Walk & Talk Episode: Semil Shah on AI Superteams, Meta’s Bold Moves, and Apple’s Missed Shots

26:06Super amazing guys. Their NYC raised maybe less than$2 million. Then they raised a Series A from Albert at USV. And then within 18 months, they were acquired by Google for almost$500 million. Great return for everybody, especially in terms of the timeline. Today, obviously, that doesn't move the needle for a large fund. that seed, they would have raised four or five in YC, not one to two. And the Series A round, had they gotten there, would have been at least double, right? So the question would be, would that have been a billion-dollar acquisition right now? I don't really know. And so that's really the debate on the field.

26:41And no one really knows because people see a lot of activity right now in AI and they see a lot of private marks. But if you think about Figma, we go back to our conversation back at the lighthouse, you know, that took for most people over 12 years to realize. And what happened in the 12 years? Trump, COVID, Ukraine war, you know, interest rate regime over. Trump back again. Right. A lot happens. So it sounds like your options are either you stick with the model, like you said, or what do you do? You get bigger? You go from being a seed fund to being a seed plus A fund? Well, it's a catch-22. Right.

27:18So a lot of people decide to go bigger and they'll say, well, they'll tell their LPs, hey, LPs, you want us to get ownership 10 % minimum in the seed rounds. What used to cost one, one and a half, two now costs maybe three, three and a half, four. And to get the number of shots on goal, because the risk is the same, right? It's going to cost more money. So then your hurdle goes up. Your GP commit goes up. Everything goes up. Right. Yeah. You're playing at a bigger poker table. Right. It's a totally different game. Then when you start trying to do 7 million, 8 million seed check rounds and stuff, guess who's looking around the corner?

27:53Sequoia, Excel, you know, and these funds are not dumb. Like they know how to move in and out of seed very well. it seems like if you're an early stage founder, there are more and more ways to get less and less dilutive capital than ever before. You can raise from a traditional seed fund like Haystack, you can go to YC, you can go to HF0, you can go to pick your accelerator SPC. And like I said, it's not that dilutive. And so you have options if you're a founder, right? Now, that means that the seed funds like Haystack, but also the bigger funds that you just mentioned and folks like Lightspeed.

28:32Oftentimes, if you want to go and compete with those funds, you're not getting the classic 25 % ownership. And so if a large fund, again, like the ones you mentioned, comes in at the seed, gets that lower ownership, they're going to be hungry for more ownership in the next round. And so they're going to do the A the moment they can, which means they're boxing everyone else out. which then motivates the bigger funds to want to be in seed even more so they don't miss the opportunity. So it almost feels like, if I can sum it up, you're moving away from a world in which you do the seed round, somebody else does the A round, somebody else does the B.

29:11And it's like, we got to go as early as possible. And then every round, take every chance we can to buy of ownership. Yeah. I mean, that's the real estate plot argument I use, which is like, let's say the next Zoom of this era or, you know, a cursor, right? Yeah, cursor. In cursor, typically you have two to three shots at the Apple for like a venture style return. And the larger funds, I think, scaled up over the last few years because they were basically saying, hey, look, there's going to be more and more of these things. And we can't get it on the first or second shot of every Apple, Zoom and Sequoia being a kind of poster example.

29:48So then they say, OK, I'll get a couple of shots at the Apple. and it doesn't matter if I got in a cursor at 40 million posts, 100 million posts or 400 million posts. It doesn't matter. Did I get in or not? I have three strikes. And that's what I think is happening. Now, the other side of the debate is that I still believe you'll have small funds with relatively smaller sub-billion dollar outcomes that move the needle for those investors and those funds and those managers. I still think that will happen. Explain that. I think there still could be a Firebase. You could have a$40 million fund, invest$1 million in the next Firebase.

30:24It gets turned around in two years to Google, 480, go home. How do you think this dynamic, so we're talking so much about the game on the field for VCs, how these dynamics are affecting VC, VC strategy. How do you think this impacts the founders and the incentives of the founders? It impacts the founders a lot of ways. So the number one way is like, I think there are two core issues that the founders face. One is like learning to over-optimize how and when to raise capital and from whom to the point where you're not really building a fan base. And so, you know, is the cheapest item in the store the right item in the store to get for the job to do?

31:08So that's one. And I think the second thing is the invitation of signal risk. There's always a debate about signal risk. But I think if you go into a big fund too early, then some founders who are more savvy will split the seed amongst two firms and sort of create that like auction format and like between the two. But look, it ultimately is up to the founders to perform where they get past signal risk. So I think those two are the key issues. And then like derivative issues for founders, I think is like raising too much money ahead of knowing what to do with it, because I do think that dampens creativity and dampens like internal angst and momentum.

31:45Yeah. And then the other issue is I don't know how you recruit in this market with Zuckerberg, you know, doing these kind of deals and other people paying through the nose for these people. So really, the money is cheaper today and finding the right aggregation of talent is much more difficult. If you're an early stage founder that comes from maybe the more traditional startup world of raise a little bit of money, prove something, raise a little bit more, add a slightly valuation, prove something else, and you go through sort of the steps one by one. I mean, can that still work? Or do you have to be in the hyper aggressive, every raise is a mega raise, light money on fire to grow as quickly as possible?

32:28I think it just depends on where you are. So to me, these are company design questions. How do you want to design your company and how you finance your company as part of the company design? So we just were the second sort of in command in the investment. There was a lead investor we know. The founder, we thought, was going to raise$6 to$8 million because of his profile and background. He only wanted two. And he was like, you know, I already have a team ready to go. I know what they're going to make. And I want to hold a feet to the fire. And I had coffee with him in New York a couple of days ago.

32:59And I said, hey, I expected the round to be totally different. And he said, no, I just, I want to have something really in hand so I can go out. And so what does that mean? One is that there's still people out there like that who want to design the company this way. Two, I think it can work, but you need somebody who's moving very fast and you need a natural fundraiser because that person has to break through the noise and say, hey, I got to work with this person. I mean, think about the deals you've done that are earlier series A's. you've basically started with in your message, I want to work with this person.

33:33That is a, like, not many people meet that bar. So when we're looking at seed, it's not just price. We really have to assess, can somebody meet that bar? Right. So a higher bar. Yeah. We even go deeper and say like, okay, if we're in conversation with this person, no props, no decks, no models, we're just talking to them. Can we envision them hanging with one of our Series A friends for like two hours unstructured, but coherent? Yeah. You know, and most people don't pass that test. Yeah. It's interesting you say that because I feel like with all the different sources of capital and the rush to get in seed because you're going to get boxed out.

34:11Oh. It almost feels like the gravitational force would lower the bar. Well, again, I've been doing this 13 years. And I believe that the first raises are not so much based on metrics or these things. They're based on what is it like to work with this person? How fast is this person moving? How does this person compare to all the people you meet? Yeah. Yeah. And that's what I tell entrepreneurs. They're like, what do I need for a series A? And I always say it's a Steve Martin quote, which is, you know, be so good they can't ignore you. Yeah. What advice would you give to a founder that is getting just constant inbound from late stage funds right now because of that?

34:54I'll tell you the same thing I've said for 10 years, which is record all the inbound, the date, the name of the person, who, and the spreadsheet, and consult your investors and advisors on who to talk to and when. And we've seen plenty of entrepreneurs get sucked into a lot of meetings. And these people are good at sucking you into meetings. So, yeah. But it doesn't necessarily mean anything, right? It doesn't mean anything. Yeah. So switching it up a little bit, there's a topic I've been following quite closely. I think you have as well. And that is this, I almost want to say this battle that's unfolding between sort of AI agent companies like Perplexity and CDNs like Cloudflare.

35:38Cloudflare has gone out to the market. Instead of just being a CDN, they're saying to the world, hey, actually, now we want to help you sort of monitor your traffic. And we want to make sure it's not being scraped by bots. We want to make sure you're not losing traffic. So we're going to automatically turn on tools that make it harder for those agents to get your content. We're going to just do it by default. We're not even going to sort of ask your permission in some cases. Some cases we will, some cases we won't. And the benefit is that now you're not losing traffic that you would normally get by human.

36:12You're not losing it to an agent. The flip side of it is users want these agents. Customers want these agents. Customers want to use perplexity. They want to use ChatGPT. And maybe just the way the content gets browsed and consumed on the internet is changing. And sort of free markets and sort of the free behaviors of the actors on the internet just needs to play out. Yeah. I'm having a really hard time wrapping my head around this one because it's so nuanced and it's all new. So the question is that the scale of these technology companies that they have and the power that these agents could do, what's the equivalent of just having a bot run there?

36:48Yeah. The part about Cloudflare is interesting because this goes back to Bryce Roberts' old post called Defaults Matter. So I think that was interesting when Matt was talking about the Cloudflare CEO, where he was like, hey, we just turned it on as a default setting. You can go in and turn it off. Now, how many people can go into their CDN and turn it off? Yeah, it's going to be a low percentage. Right. And so it'll be interesting to see because I guess there are like 15 % of internet traffic, right, what the other CDNs do. And I don't know what the model is. I mean, obviously, the model that people can envision, which is like, oh, the model provider will pay some kind of royalty back to the content source.

37:32But even if that happens, it's going to be way lower than what they would have gotten otherwise through subscriptions and other things. It seems like this is going to be a new battle between the publisher layer and the agent layer with the CDN in the middle playing referee. And the CDN, in this case, CloudFires, inserted itself in the middle with a commercial interest. So, yeah, I mean, the model, if you think about it, the model that web publishers are used to is they put their content out there. They rely on organic or inorganic sources of traffic to get human traffic, and then they monetize that human traffic via ads or subscriptions.

38:05What the agents are doing is they're sort of grabbing that content and putting it on their sites, and they're crediting it, and they're providing a link that people can click through to. But what the data that Cloudflare and companies like Tolbit are showing is that the click through rate is extremely low. And these AI agents, in many cases, are actually really significantly damaging. the traffic. And so you could kind of see the other side of this. I know a lot of people are sort of upset with Cloudflare right now, but you can also kind of see the other side. Oh, absolutely. We're going to try to protect your business model.

38:37Yeah, yeah, yeah. I mean, I think it's an interesting battle. I think, look, the Cloudflare team is amazing. I mean, they've been way ahead of the game in a lot of ways. So it's not a surprise that they're inserting themselves here. And there's no rules. The way I think about it, though, is that there's a lot of bad content on the internet. And the downside of letting everyone read, write, publish is that you get a lot of noise. And so maybe the economics of just doing stuff on the web is changing. Whatever happened to newspapers now may happen to content creators who don't modernize with the times.

39:13So then you have little things like, well, there's AI search optimization. Can you do that? I mean, we'll see if that field comes along, but All these companies now control that access point. And it's not really clear what they're going to do. And they seem to be moving more towards subscriptions and not doing ads. That's true. Right? Yeah. So I wonder if the ultimate outcome of this is one way or another, ads are in trouble. Although, actually, sorry, let me take that back. Because it seems inevitable to me that these AI agents are going to have ads. ChatGBT is going to have ads. Oh, 100%. 100%.

39:45At the scale, I mean, it would be silly not to. Yeah. It's really fascinating. I mean, the backlash against Cloudflare has been strong. Do you think there's any chance Cloudflare just backs out of this strategy and says, oh, never mind? Normally, I would say that it's possible, but Matthew strikes me as someone who I don't know him. Me neither. He doesn't do things willy-nilly. Seems headstrong. Yes. Determined. And, you know, been at this for a really long time. Yeah. It'll be interesting to see if the other CDNs jump into the game and create this sort of United Front, the Akamai's, the Fastly's, et cetera.

40:20I know a number of them have parted with it. But, you know, I think like at like a base foundation level, like if you wanted to start your travel blog or your surfing blog here at Ditch Plains, like the internet enables you to write and do everything you want and create videos and have a subscription service. But you're not entitled to that in perpetuity. Yeah. So the distribution. No. And like MG, you know, our M.G. Siegler, he would cite Gruber's blog or Ben Thompson cites M.G.'s blog, and they'll put the quote in there and they'll provide a link. But you can write on top of other content. You can remix content.

40:55That is a good point. This is remixing on steroids. This is remixing on steroids. And so if you take your travel blog or your surf blog and you go back into an inner real life format, you have a storefront, right? Now, no one's taking the opportunity away for you to have your storefront downtown. The problem is with the foot traffic and no one's guiding your, you know, you don't control the foot traffic. You don't control the foot traffic. And we did take it for granted in a Google first world where it's the front door to everyone's web browsing experience, right? It's just, it's just guaranteed search traffic is going to be there, but it's not really going to be the case for much longer.

41:33It seems like. No. And that's why I think you'll see things like media driving people to sites. I mean, how many people are on Instagram that are creators that are saying, I'm hosting a meetup, click on my link. So you have to engage them where they are. That's going to be in a perplexity, in an Instagram feed, in a WhatsApp channel. And then you got to move them over here. You got to be so good, they can't ignore you. And they move over to where you are. So we got to meet our friend for lunch soon. So why don't we get rolling over towards Derriere's and we'll pick up the conversation there. Let's do it.

42:24All right.

42:28So we're obviously here on Long Island. We talked about it earlier. And it feels like podcasts are getting more experimental. It feels like media is changing. And this is always a topic I like to discuss with you because in addition to being a great seed investor, you're a fan of media. You're a content creator. You've been blogging on the internet for, I don't know, decades at this point. What do you observe is happening to podcasts right now? Again, we talked about TBPN Live. We talked about Bryce. There's interesting things coming out of Colossus and Invest Like the Best. It's changing. Yeah, yeah.

43:09Yeah, I think this is a topic we should always check in on because it may be changing every six months. Yeah. Right now, like you pointed out, the pace of acceleration of change is very noticeable because things are working. Yeah. So I think a couple of things come to mind is one of probably a few episodes ago, we just talked about YouTube killing podcasts, right? Which essentially means videos eating podcasts, which means that's the point of creation for most people. And then that's splintering into live format, right? live audience format and taped live and then broadcast and then there's an archive and then there's like richer production like i mean i'd love to hear where you're going to take this podcast or you know what you've observed and bryce doing his or other people doing that where they're saying okay maybe i want to do fewer of these things and uh do it longer form and in more interesting locations and formats because we're kind of out of the era of seeing everyone on their headphones and in their home office, you know, in front of a, or in their home bedroom or whatever, you know, it's, it's a little bit too static.

44:10I totally agree. And I think it actually relates to something we talked about earlier, going back to the Figma conversation, the premise for Figma was that design is an enduring differentiator, especially in a sea of mediocre applications, the things that are beautiful and that people love to use stand out. I kind of think that's what's happening to podcasts right now. The things that are beautiful and people want to watch, they want to put their eyes on, they want to immerse their sense of it. But remember, tens of millions of people, every time a Joe Rogan or Lex Friedman pod come out, have it on their big screen TV.

44:42Oh, totally. Making pizza, hanging out with their friends, reading a book or doing a crossword. And there's nothing going on except two people talking. Totally. Now, everyone's not Joe or Lex. And so the question is, how do you create a richer content experience 100 % for that next person. And I think all the things we're talking about, whether it's Kill Tony format or all the way to what other investors are doing, or my friend Jack Altman has been doing this kind of style pod, but in his office. And people are experimenting with all these new things and there's tons of space open for it. Yeah, my point is, unless you're Joe Rogan or Lex, you can't really stand out with that format.

45:22You just can't catch up to their distribution. Unless your guests are amazing, right? So Jack has an ability to get guests in the tech world or unless your angle of attack is unique. So you may have you may have really good people on, but you're an expert in a certain way, you know, like a like a Tim Ferriss. Right. And then you sort of go. But yeah, I think people want to see people in different places. You think about why the All In podcast is so great. You know, one of my colleagues just mentioned people in her generation. They want to watch like friends had a resurgence. They want to see people who are friends hanging out and working.

45:55And that's sort of one version of it. Yeah. I also think we're going to start to see other formats and modalities creep into podcasts and videos. As an example, we mentioned Colossus and Invest Like the Best. So Patrick, you know, he's been doing these great long form interviews forever. And now I'm pretty sure there's like a zine component. There's almost like a magazine component where he'll do a long form interview with Neil Mehta or Romton from Abstract. And then there's a written long form interview and it's got beautiful photography. And I think there's actually like a physical copy of this thing.

46:32I think that's a pretty interesting strategy. Yeah, I mean, if you think about even just selling that for 10 bucks, someone who's an investor or wants to be a VC or coming into VC can read about from Neil Mehta or read about the Josh Kushner interview in like a more luxurious format. even if it's 10 bucks, you know, his physical item, whatever he would sell it for, is great because you would normally buy those books at the bookstore. They just don't exist at the bookstore. Yeah. It reminds me of kind of like music fandom and pop culture. The first thing I thought of with the zine, with the Colossus thing, is it actually, speaking of Montauk, reminds me of Whalebone.

47:09You know, Whalebone is like this surf brand, which has merch, but they also make this magazine. And like, I actually don't even know what people do with the magazine. It's just nice to look at just like beautiful photography well airbnb tried to do a magazine oh that's right which i thought was actually a really great idea and i guess didn't have what was in the magazine i think it was like a travel and inspiration magazine it's a cool you know like super you know for a company like that it's not a huge line item but yeah i mean i think uh famous andreessen line right we're just a media company that happens to invest in startups yeah and so a lot of this thing is about media a lot about is about attention um and we're just talking about tech in bc but this is much broader.

47:48I mean, I know we follow a lot of musicians online who are maybe like truly indie musicians are never going to make it. And then, you know, I follow a lot of cooks online, you know, and try to just watch what they're doing. And like just the innovation in the individual cooking blogger, the way they create videos or cut videos is, I mean, it just blows me away. And I'm sure that's true in like 50 other niche verticals. Right. So they're just getting more artistic and creative with the video or what are they doing? I mean, I'm going to sound like a complete noob saying this, but like there are multiple editing tools that these folks can use online and pay whatever four bucks a month for.

48:26Yeah. And they can remix content, make it more interesting, more catchy, right? Because you can look at making, how to make the perfect grilled cheese with bacon, you know, out there. But if you get someone who's like really catchy format, you know, those people aggregate a lot of audience members. Yeah. It's just, I guess, again, something artistic, something creative, beautiful is another way to stand. Actually, you know what it reminds me of? The Instagram video you sent me two days ago about the Mets telecast on SNY, the New York Mets. Whoever is, I don't know, directing that segment is doing really, really interesting things with the camera work and the transitions.

49:05It looks like classic films. And I certainly noticed it as a Mets fan, but I didn't really realize what was happening. It's a very, very intentional choice by this guy. I don't have his name. I should have brought it with me. Yeah. I mean, I think that analogy can stretch out to a number of different places. Like, I think what that reminds me of when you were saying it, we're like, you know, when we were probably growing up and there was just terrestrial TV, you would have like QVC and online shopping networks on TV, TV shopping networks. And you would call and order whatever set of glasses. Now, when you see like the live e-commerce stuff especially coming from asia they've got it looks like a horse race board right all the stats and everything on it so the overlay on top of video or live video you know is just happening with captions and all this stuff we take for granted now and that stuff's baked into whatnot i think right yeah so you can do all sorts of things and this guy who's doing the mets broadcast is like more film inspired imagine someone who's anime inspired or imagine someone who's fill in the blank inspired, you can take it in a lot of different directions.

50:08And AI is obviously gonna make all this stuff easier and easier, right? Easier and easier. Tap a button and get the Fellini transition. Really the way it should go from like a content perspective is like you should get your raw feed that you paid for and you can upgrade for other kinds of feeds, right? And it could just be a very small upgrade. You pay the one person who's creating the thing. I mean, it would be massive for them. Not to make everything about tech and startups. We talked about actually how it impacts VC a little bit with, you know, invest like the best and things like that. But does this find its way into startup building in some way?

50:41I don't think so. I think like for folks that have consumer facing or like small business facing ones, media is important at the right time. Yeah. But if Airbnb couldn't launch a travel magazine when they had oodles of cash, you know, I think it's one of those traps, one of those startup traps. I think what's more interesting is like the cooking blogger I may follow or the interesting percussionist you may follow on Instagram. What can they do if they get a little bit of an audience now in terms of creating multimedia like a zine you mentioned or something? Like I'm sure you followed a new artist you met on Spotify and you go to their merch store and they sell a record.

51:23And you may not be on record player, but you buy the record as art. Yeah. Right. So a lot of things are coming back and, you know, wouldn't be surprised. I do think it probably has an impact, maybe somewhat to your point, on startup brands and the importance of a great brand. You know, we were talking a lot about how a lot of other companies want to partner with this company to be able to put that company's brand next to their brand. And they want to be associated with this company. And that's why they want to partner with it. One of the first VCs I ever met in Palo Alto, who's still doing it, been doing it probably 30 years, said association is very important in our business.

52:03Yeah, in all aspects of it, not just the startups, but obviously the VCs who you choose to raise money from, your hires, we're seeing this with Zuck and Meta. Obviously, they want to have the big, splashy hires, not just in talent, but also in name. These are just the tech VC versions of collabs. Right, right, right. Yeah, super interesting. Speaking of things coming back, I have actually noticed that like these vintage band t-shirts and merch. Oh, yeah. They're so expensive. There's a vintage store actually in New York right near the Lightspeed office. And I walked in there and I went to go grab this Nine Inch Nails shirt.

52:41And I put it on the counter and the clerk said, oh, it's$3.50. And I said,$3.50? It's like a 30-year-old shirt. It's falling apart. He's like, no,$350. so it is crazy the premium that people will put on great iconic brand and artistry and imagery I wonder what that will mean about the longevity of some of these shows and podcasts we're talking about I had this uh feeling that like the best shows or even think of like Mad Men or Breaking Bad within year three to five you're sort of pushing the edge of what's possible and I found that same cadence to be true in like group chats I'm in which are kind of like their own podcast series or TV series.

53:19And those kind of last about three to five years. And so I think the rate of decay in a space like this, where it's so easy and there's so many people and the tools - It's going to be faster. It's going to be way faster than that because to match a production and writing really Mad Men was like a visual plus sort of text-based story. You're not going to have someone randomly AI do that no but i think for a lot of other things or podcasts or what people are trying to do not to call it out i just think the next person there is going to be there try it copy it go one step beyond yeah there's really no rules remix culture on steroids with ai are there any brands in ai right now that you think are doing a really good job and we should revisit this question in a few months because per your decay point yeah i'd be curious to know you're still impressed by the brands a few months from now.

54:10Well, I don't think anything I'm going to say is unusual, but I think like the leaders in their categories have done a great job. So like, I think perplexity has done a great job internally in our insular world of tech and VCN startups have basically grabbed the mantle even before it was given to them and said, we're going to do this, which is open AI, it's a bold prediction to just own the browser. And I think externally involved enough investors involved enough. people did enough press in their own sort of writing that like they sort of claim that headspace. Yeah. For people who are looking to bounce from Google search, they provided a home.

54:48So I think they did a great job. I think cursor has done an amazing job just in the branding of the name and like what it means to be able to like write from there. And, you know, that's really a unique story. That's a good call. I think they've done a great, great job at branding. I think OpenAI, even though it's very clunky branding, it's like, you know, you have the same number of models you can choose from as you have like iPhone. Oh, 4.0 and yeah. But I think that their word of mouth amongst a wider variety of people, it just speaks to how powerful the AI prompting is. Not the brand, the technology.

55:29Which then conversely makes the brand more interesting, right? The product is driving it. Those are the ones that really stand out to me that have like owned the consumer mindshare space of like create with cursor, search in a new way with perplexity. You know, those are the things that are driving it right now. And I think to a smaller degree, you have like a mid journey, right? And you have like a Suno, I know a company you're involved with. but I don't think it's broken. Crossed over to mainstream. Well, Samil, this has been awesome. Thank you for doing it. Thank you. Great to do an experiment and take the pod out into the wild.

56:09Maybe we'll do it again and always looking forward to the next one. First of many.

56:17For those watching or listening, thank you so much for tuning in for this special edition of the podcast. We hope to do it again soon. See you then.

56:29you