, tell us which bottle we should buy. Oh, so go to wines and you will see we have. So we named our wines after terms of the.

Starting point is 01:08:26 B. Series B over red wine. Hey, you want some liquidity? Jeff's got liquidity. And wait for it. The next one, the next one is called carried interest. So you'll be able to buy our carried interest later and it will be an exceptional wine. On the left, we have your big fans

Starting point is 01:08:49 of Spanish wines and you decided to recreate that flavor profile. Is this a a cabernet? It's a, it's a, it's Grinash, Cabernet Sauvignon, and 12% Syrah. Typically, it's Scarellian, but syrup is what you decided to use. What's the flavor profile? What am I going to get on the tongue there? And then what would I pair it with?

Starting point is 01:09:10 It's very grunashy. So lots of red fruits, sort of lighter, very easy to drink, awesome sort of wine. Very, very pleased with the work. You put that with some chicken or something or roast chicken or steak? Yeah, chicken or pork. Now, steak, I would still use something heavier, like Cabanets Sauvignon, like I'm typically our carried interest will be perfect for that. And then the inspiration for the white wine, which is 100% the Eurnier, is French conglue.

Starting point is 01:09:43 So this one is slightly acidic, extremely pleasing again. We just started to ship the bottles to our customers. and I'm trying to see on social people enjoying our wines, it's really awesome. But to be clear, 95% of my time is deep tech for NKork. Got it. Okay. Well, to my team, I want a case of each, please, to my team, ordered immediately. And then I want everybody on the staff to get to take a bottle, okay, from the office.

Starting point is 01:10:13 Thank you. Thank you. I'm ordering this for the staff. And when we'll have our staff dinner on Wednesdays, we have our management team meeting, maybe we'll crack one of these open to a little taste test. So do me a favor. just decant the series B for at least an hour so that it opens up because it's still very, very young. So you will enjoy it a lot more if it's decam.

Starting point is 01:10:33 Love it. This week in Wine is now, it's a first episode here. All right, as we wrap here, I want to talk about investing in a dovetail two things, investing in unpopular categories here. Maybe, you know, people are obsessed with agents right now. They're obsessed with on-prem models. We have a great investment in Go AI, et cetera. training data. I got a great investment in Micro One. We got some great investments there, but they were made three years ago. But today, Dave and Jenny, where are you looking,

Starting point is 01:11:05 where there's not a lot of people looking that are not the obvious categories, Dave? Is it possible to invest in 2026 into 27, the year of our Lord, my brother in Christ? Whether you believe in Christ or not, we are still brothers in Christ. I could be a lowercase thing, if you like. But my brother, where can we find value going into 27? Well, Dave just told me that fintech and healthcare were, you know, no-goes. These are not sexy categories. That was news to me as a fintech and health care investor. They're not no-goes for us. I'm just saying that other people don't seem to care about them. I know. It's insane.

Starting point is 01:11:43 Yeah. How could they not care about fintech when, like, my Robin Hood investment, I wake up as a I own more Robin Hood shares now than I did when it was a private company, because Because when it hit like $8 or $9, Jenny, I was like, wait a second, this stock is mispriced. I'm going to back up the truck. I bought a bunch at $8 or $9 and now over $100 a share. I, without getting into details here, but I'm making a significant percentage of my gains in Robinhood as a public market investor now, like double digits on my returns. And I invested before I was released as a product. So maybe talk to us about why people are missing fintech specifically.

Starting point is 01:12:21 Well, get this. I talked to one of our portfolio companies today at the intersection of FinTech and healthcare. They raised like a 2 million pre-seed and they've been, you know, just crushing it. They're doing about 3 million ARR. And they kind of tested the waters on a Series A and that was like no go. Everyone said they didn't like the margins, the categories. There was a million reasons. And so now they're going to kind of raise like a, you know, a 3 million seed.

Starting point is 01:12:50 And actually, I mean, they just started that, but they don't think it's going to be as easy as they thought. So that's pretty surprising, right? A company that's done really well in the last two and a half years has gotten to like a pretty significant run rate. So it hasn't been like an endless amount of time in a category that, I mean, may not be sexy now, but I think, you know, we all agree. These are massive markets, Vintech and healthcare. And great founders, technical founders. I mean, literally check every box. And no one was interested because they weren't growing at, you know,

Starting point is 01:13:19 500 X, right? And they weren't in a category that everyone loved. And I mean, I was surprising. I think the founder even said 18 months ago, I would have gotten a series A done. And now, you know, they're kind of just like, oh, we'll raise three million and then, you know, we'll get to a real so bridge. They're just bridging it internal team, internal investors doing the majority of them. They'll get one, one new investor. But that's pretty shocking to me to be doing, you know, three million error in great category. What's the lesson for founders? What's the lesson for founders here? lesson is timing. They should have raised 18 months ago, honestly, because they were kind of the darling and people wanted to give them money and they thought, okay, we want to get from one million

Starting point is 01:13:59 to three. Obviously, you know, that Jesus evaluation and, you know, that was a personal milestone that they wanted to hit. But, you know, timing is everything in this business. So that was a learning. In this environment, if someone wants to give you money, take it. It's sort of the lesson. And if you're not raising, then, you know, sort of figure out how to raise whether it's a, it's a safe, whether it's whatever, but the issue that we have is, as the people who help founders figure out their fundraising strategy, it's never been harder for us to figure out when and how much to raise because of what Jenny is described. We've had a lot of fundraising happen in series and series B, but it's sort of unpredictable, meaning like it happened awesome,

Starting point is 01:14:49 but could we have sort of had the foresight of saying, yes, this will happen? It's easier when the companies sort of raise a growth round. Like today in Numeral, one of our companies is announcing a hundred million dollar kind of series C from inside. I mean, it was clear to us that they were going to raise, we just didn't know the valuation and the actual size. But given the company's progress, which is very, very quickly sort of growing to tens of millions of dollars in revenue, like it's predictable.

Starting point is 01:15:19 The issue for most founders is if you show up with a triple-triple-triple-double-double, which guaranteed you financing three, four years ago, right now, nobody cares. Nobody cares. It's kind of pushing founders to teeter on the edge of profitability, or at least being able to get there. And so that also isn't great. These are seed-stage companies. We want them to invest and grow, and they feel nervous because, you know, the ground is so shaky.

Starting point is 01:15:48 Here's my best advice to founders. I always tell this to founders, Jenny. Are these good terms? Is this a good investor? I'm sorry, is this a good valuation? Is this a good investor? And are these clean terms? Take these three circles and overlap them.

Starting point is 01:16:04 And in the middle, I write these three words, just do it. Good valuation, good investor, clean terms. Just effing, do it. What are we talking about it? What are we optimizing for? And I would like to thank Gemini for making me graphics that I would have paid an illustrator $300 for and waited six weeks. I mean, it's crazy, right?

Starting point is 01:16:26 Like I just made that while you were talking, Jenny. I was like, I need to write a blog post on this or explain this to founders. I literally had this with my friend Ali, a micro one who's been on this week in AI a couple times. One of our, you know, like literally it's my next Uber here in terms of we were able to invest in it in a $12 million valuation. He hits hundreds of millions of dollars in revenue. he's been public about this. He's massively profitable. He's like, yeah, people want to put more money in, but we're not going to take it because we're profitable. And I was like, good terms. He's like, great terms. I'm like, good investor. He's like, great investor.

Starting point is 01:16:58 He's like, yeah, great valuation. He's like, yeah, great valuation. Good investor. No, great investor. Okay. And the terms clean. He's like, yeah, they'll do whatever, you know, like standard paperwork. And if you look in the news, you'll see a news story. He just, I think, announced he raised $100 million at $4 billion, which, is unbelievable and absolutely fantastic congratulations to them. And they were like, had more money potentially coming in. And I'm like, hey, son, hey son, hey son, make hay when the sun is shining. But Ali is young and a force of nature. 25 year old. There he is. Well, thanks. Congratulations. I mean, we own 6% of the company.

Starting point is 01:17:48 It's going to be a fun returner and the syndicate's going to be quite happy. You know, if things keep going according to this. All right, we've got to wrap, guys. You've been amazing. Jenny, you've got a favorite company right now. You want to give a little pluggy poo to and how can founders reach you? You can reach me over email, Jay at everywhere.vc. Pretty easy.

Starting point is 01:18:09 I love it. And what should they include in their email, you know, in terms of like a chart? short, long deal memo? I just want, I just want companies that fit our thesis. So we do all precede. We're one of the last surviving 250K. I think what Jeff did back in 2007. It's what we do.

Starting point is 01:18:29 250K, precede rounds are usually like a million, all B2B across categories. So if you send me something like that that's focused, I'm happy to chat. So you're the last lady standing in this is a very important preseed space. I mean, it's pretty, yeah, I mean, there's like 10 of us left. Trust me, I'm on the boat. I'm on the raft. I'm like, why doesn't anybody like this boat? It's incredible.

Starting point is 01:18:53 Yeah, they hate the boat. They hate this boat. The boat's not moving fast enough. It's not big enough. I'm like, but the fish, look at the view. It's great. Jeff, you got a favorite company you want to give a little pluggy poo to. And can people reach out to you with their deep tech?

Starting point is 01:19:10 Deep tech. Sure. I mean, Uncorg does pretty much. everything. So happy to get people's email, jeff at uncork capital.com. And I'm happy to route to all my partners because as a firm, I invest in deep tech, but we invest in a bunch of different categories. So if you have exceptional founders building something really interesting, new, different in a market, which is going to become monstrous, send it to me and I will get back to you. Plug to Loft Orbital, which just announced a big.

Starting point is 01:19:43 million dollar investment from Abu Dhabi in a constellation that they're building, which will be the largest AI compute constellation in space. They already have more compute in space than any other company that manage missions for others have investing the company and being on the on the board for about seven years now. Loftobel.com, super exciting. Did they start in computing space or did they start in another area? because this has become a new opportunity. They started as shared missions.

Starting point is 01:20:15 So they basically sort of get a bunch of sensors on satellites that are shared across a bunch of customers. And so as opposed to forcing you to launch your own satellites, you just give them your sensor, your camera, your infrared sensor, and they sort of launch it and manage it for you. So they make space look at like a data stream. They hide the complexity of space. and they basically plopped GPUs up there,

Starting point is 01:20:43 and that gives them the ability to take an image, run the AI models, and send you down the information, the action that you have to take, as opposed to waiting for a multi-gigabyte image to actually stream down to the planet. Okay, very good. And Dave, what's exciting in your world and how can venture capitalists?

Starting point is 01:21:09 who need to raise their fifth fund and would like to clear some positions in fund one, two, and three. Not that I'm thinking of anybody with the initials, JC, but, you know, I'm raising my... That's not Jeff Clavier, by the way. To be here?

Starting point is 01:21:23 It's not Jeff Clavier, nor is it Jesus Christ, nor is it Julius Caesar. It could be another JC. But Jeff understands what we're talking about here. We get put into these short lists all the time. Jason, both of you and I are going to be at the sidecar.

Starting point is 01:21:39 SPV summit in New York next Tuesday, and I'm going to be speaking about liquidity vehicles for fund managers, what I call a single asset continuity SPV. Okay. The basic concept is just selling a piece of some of your winners into the SPVs that you might be using to raise parata so that you can generate DPI for your own fund and for your LPs. Interesting. And you would manage that or you would be the LP in that?

Starting point is 01:22:06 Well, we might be helping to price those. but I would say that a simple way to do this is after a series B or after a series C within three to six months, you can use the post money price on that round as a legitimate objective transfer price for the asset from your initiating fund to an SPV. If your LPs want liquidity, they can keep that. If they want to roll with the position, they can roll into the new vehicle. But it's a way for fund managers to generate DPI probably in years five to eight. And the important part of this is they're not necessarily selling to somebody else. and maybe having a disadvantage of his price

Starting point is 01:22:41 or missing out on the upside, they can lock in some of the gains in their initiating fund to generate DPI, but they can still manage the position in the SPV going forward. We'll talk more about that at the Sightcar Summit coming up on Tuesday. I will be there,

Starting point is 01:22:57 and I'll be going from onstage to then talking to you. Talking. Well, no, I'm going to be in the audience for this. I want to, yeah, I want to ask some questions. It's so clever, and Dave had to explain it to me like three times for me to like you understand it. So I will also be in the audience trying to absorb it again. But it is a very smart idea. So. The other thing I want to be pitching to people,

Starting point is 01:23:18 especially for anybody who has 2026 taxable events is donor advised funds. And Jacob, if you could show practical VC.com slash DAF or DAF. We're actually going to be doing two education webinars on this one tomorrow morning and one Monday morning, which you can sign up for on our site. And this is really about how to donate illiquid assets, which could be stock in a startup company or GP or LP or LP interest in a fund, even carried interest in a fund. Instead of, you know, donating your cash, you should really be donating the assets that you haven't already been taxed on. And not all donor advice funds accept these assets, but UI Charitable, which we work with

Starting point is 01:24:01 out of Utah does. So you can get a third-party fair market value assessment for. or assets that you're holding, whether those are startup company equity or fund interests, you can donate them to your own donor-advised fund, and you get a tax deduction action immediately upon donating, whether or not they're liquid or could be liquid in the future. And you can use that to deduct against your taxes for this year and up to five years in the future. This is really important when you have a liquidity event.

Starting point is 01:24:29 So for any employees who are selling secondary, instead of selling a million dollars worth of stock and paying three to five hundred thousand dollars in taxes, you can donate a portion to your DAF, and instead of sending money to the IRS, you're sending it to your own philanthropic pocket. You can also do that with assets that are illiquid, and let's not say that any of us are sitting on busted unigorns from five years ago, but you might be able to get a favorable evaluation for that asset, which is highly illiquid. Donate that to your own donor advice fund and cut your tax bill, even if you don't have a liquidity of any. in the current year. But important to get that stuff started and do it now, you need to get

Starting point is 01:25:06 that stuff done by December. We'll probably take you about 30 days to get fair market value opinion and do the donation process. Amazing. I just want to give a shout out to tax GPT. This is a surging company from our pre-accelerate. We call a founder university. We now do founder university in the U.S. and globally twice a year. We do it twice a year in Saudi. we'll be doing our next edition in November. And twice a year in Japan, founding university is pre-accelerated. You just need to have a team. You don't have to be even incorporated yet.

Starting point is 01:25:41 Hopefully you get incorporated while in the program. And Founding University is a 12-week course, just helping you get your product finished to market. We invest in some of the graduates, and one of them was tax GPT. They went on to our accelerator. We invested, and we invested. Then they got into Y Combinator, so they went to Founder University launch. and then why accommodate all in like the same 12, 18 month period.

Starting point is 01:26:05 And they've now hit tens of millions of dollars in revenue, thousands of tax and accounting firms using it, and it is a fantastic startup than I am enamored with. So check out tax, GPT, if you're an accountant. And yeah, if you want to angel invest alongside of our firm, our best companies will sometimes syndicate at the syndicate.com. You have to apply and you have to do an onboarding call now. We have too many members.

Starting point is 01:26:37 I say this with peace and love. There are too many members, so you cannot invest with the syndicate out of the gate. You have to do an onboarding call, then you put on a wait list, and then we invite you to join. So we get that process started now. We have over 4,000 active members and the syndicate.com. We do early stage and we do late stage. Barbell Strategy. This has been this week in Venture Capital.

Starting point is 01:26:59 we'll see you all next time bye bye

This Week in Startups - VCs Would Bet on Open-Source AI Over OpenAI and Anthropic | E2341

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.