Latest / Newcomer Pod / The AI Startup Fantasy Draft
Transcript
- 0:04Welcome to the AI draft on the Cerebral Valley Podcast.
- 0:08I'm Eric newcomer. We've got a probably my favorite
- 0:10episode of the year coming up for you.
- 0:12We're about to pick what we think will be the most valuable
- 0:17AI companies in four years from now.
- 0:19This is a follow up from our draft a year ago, but we'll
- 0:23catch you up on everything that happened last year.
- 0:25We'll be picking a bunch of new companies.
- 0:28I'm here with the James Will Sturman.
- 0:30Hey, Eric, happy to be here. Excited for the draft?
- 0:34James is the the brain behind the draft.
- 0:37He'll explain it in a second. And Max Child.
- 0:40Hello. Hello.
- 0:43Super excited. We a very contentious game of
- 0:47forecasting the top companies in AI ahead of the Cerebral Valley
- 0:52AI Summit on November 20th in San Francisco.
- 0:55This is our way of thinking about all the top companies in
- 0:58AI before we host the AI Insider event of the year.
- 1:03All that said, James, do you want to tell everybody the rules
- 1:07so they know why and how we're operating here?
- 1:11Buckle in. It's a complicated set of rules,
- 1:14so I will try to go through them efficiently.
- 1:18If you recall from last year, some of you may have listened to
- 1:22that episode, we drafted teams. As Eric said, we drafted the top
- 1:27private companies in AI. We're all looking to create a
- 1:32team of companies that by the end of 2028, will have the
- 1:38highest value on either sort of a private market valuation, an
- 1:42exit or an IPO. So just to caveat a few
- 1:46different rules there, last year we each drafted 5 teams.
- 1:50And in order to get the 1st pick, we agreed that you'd have
- 1:54to take a big handicap because there was such an outlier pick
- 1:57with Open AI last year. So we agreed that Eric, by
- 2:01drafting Open AI first, would have to take a $75 billion
- 2:06handicap. Auction for that so I I paid the
- 2:09most the right. How are you feeling about that
- 2:12decision in retrospect? Excellent genius, I'm dominating
- 2:16right now. Yeah, Eric's team's looking
- 2:18really good. We'll get into that in a minute.
- 2:20A couple other rules. You have to be on this list at
- 2:23all. To be a one of the companies
- 2:25that you can draft, you have to have raised over $100 million.
- 2:29Fortunately, a lot of AI companies have raised well
- 2:32beyond $100 million. It's a long list.
- 2:34Very short amount of many. Companies we have never heard
- 2:37of. It's a quite, quite a long list.
- 2:40Many of them I haven't heard of, heard from since, but they they,
- 2:45they have raised over $100 million.
- 2:47A couple other rules. You have to have had a core use
- 2:50of generative AI or other generative AI infrastructure use
- 2:55cases within your product. However, we're excluding certain
- 3:00industries as well. So if you're purely a bio
- 3:02healthcare industry, AI provider, a defense provider,
- 3:06silicon chips infrastructure provider, you know, such as core
- 3:10Weave or Lambda Labs, we're also excluding you from this draft
- 3:13robotics and a VS as well. We're not drafting Waymo, we're
- 3:17not drafting crews and other AV companies.
- 3:20And we're also excluding a couple companies that are just
- 3:22purely based in China. We don't seem to know much about
- 3:25them and and that makes it hard to draft them.
- 3:29And then, yeah, you have to have been founded within the last 10
- 3:33years. So before, you know, excluding a
- 3:35couple companies that maybe have a generative AI story, but they
- 3:39were founded before 2013 or 2014.
- 3:41And then lastly, this year there will be a way to add and drop
- 3:45companies. So you can actually decide in
- 3:48advance of this podcast recording whether you want to
- 3:51drop one of your companies of your team.
- 3:53You can pick up new companies off of waivers.
- 3:56If you've ever played fantasy football, you'll be familiar
- 3:59with that concept. And then we also will extend the
- 4:03teams a little bit this year. So we're keeping our teams from
- 4:06last year, but we're adding two slots each to the team.
- 4:09So you'll have opportunity to at least draft two more companies
- 4:12for your team. How's that sound, guys?
- 4:14Great, great. I emphasize, and I think you
- 4:16said this, but like the goal here is to get the highest total
- 4:21sort of market cap on your team. It's not like a return on your
- 4:24investment. So there's definitely an
- 4:26incentive to buy highly marked companies if you think you know
- 4:30there's still room to grow. But even if there isn't, you
- 4:33know, they may, it may be strategic to pick a maximally
- 4:36valued company. The other thing I want to say is
- 4:38play along at home. Comment on the post with your
- 4:42picks. Obviously you have the advantage
- 4:43of not competing with anyone else, but there are still so
- 4:46many companies that it's a challenge to come up with your
- 4:50What's going to be the total number of companies we each have
- 4:53by the end of this, James? Seven.
- 4:55Yeah, seven each by the end of this.
- 4:57Yeah, so pick your 7 and let us know which companies you draft.
- 5:02Or let us know, you know, what were our genius picks and what
- 5:05were our terrible picks? Who are you reading for,
- 5:09Allison? All genius.
- 5:14Newcomer subscribers are probably rooting for Newcomer.
- 5:16Are they? I don't know.
- 5:19Maybe they're all rooting against newcomer.
- 5:21You never know, could be a hate watch.
- 5:24Revenge for scooping their fundraise.
- 5:26What else do we need to cover? Anything else?
- 5:29All right, so a lot of rules, but we know there are a lot of
- 5:33nerds on here who want to make sure, understand the strategy
- 5:36behind it and follow along very closely.
- 5:39So those are the rules. And now we get to the fun part
- 5:42and make our picks. Give it a listen.
- 5:44This episode is presented by Brex, the financial stack that
- 5:48founders and VCs can truly bank on.
- 5:51Imagine what your founders could do with their runway if they had
- 5:54a banking solution that had no minimums, no transaction fees,
- 5:58and 20 times the standard FDIC protection.
- 6:01Plus they could earn an industry leading yield while maintaining
- 6:05access to funds whenever needed. Brex simplifies financial
- 6:09services for startups so they can focus on building.
- 6:12Connect your portfolio to the financial stack that one in
- 6:15three US venture backed startups already use.
- 6:18Check out brex.com/banking Dash solutions.
- 6:26The draft started out strong with a big pick from Eric Open
- 6:30AI, but he had to take a handicap on that so he had to
- 6:35sacrifice 7 to. Guarantee that I would have the
- 6:37right to pick open AI, and I've been extremely validated by that
- 6:41already. Yeah.
- 6:43So you took a $75 billion handicap off of your own total
- 6:48score. And by the way, this draft ends
- 6:50on in November 2028. However, you, Eric, have also
- 6:55had some exits or semi exits, both inflection and character AI
- 7:00semi exited. We've decided to award you their
- 7:03valuations at the time of exit. Do you want to talk about that
- 7:07nuance for the listeners? A key subplot of this year has
- 7:11been sort of the licensing deals Aqua hire somewhat exit.
- 7:16Basically, leadership at a company gets bought shareholder,
- 7:22you know, the acquirer sort of acquirer pays the company that
- 7:25start up a big licensing fee, then the startup uses that money
- 7:29to buy out the share, the preferred shareholders,
- 7:32basically wiping the cap table. So it's a weird, the company
- 7:35still exists, but the investors got paid out and the
- 7:39leadership's gone. And so you know, I think the
- 7:42logic we came to agree on was that these were effectively
- 7:45exits for the investors and therefore you should be able to
- 7:48cash out. So the the licensing deals
- 7:52character which I think we're putting at 2.5 billion
- 7:59inflection. Around 1:00. .4 what James you
- 8:02had adept. Yeah, at around a billion.
- 8:05Yeah. And tech, I don't.
- 8:08I guess in the abstract you could rebuy those companies,
- 8:11though. We've also put in the
- 8:12requirement that companies need to have raised $100 million.
- 8:15Are we allowing rebuys of these? I'm never going to rebuy any of
- 8:17them to. Be clear.
- 8:18I don't think we should. I don't think we should allow
- 8:21rebuys because like you said to us, the, I believe reported that
- 8:26the cap tables were kind of completely wiped out.
- 8:28So these companies haven't, the new shells of the former
- 8:31companies haven't really raised $100 million yet.
- 8:33So they're no longer part of the draft.
- 8:36Your other companies are doing quite well.
- 8:38So yeah, Open AI reportedly raised $157 billion evaluation.
- 8:43Glean raised an E round at a $4.5 billion evaluation.
- 8:49Mistral also raised the B round around $6 billion.
- 8:55So your total including the $75 billion handicap is roughly
- 9:00around 100 billion right now, 96.4 So you are way ahead of Max
- 9:06and I Max had Data Bricks, Pine Cone, Cohere, Modular games, and
- 9:11Imbue. Using to value Data Bricks by
- 9:15because it hasn't raised a new round since.
- 9:17But like Data bricks, is an extremely valuable company, so
- 9:20Max is worth at least half of what I am by.
- 9:23There's no way. So.
- 9:26Thank you. Thank you, Eric.
- 9:27I appreciate the defense of my honor here because the
- 9:30spreadsheet says my total value is 5 compared to your 96.
- 9:34So yeah, it's like I was like, is this just going to be a dump
- 9:39on Maxfest? Because database still looks
- 9:42pretty good to me, Yeah. You just, you, you need to get,
- 9:44you need to get a new fundraise to get to get points on the
- 9:47board. But I agree you're still still
- 9:49holding strong. But your only points on the
- 9:52board would be Cohere, which raised the Series D and came in
- 9:57at about 5.5 billion. And then my own team, I have
- 10:02Hugging Face, Anthropic AI 21 Labs, Replit, and Adept.
- 10:07As we said, Adept kind of exited at that 1 billion mark.
- 10:10And then Anthropic also did a fundraise in the last year that
- 10:15valued it at a 18.4 billion post money valuation, putting my
- 10:20current score around 19.4 billion.
- 10:24So yeah, I still have some, I think. 30 to 40 so you'll you'll
- 10:28get a boost. Yeah, I was gonna say James and
- 10:31I are kind of in this the same category of like we're around 40
- 10:34probably because I think data bricks is probably around 40 and
- 10:37in tropics probably around 40. Yeah.
- 10:39And so Eric is crushing us in 90 for sure.
- 10:42Yeah, yeah, sure. My my roster is pretty weak.
- 10:45I went with the stars and scrubs strategy, except I only have one
- 10:48star and that is data Bricks, something like 40-40 or fifty.
- 10:52I have Pine cone Cohere which hung in there at A55AS, James
- 10:56said. I have modular and I have Imbue.
- 10:59The scrubs are still scrubbing pretty hard, so there may be
- 11:03some foreshadowing about. You biased towards some.
- 11:06I may do some dropping. You interviewed on stage, which
- 11:09was perhaps yes. I did go with a personal
- 11:12connection vibes based approach it, it may not have panned out,
- 11:17but I got data bricks. So I'm arguably still in 2nd
- 11:21place and I will take that, all right.
- 11:22Let's let's go into the drops. So Eric, you freed up two spaces
- 11:27already with your inflection and character sales.
- 11:31Are you dropping any other companies?
- 11:33I'm not I'm not dropping anything else.
- 11:35I like obviously open AI clean mistrawl.
- 11:39I'm keeping. I want to say one thing.
- 11:41You know, I was very bullish, I think impression Lee on opening.
- 11:45I just wrote a bear column on open AI at 157.
- 11:50Max and James are not giving me a way to like somehow like sell
- 11:54my current markup so there's no I would never drop.
- 11:57Exit Exeter IPO but. I'm not, I'm not allowed to like
- 12:02lock in where it is and then sort of be short.
- 12:05I'm still. So you're saying go up, you
- 12:07would do it up, Okay. I think it'll get marked up.
- 12:10Got it. But you're just.
- 12:11Forty, $50 billion. But I'm skeptical beyond that.
- 12:16Got it. I would say if with open AI this
- 12:18may be a future debate, but the every new round is getting
- 12:21increasingly bizarre in terms of.
- 12:22Stress. I know, right?
- 12:24Exactly. I talked to some evaluation,
- 12:26some investors in the open in open AI round and they're like,
- 12:28well, we get paid back in two years if they don't grow fast
- 12:31enough, exactly starting to become like a sort of debt like
- 12:35structured. I know if I could cash As for
- 12:37the money it's worth, I'm I would be tempted, but I want to
- 12:41be able given I'm keeping it, I'm keeping it with pride and
- 12:43I'm going to pat myself in the back if I if I.
- 12:46Yeah, all with it. All right.
- 12:47Well, if if the next opening eye round involves like, you know,
- 12:50collateralized obligations or something like we need to start
- 12:54talking about what evaluation even is at this point.
- 12:57So. There's a lot of nuance in this
- 13:00VC game, I think is what we've discovered from this draft.
- 13:03Who knew? I thought it was easy.
- 13:07OK, Max, are you dropping anyone from your team?
- 13:11I I am dropping some of my scrubs.
- 13:14Sad. I will yeah.
- 13:16RIP Max's team. I will.
- 13:21I was honestly considering dropping everyone except data
- 13:23breaks, but I decided to dial it back from that approach.
- 13:28I will stick with Cohere given I'm starting Mark leave it
- 13:32found. Yeah, the markup Also, I think
- 13:33foundational models like are becoming like state level assets
- 13:37and potentially like Canada will want their own foundational
- 13:40model. You know, like safe super
- 13:42intelligence might be like Israel's foundation model.
- 13:45So I'll keep I'll keep Cohere and I will keep modular because
- 13:49I still ride with my boy Chris Ladner.
- 13:52You know, it might not workout, but he's trying to create a new
- 13:55language. Yeah, OK.
- 13:57Yeah. So I will be dropping pine cone
- 13:59and Imbue pine cone. I went kind of back and forth
- 14:02on, but I, yeah, I the whispers in the VC circles.
- 14:06I, I roll in quite regularly. We're not very positive and
- 14:10stole volume. So there's a little imbue.
- 14:13Stole stole Volley's office, which.
- 14:16You think is a possible sign they've got a lot?
- 14:18Of kids to go around. Yeah.
- 14:21Yeah. Outside of the bad blood on the
- 14:23office front, Eric, would you agree?
- 14:24Imbue. Not a lot going on since I
- 14:26picked up. I haven't heard.
- 14:27I haven't really heard the radar as much, yeah.
- 14:30So either they're working on the greatest takeover the world
- 14:32strategy ever, you know, or there's not a lot going on
- 14:35there. So I'll go without a lot going
- 14:37on. I'm dropping those two pine cone
- 14:38in view it. I wish I could say it was fun,
- 14:41but it wasn't really. All right.
- 14:42All right, an upward. My team, I'm feeling good about
- 14:47Anthropic. As we've said, I think you know,
- 14:49even in the last year we've seen a lot of kind of progress in
- 14:52terms of catching up to open AI in the closed model race and a
- 14:57lot of success with their specifically their coding
- 15:01capabilities and enterprise adoption.
- 15:03So very excited about Anthropic Hugging Face as well.
- 15:07Still I would say core player in the space.
- 15:09So I'm keeping them. I was decided to drop AI 21
- 15:13labs. They essentially have an open
- 15:16source model, but it seems pretty behind other open source
- 15:19providers And they, you know, their claim to fame is a lot to
- 15:23do with the context windows and the kind of inference time.
- 15:26But I think that is getting a little bit commoditized as we've
- 15:29discussed in in earlier segments here.
- 15:34So dropping them I was debating reblet but I've decided to keep
- 15:39them on my team. Reblet.
- 15:40Keeping Reblet, Yeah, that was an interesting, yeah.
- 15:43It was. Give us the case for why you're
- 15:45keeping reblet here. Yeah.
- 15:46Yeah, so coding, so reblet. Coding is hot reblet.
- 15:51If I were starting as an engineer again, I would probably
- 15:55be using them. I think that it is the best way
- 15:58to get started coding. They have a vision of, you know,
- 16:00empowering like billions of billions of creators to create
- 16:05applications. I think it's a little bit of
- 16:08like a long run or a high upside bet.
- 16:11And I still think it's the right.
- 16:14They have product market fit. They have customers that have 20
- 16:18million users or something. They're working on AI agents.
- 16:21They're really in the mix. I hear about them a lot, even on
- 16:25X, so I think there's, yeah, there's still some upside.
- 16:30To be had there, they're good. It was a tough call out.
- 16:32There. Prominent FX Yes, I was can.
- 16:35I is that my take? My contrarian take here, well,
- 16:38not contrarian at all, but contrarian to you apparently is.
- 16:42I think what we learned about the first year is we are not in
- 16:46the Unicorn hunting business. We're in the deck of horn
- 16:49hunting business because nothing shows up on this chart other
- 16:52than the the $10 billion plus evaluations right you guys
- 16:57congratulations you had X like 1234 billion whatever Yeah,
- 17:02exactly. Well, I think real VC is if
- 17:03you're a CVC you can be in the Unicorn business.
- 17:05But we're we appear we're basically gross stage VC is
- 17:08we're 100 million minimum. So we're.
- 17:10And we're able to get in like $3 billion.
- 17:13Yeah, I, I just I'm yeah, right. So I'm not, I'm just not sure.
- 17:19I don't. I'm not sure I see the path to
- 17:20deca corn for replit. How?
- 17:22How would you see that for modular?
- 17:24Explain the deca. For modular, OK, here's the Deca
- 17:28corn case for modular. Modular replaces CUDA as the
- 17:32core language for how you code and gets bought by like, you
- 17:36know, AMD or ARM or something as like the way you code on non
- 17:40NVIDIA chips. And there's so much demand for
- 17:43getting off NVIDIA, not paying Jensen absorbent fees that
- 17:46there's huge value in having an alternate AI coding ecosystem.
- 17:50So there you go. That's what I believe.
- 17:52Look, I think, I think it's a very asymmetric upside bet where
- 17:56it would probably be worth nothing in 90% of cases.
- 17:58But I like my 10% upside case and I'd be interested in your
- 18:0110% upside case for repple. Like give me the deca corn.
- 18:03Repple it's it's basically that that becomes how everyone learns
- 18:07to code the number of engineers, the number of engineers in the
- 18:12world. Seven, I mean, that's there's
- 18:14no, you know, everyone codes in the world like the.
- 18:16I think it'd be good. You didn't let me finish.
- 18:19I think that, I think that, you know, let's say you have
- 18:24essentially hundreds of millions or billions of engineers in the
- 18:27world, Like it becomes, you know, the spreadsheet or
- 18:31something, right, coding. Everyone became like a sort of
- 18:35an engineer in the PC era through spreadsheets, but maybe
- 18:40that maybe true built application building becomes
- 18:43like a core part of people's knowledge workers jobs in the
- 18:45future. And yeah, maybe it becomes a
- 18:47true platform for that. Or there's a bit more of like a
- 18:50consumer take, I guess where, you know, somehow creating these
- 18:53applications becomes, you know, part of social network type
- 18:57experience. But yeah, I, I think it's a,
- 18:59it's a stretch, but it's, it's possible.
- 19:00I think it looks like we all have 4 spots to fill given our
- 19:05drops and our new slots. So we're gonna pick up exactly
- 19:09where we left off the snake draft.
- 19:11I had the last pick of the draft last year with Adept at #15 so
- 19:17it'll be my pick coming in hot with draft pick #16 and then
- 19:23we'll snake draft from there to Max and Eric.
- 19:26So yeah, my, my pick. I don't think this will come as
- 19:29much of a surprise, but I'm picking Xai, which has been
- 19:34rumoured to be raising at a $40 billion evaluation, did.
- 19:39You give it an extra premium for being anti woke AI.
- 19:44Well, Eric, they told me that they were.
- 19:46Yeah. They're going to take over the
- 19:47world. So very, very, very.
- 19:50Really. Boosted, Yeah.
- 19:53It's a good figure. Obvious pick.
- 19:54Yeah, yeah, yeah. Yeah, I think it's a good pick.
- 19:57I think they spun up I think 100,000 H one hundreds in the
- 20:02last four month months, which Jensen himself said it wasn't
- 20:05possible. But but then according to Elon,
- 20:07they've done it. They're they're they appear to
- 20:09be like. Trucking in gas powered
- 20:12generators to power this thing they have, they're just going as
- 20:19fast as possible I think, which is pretty, pretty compelling and
- 20:21kind of what you need to do to be a foundation model competitor
- 20:24in the environment today. So yeah, don't short Elon, I
- 20:28guess. And they have that Twitter.
- 20:32In which case you probably should have shorted Elon, but
- 20:33yeah, yeah. That's.
- 20:36AI AI does feel like more of a hardware problem and an energy
- 20:39problem than like a more social. Networking problem so more in
- 20:42his wheelhouse yeah, right. I think and maybe X AI helps
- 20:46Twitter yeah, maybe. I don't know yeah.
- 20:49I think like, I think or vice versa.
- 20:51I think it's, you know, kind of an interesting.
- 20:53I, I'm actually very intrigued by how that works from a
- 20:56business perspective, like because the current only product
- 20:59of XAI is this Croc chatbot that's available only in X
- 21:04Twitter. So I I don't know how this looks
- 21:07as a partnership long term, maybe they become the same
- 21:10company or something, but this feels like a good pick to me.
- 21:15I think it's the obvious first pick.
- 21:16I think the short case is. It's like the third best model
- 21:19right now, but the long case is Elon's good at bending the laws
- 21:22of physics, and AI is mostly a laws of physics problem in some
- 21:25ways with yeah chips. And recruiting and get it
- 21:28raising fundraising. Yeah, great.
- 21:30Sure, totally. Yeah, he's.
- 21:32Yeah. All right.
- 21:33Max, you're up next with the next pick of the draft.
- 21:37I will take kind of the obvious second pick set up.
- 21:40Eric whining about how his open AI first pick screwed him.
- 21:44For this redraft, I'll take Scale AI, which is far and away
- 21:49the second most valuable open slot on the board.
- 21:53They just raised to 14 billion. They provide data labeling
- 21:56services to a bunch of different AI companies.
- 21:58I think what's interesting about Scale is they are very
- 22:02polarizing in Silicon Valley. I would say a lot of people have
- 22:06joked for many years that they're just a like sort of
- 22:10random third world worker, like Mechanical Turk business where
- 22:15they offload a bunch of menial tasks to people in cheap
- 22:19countries. But on the other hand, they
- 22:24basically like 4X revenue in the last year, some crazy thing from
- 22:28like 1 to 4 billion. And so if I'm getting at this at
- 22:31like 3X revenue, I'm feeling pretty good.
- 22:34And I, yeah, I mean, the lesson of the last draft is you gotta
- 22:37gotta ride the horses that are winning.
- 22:39And this is a clear #2 right now.
- 22:41So I'll grab scale somewhat conflicted about whether or not
- 22:47it's a long term multi deca corn, but you know, can't not
- 22:52take the the only other ten billion.
- 22:54Plus so the data piece of the AI puzzle, which we talked the
- 22:57least about but couldn't. Right, yes, there.
- 23:00Yeah, we haven't really weighed in on the take that data, you
- 23:03know, quality of data is long run a huge piece of the puzzle
- 23:08maybe equally or you know, as important as as chips and
- 23:13compute and scale and energy and so on.
- 23:15Because there's a lot of garbage data out there and especially we
- 23:18just scrape the whole Internet and you don't have good stuff,
- 23:20you maybe won't get a good AI or same with, you know, imagery or
- 23:23video or whatever. So that's the scale case.
- 23:26The scale short case, as I said, is that oddly enough it's
- 23:29business might be very automatable by AI in some ways.
- 23:32So yeah, that's my take. And we have the CEO joining us
- 23:36at the conference. Yes, we have CEO and Eric, I'm
- 23:38sure we'll ask him some hard questions.
- 23:39So it'll be fun. I'm excited.
- 23:41It'll be fun. I'm up.
- 23:44Eric, you're up. Yes.
- 23:45You get 2 picks. 2 picks, 2 picks.
- 23:47OK. I think the first one is what
- 23:49you guys would pick partially hidden on the stock because it's
- 23:54valuation increase. I think it's only reported and
- 23:57not locked in, but I'm picking perplexity.
- 24:02Smart. Yeah, I mean.
- 24:05Bad. Highest chance of disrupting
- 24:10Google search. I do think a lot of investors
- 24:12are very excited about it even though it you know as a reporter
- 24:17it gets me a little annoyed with their seeming story RIP offs.
- 24:20I do like the attribution philosophy and excited to see
- 24:25them continue to develop. I did not realize I was going to
- 24:28get to back-to-back. Can I get of a genuine
- 24:30perplexity short case like? Can you give one or?
- 24:34Can I give one? Yeah.
- 24:35Can I? Can I actually critique this?
- 24:38Picks, isn't it? Next week, yeah.
- 24:39Yeah, I just think if you look at the data, like if you look at
- 24:43App Annie or you look at any of the open, you know, data
- 24:47services that tell you how much usage perplexity has, it's like
- 24:50shockingly small. It's like really, really not
- 24:53that popular. It's very popular among nerds on
- 24:57Twitter, which is a perfect overlap with venture capital.
- 25:00I know there's this hot take that it's gonna disrupt Google
- 25:03or whatever, but you know, most people are pretty just used to
- 25:08typing in the search box that's on their browser, that's either
- 25:11on their iPhone or on their computer, and that's going to be
- 25:13Google forever and ever because they have the money to buy that
- 25:16box from Apple. According to the information, by
- 25:19the way, they have 20 million ARR.
- 25:21Is that surprising? Based on what?
- 25:23You're, no, I mean, I think like, you know, they probably
- 25:25have a couple million Mau, you know, somewhere between 2:00 and
- 25:285:00. And you're monetizing those
- 25:29people, you know, reasonably effectively, but not crazily
- 25:31effectively. So that's a ballpark, you know.
- 25:33Yeah. So let's say they have 5 million
- 25:35Mau or whatever, Google has like 2 billion Mau or something, or I
- 25:39don't know. It's it's, you know, it's 1000
- 25:41times bigger or five and it's free.
- 25:42Or something I know. Yeah.
- 25:43And it's free. Yeah.
- 25:44So I just, I just would say, like, I'm not convinced the
- 25:47differential against Google's AI summaries or whatever is going
- 25:51to stay there for very long. A bunch of people gonna RIP off
- 25:54their interface innovations and I don't really see how they're
- 25:58going to get a distribution advantage.
- 26:00So yeah, I'm, I'm definitely short at whatever the what's the
- 26:04latest valuation like 10 or something?
- 26:05It's reported 8 billion eight. Yeah, I'm I'm short at 8 for
- 26:08sure. I'm I'm probably long at like 1
- 26:11or I don't know but I'm I'm short at. 8 The next Google
- 26:15we'll see no. I just think that I think that's
- 26:17an overrated take. Yeah, I'm sorry.
- 26:20I. Eric, who are you picking you?
- 26:22Know I went all over the place. I certainly have companies down
- 26:25bit lower on the valuation list that I'm probably more amped
- 26:30about. I thought about Sierra, like
- 26:34customer support being sort of a promising area where I could
- 26:39actually get deployed. But just looking at my strategy,
- 26:42like we're going for big ones. You got to go with the guy who
- 26:46was the genius behind open AI that could either lead research
- 26:50at another company, could invent AGI, VMS AGI.
- 26:54I'm going to win like, so I'm going safe super intelligence.
- 26:57Ilya, I'm betting on you. And then obviously, I mean,
- 27:00Daniel Gross is, you know, one of the most connected people in
- 27:03Silicon Valley these days. Music Co founder.
- 27:05That's what I'm going with. Yeah, Yeah.
- 27:07For the record, I thought that was where you were going to say
- 27:09the obvious next pick was because that's definitely I
- 27:11would have taken injuries. Why would you get?
- 27:14Why are you guys so excited about this company?
- 27:16I'm I'm short like. You're short, right?
- 27:19I guess as an as an owner of XAI you.
- 27:23You believe in dangerous super intelligence.
- 27:26You're like, you're like now. I don't know. the Super
- 27:28intelligence has to be dangerous to be valuable.
- 27:30Like the safe. The safe is no good.
- 27:32I have a balanced diversified strategy of having anthropic and
- 27:37XAI. Save the world, Kill the world,
- 27:40Who can go to safety? Give us the short case.
- 27:43Yeah, What's the short case? Yeah.
- 27:45I just think that, yeah, like I think that he, you're basically
- 27:52basing this off of, you know, a billion dollar evaluation and he
- 27:56can create a new foundation model, right?
- 27:58I think that's the only explanation for why this is
- 28:01important and and it just feels like he's so far behind.
- 28:04All right. So my pro safe super
- 28:06intelligence take here to stand behind Eric is I think that
- 28:12these foundation models are becoming so expensive that their
- 28:14borderline have to be nationalized in the next like
- 28:16two to four years. And I think safe super
- 28:19intelligence is basically going to be Israel's national model,
- 28:23at which point, you know, I don't know if that how that
- 28:26impacts the valuation. And we could debate that a
- 28:27couple years down the road. But fundamentally, like it's
- 28:31going to be a nation state level concern to have a really high
- 28:33quality model. And I think Ilya will probably
- 28:36be the the flag bearer of Israel, which maybe was the AI
- 28:4021 case for for James a year ago, but they might have lost
- 28:43out to SSI. So that would be my pro case.
- 28:46And that's why, yeah, I'm a fan of the pick.
- 28:48All right, let's keep moving. All right, Max, you're up with
- 28:52the next pick. Well, Eric kind of foreshadowed
- 28:57this that I think, you know, a solid option for next pick is
- 29:00Sierra, which is X Salesforce CEO Brett Taylor started a AI
- 29:07kind of customer support AI company.
- 29:10You know, I find customer support super boring.
- 29:12There's like 250 of these fucking companies.
- 29:14I don't really see any obvious reason these guys are going to
- 29:17win because I haven't seen the actual product or compared them.
- 29:19But like any good VC, I'm just riding the heat of all the other
- 29:24hot VCs being in this company. And everyone I talked to in the
- 29:27biz is like, this Is it baby Sierra?
- 29:31Yeah, exactly. Benjamar and Sequoia get me
- 29:33wrong, baby. It's never lost me money before.
- 29:35So I'm going to freely admit I know almost nothing about why
- 29:39this is supposed to be the best of these 200 different customer
- 29:42support companies and say, take me to the top.
- 29:45Brett Taylor. I love it.
- 29:46Basically just you're just going with the the VC cred.
- 29:50Hell. Yeah, OK.
- 29:52I believe in that philosophy. That was my Glean bet last time,
- 29:57Kleiner and Sequoia. Totally worked.
- 30:01All right, I will take the next pick of the draft.
- 30:05I guess I got 2 picks here. I think this will be maybe a
- 30:10slight surprise, but so with the next pick, I'm taking 11 Labs.
- 30:16Yeah, 11 labs. We are already a customer.
- 30:20They definitely have a lot of product market fit in this voice
- 30:24cloning and text to speech space.
- 30:27They seem to have been making moves, I think I alluded to
- 30:31earlier, buying companies that provide consumer applications so
- 30:36that you could, you know, read articles with these generated
- 30:39voices. They have been, they've released
- 30:44a lot of new technologies, including more.
- 30:47Most recently they announced kind of a new agent framework
- 30:50for voice AI applications. They also have a sound effects
- 30:55API. Yeah, they ship a lot of
- 30:56product. They have a lot of fans in the
- 30:59developer community. They're taking an eye towards
- 31:03some consumer applications. So I'm going with 11 lamps.
- 31:07I I'm short 11 lamps. I I just as a customer, I like
- 31:14the product. It also seems immensely
- 31:18replicable by the large cloud services.
- 31:20I think once Lena Khan is no longer blocking acquisitions, it
- 31:23will get bought by AWS or Google or Amazon or Microsoft.
- 31:27All right, And with my next pick of the draft, I'm going to take
- 31:35runway so all. Right, that was my next.
- 31:38Pick. Really, I would.
- 31:40I love them. Wow, fascinating.
- 31:44OK, You think that's a bad pick, Eric?
- 31:48Yeah, I mean, he was the one at Cerebral Valley New York making
- 31:51promises that our own AI crowd didn't even believe.
- 31:54I don't know. And now they're now they're sort
- 31:56of saying they're in like the video production business
- 31:59themselves. I don't, I I'm very bullish on
- 32:02Adobe figuring out what creatives want and monetizing
- 32:05it. Well, yeah, so no, I'm not.
- 32:09I just think what we're in the business of the DECA corn
- 32:11hunting runway, the the hype train is why I'm investing in
- 32:16them here. They said, you know, we'd be
- 32:19creating Oscar winning fully feature length movies, you know,
- 32:23in two years or something. They told Max that on stage.
- 32:26I think there is a lot of high upside scenarios where video
- 32:31Gen. tools become like a much more interactive medium also,
- 32:36right. So if you could really generate
- 32:38video that's photorealistic and and sort of in that low latency,
- 32:41you can create almost gaming use cases.
- 32:43There's a lot of like simulation type products they could build
- 32:46and then there's consumer use cases, just, you know, creator
- 32:49tools. So they do seem to be the leader
- 32:52in this important AI video space.
- 32:54We also haven't seen, you know, open AI launch Sora, even though
- 32:58that received a lot of, you know, attention when they
- 33:00announced it. So I don't know super.
- 33:02Expensive. Yeah.
- 33:05Do you think Runway is just burning burning capital then or?
- 33:08I don't know that specifically, I'm just saying if all these
- 33:10models cost a ton, like video models sound like they're going
- 33:13to be like, you know, so, so cost in terms of.
- 33:17Yes. And right now, I think, you
- 33:19know, the main customers of Runway are like a feature length
- 33:22Hollywood, you know, production studios or maybe like doing, you
- 33:25know, a few like one second clip generations or something to like
- 33:30fill in part of their films. But I think that, you know, as
- 33:34cost curves come down, it just opens up a lot of use cases.
- 33:36But yeah, if cost doesn't come down then maybe it's a bad pick.
- 33:39I like I like the asymmetric upside bed to take over the
- 33:41world plan. So I'm kind of in.
- 33:43Should I take the next pick? Yeah, go for it.
- 33:45All right? We're definitely getting into
- 33:46the harder part of the draft where I think you have to go a
- 33:49little more wild with a little crazier pick.
- 33:52I'm gonna take Sakana AI from one of my VC referrals here.
- 34:00Said this is the hot thing. You should check it out.
- 34:02Apparently, they're literally trying to use AI to automate
- 34:07fundamental scientific discoveries.
- 34:10Essentially, they're huge in Japan.
- 34:12I believe in basically the core mission that it is possible to
- 34:15make scientific breakthroughs with AI.
- 34:17And so I don't know, that sounded like an asymmetric high
- 34:20upside bet. And I'm getting good buzz from
- 34:22the VC networks. And at this point, it's pretty
- 34:24hard to kind of dig through the draft.
- 34:26So Sakana, welcome to the team. Is it OK if I show my hand and
- 34:30sort of talk through the four and then picture?
- 34:32Yeah, sure. I mean, yeah.
- 34:34You're just giving you guys Intel leak.
- 34:35Ideas to us, so it's. All good?
- 34:37Yeah, maybe I won't. No.
- 34:38No, no, come on. Is it a podcast first or is it
- 34:41winning first? Do it for the content baby.
- 34:44All right, I'm debating Thesis 1 is like man, we keep saying it's
- 34:48all about applications. We're bullish on coding and the
- 34:55other big hyped application category is legal.
- 34:58So Codium is coding, Harvey is legal.
- 35:02Those are the applications. Cursor is very buzzy, but I'm
- 35:05not allowed to take it. They haven't raised $100
- 35:07million. I met an interesting services
- 35:10legal startup whose name on blank.
- 35:12Yeah, I'm sorry guy, but I don't think they've raised enough
- 35:15money. So so that would leave Harvey
- 35:18and Codium weights and biases. I think is a company that's
- 35:21that's done well in serving a lot of like the infrastructure
- 35:24layer and could be like a solid company.
- 35:28Listening to myself, I'm clearly excited about getting hits on
- 35:32coding and law. So I'm gonna take Codium.
- 35:36I'm gonna take Codium first and then Harvey second.
- 35:43All right. All right, actions.
- 35:44What do you think of the Harvey controversy that?
- 35:47What do you think of the Harvey controversy that.
- 35:49Well, there's that random tweet that is like vaporware.
- 35:52You talk to some of the legal startups and they're like, we
- 35:54don't see them out there, but. Well, I, I've, I've talked to a
- 35:57couple lawyers and they're like, it's not that great right now.
- 36:01Now maybe you're like, I'm long models getting better, so it
- 36:03doesn't matter. They have a better sales force,
- 36:05yadda yadda yadda. But.
- 36:06Both of them I'm betting on the category without knowing a lot
- 36:10about the culture. You're just basically trying to
- 36:12grab the leader in a category you believe in.
- 36:14All right. Exactly.
- 36:15Fair enough. Yeah, I, I think that's a
- 36:17reasonable thesis. I I do think the buzz that
- 36:20Harvey is not that great is it seems to have some where there's
- 36:24smoke, there's fire kind of element to it from from my
- 36:27research. What makes you so sure that
- 36:29Kodium is the leader in this category?
- 36:33Why is there? Here's my other option on this
- 36:35list. Well, I don't know.
- 36:37I'm just, I'm just wondering like what's the evidence of
- 36:39that? I guess they're.
- 36:40Just raised enough money to get on the list I guess.
- 36:42Isn't isn't sure. Isn't magic the same thing?
- 36:45Kind of or similar. Magic.
- 36:47Yeah, there's. And obviously cursor hasn't
- 36:49raised enough, but Magic was like Daniel Gross's play into
- 36:52that space, right? Yeah, yeah, I've heard of
- 36:54Kodium. Fair enough.
- 36:58The end of the draft is tough, man.
- 36:59I think you're just grabbing ideas here.
- 37:01I mean, cognition is also codium, to be clear.
- 37:04Yeah, I could have taken. Yeah.
- 37:07Exactly. Cognition.
- 37:08Yeah. Poolside is coding source, Scrap
- 37:11is somewhat code related. There's just a lot of coding
- 37:16companies up here. So I was just curious.
- 37:17But yeah, yeah, why you went with Codium specifically?
- 37:21But it's all good. It's all good.
- 37:23It's done. It's done.
- 37:24Is done. All right, I will.
- 37:26It's done. What's done is done.
- 37:28The bottom of the draft is tough.
- 37:31Max your last. My final pick, I will take
- 37:36Hebia, which is sort of like Eric's legal thesis, but for
- 37:43Excel. In some ways it's sort of like
- 37:46the finance, like private equity, you know, investment
- 37:52banking to some degree, whatever.
- 37:53Excel, Jackie, AI, basically, I've heard very good things
- 37:56about the products. It has a, you know, super elite
- 37:58investor list with Peter Thiel and Andreessen and everything.
- 38:02And I actually talked to someone who worked there and they were
- 38:04like, this is a pretty sick product.
- 38:06And I don't think they were just talking their own book.
- 38:08So yeah, Hebia AI for Excel jockeys, which I was an Excel
- 38:13jockey 12 years ago. So that that sounds pretty good
- 38:15to me. Would never have considered that
- 38:17in my life. It's.
- 38:20The bottom of the draft. It's the bottom of the draft.
- 38:22You got to take some sleepers, you got to, you got to, you got
- 38:26to have a dream. All right, last pick of the
- 38:31draft, James. Last pick of the draft, a lot of
- 38:34companies still on the board. Yeah.
- 38:37So I'll walk you through some that I'm considering kind of a
- 38:42debate between I guess three companies here.
- 38:46Poolside, which I mentioned is another AI powered software
- 38:49development platform. They raised $500 million in a
- 38:53Series B this month, which I thought was pretty interesting
- 38:57quantity of. Money.
- 38:58For clear purposes first, theoretically for a foundation
- 39:03model around trained around coding.
- 39:07The other one on my list. Still, I think Eric mentioned
- 39:10them earlier, they do seem to be kind of a leader in this
- 39:14developer space where if you wanted to, you know, use open
- 39:18source models and do fine tuning and host them and and have the
- 39:22infrastructure for that, they seem to be one of the leaders in
- 39:25that space. And then the last one would be
- 39:27very much a flyer, which is just World Labs, which is Fei Fei
- 39:32Li's startup, which just raised about 100 million in July at a
- 39:36$5 billion valuation. So yeah, there's some
- 39:41opportunity there. But I think similar to my bear
- 39:47case on Safe Super Intelligence, I think I will be rolling that
- 39:50one out. And in terms of going after that
- 39:55sort of the software space, I think I'm gonna have to pick
- 40:00poolside. Wow.
- 40:02Lots of money they could sell. They could sell for a billion,
- 40:06you know. Exactly just based.
- 40:07On the money just. For the cash just to get to.
- 40:09Lock in the money, you know. I mean, yeah, I don't know.
- 40:13I don't know what they're, what they're working on.
- 40:15I hope it's worth. A lot, yeah.
- 40:18I upset I upset I. Upset yeah should we do some
- 40:23bring out your dead any picks that we.
- 40:25We think you should have picked you.
- 40:26Should have picked. Yeah.
- 40:27I mean, I don't know. I'm shocked you didn't pick
- 40:29Suno, James. You're just banging on Suno like
- 40:31every weekend. Suno's a great product.
- 40:33Grab some SUNO stock. I just, I'm very nervous about
- 40:38the music industry, having worked with the movie.
- 40:40Yeah, you. Worked with the music industry.
- 40:42You. You have some?
- 40:43Yeah. When do we get sort of picking
- 40:44volley in this? Like no, you haven't raised.
- 40:47Next round, we haven't raised. Next round.
- 40:49Yeah, Yeah, that'd be good. Yeah.
- 40:52I think Magic is interesting. I like that as another coding
- 40:54pick, my VC Whisper source. Hated together.
- 40:59Hated together. Oh yeah, OK, I didn't go it
- 41:02together then. Yeah, it's like, I think
- 41:04Together's A0 Max, you're behind, so you're doing more.
- 41:08High up. Research now I've no yeah I see
- 41:10like last time I did a lot of research and this time I sort of
- 41:12just like. Just for it, yeah.
- 41:14I'm coasting. I'm winning.
- 41:16You. Know I'm not yeah, I'm not sure,
- 41:18but these also ran kind of models we have here.
- 41:20I think, you know, Pika is, is you know, the third best video
- 41:24model. So I'm a little, I'm a little
- 41:25concerned about that. Cognition's interesting with
- 41:29Devon, but I guess like seems to be less of a foundation model
- 41:35play and more of this agentic play, but maybe the hype has
- 41:39died down a little bit. It is amazing how just like in
- 41:42six months, the hype cycles are just changing so rapidly.
- 41:45That stuff, it was just like crazy hot, you know, at our last
- 41:47conference or two conferences ago now feels like, oh, that's
- 41:50tired. You know, they haven't done
- 41:51anything in eight months or something.
- 41:53And it's. Like.
- 41:54OK, well, yeah, the cycle time in the AI industry is just
- 41:58ludicrously short right now. Yeah, cool.
- 42:04Should we wrap? Should we wrap up our text?
- 42:05Yeah, yeah, yeah, yeah. OK, just let's let's compile
- 42:11this quickly. So Eric, your team now consists
- 42:16of open AI. You banked some money from
- 42:19inflection and character and you you continued to ride with Glean
- 42:25and Mistral and then you added two pick or 4 picks to to your
- 42:30team. Perplexity.
- 42:32Great pick I think, but you know, risky as Max said.
- 42:36Safe, Super intelligence safe Pick safe.
- 42:41Safe. Safe pick safe.
- 42:43And super intelligent. Pick safe, super pick Kodium and
- 42:48Harvey. How are you feeling, Eric, about
- 42:50your team? I'm I'm happy, honestly, I
- 42:52think, you know, I wish I could have gotten X AI, but given what
- 42:59was available, I'm I'm happy with my picks.
- 43:03Max, you dropped a few folks from your team.
- 43:07You have you kept data bricks cohere and modular and then you
- 43:11added scale AI high valuation play there you added let's see,
- 43:22sorry. Sierra.
- 43:24S s Sakana Sakana and Hebbia yes, I like things that.
- 43:29End in a apparently. And start with S.
- 43:34And start with S. Yes.
- 43:39And then how are you feeling about your team?
- 43:42I mean, I'm feeling better than last year.
- 43:44I'm glad I banked scale. So I've got some, some valuation
- 43:48games. I still have data breaks.
- 43:50I still have scale. I've got some valuation.
- 43:53You know, I, I, I still think I'm basically running the stars
- 43:56and scrubs play here, which which didn't really work last
- 43:59time. So other than other than again,
- 44:02just data breaks. So we'll, we'll see.
- 44:04We'll see. Yeah.
- 44:05I'm I'm a little less optimistic perhaps than I was last time
- 44:09around. And then my team, I have Hugging
- 44:13Face, Anthropic Replit, I banked some cash from the Adept Amazon
- 44:18deal and I'm adding XAI with the highest valuation on the board
- 44:24right now, adding runway, adding 11 labs and adding poolside a
- 44:35lot of like sort of foundation models, developer tools.
- 44:38I think with Anthropic and poolside getting that kind of
- 44:43coding play. But also Anthropic has one of
- 44:46the best, I would say coding LMS right now with Claude, But I am
- 44:52a very bullish on that. That is, you know, valuable use
- 44:57case of this technology. So riding, riding the developer
- 45:00tooling ecosystem. You're you're like, you're
- 45:03running like the murderer's row of companies I think are
- 45:06overrated, basically, which probably means I'm wrong.
- 45:12And one of them will be like a $200 billion company.
- 45:14So yeah. You only did one necessarily
- 45:17right? You're just like, you're just
- 45:18naming them off. You're like, you're like
- 45:20poolside Rep lid 11 Labs. Like, what was the other one
- 45:24like? You know, Hugging Face.
- 45:27Anthropic is great. For the record.
- 45:28You know, Anthropic is Immaculate, but you're you.
- 45:31You. Yeah, you just got a lineup of
- 45:33like, Twitter memes that I hate, basically.
- 45:36Just because you feel like they're overvalued or?
- 45:39Yeah. I just like, I don't yeah, I
- 45:42think overvalued or or just, you know raise too much money in the
- 45:44case of poolside with no product or or I think the product has a
- 45:48sort of lack of, you know, large market.
- 45:51Would either of you pick what Kodium or Harvey?
- 45:53Were they on your list anywhere? Well, Harvey was on.
- 45:57My. Harvey was on my list.
- 45:59Harvey on your hater list? Yeah, yeah, yeah, Harvey was
- 46:02probably on my hater list. Yeah, so.
- 46:04In some ways you have. The company to be a little
- 46:06polarizing you want. People to be thinking about,
- 46:08right? Exactly.
- 46:09Exactly. Yeah, There's a top of mind.
- 46:12I bet the Max anti portfolio is better than my actual portfolio.
- 46:16So don't worry it's like I need I need the the inverse of my
- 46:21hater list. We'll probably.
- 46:22Do better than my. My actual list, I mean, at the
- 46:23end of the day, I'm still like, really exposed to open AI more
- 46:27than anything. It's like, yeah.
- 46:30Yeah, yeah, maybe. Maybe the whole industry is,
- 46:34yeah. I mean that.
- 46:35That's been the question for. Lt. events just is it really all
- 46:38just about open AI and still to be seen right?
- 46:43One update from our overrated, underrated episode.
- 46:48Minor correction on one of the revenue figures, Max.
- 46:51Yeah, I would say a major correction.
- 46:54A major correction, but we got the source, we relate what the
- 46:58source said accurately, all right, it relates outdated.
- 47:02All right. To make this clear, the
- 47:04Information Special Pro subscription told us that
- 47:07Anthropic was making around 100 million in revenue per year.
- 47:11Apparently CNBC reported that Anthropic is in fact closing in
- 47:15on making a billion dollars in revenue a year, which is quite
- 47:19quite a bit larger and is growing its revenue at basically
- 47:2410X year on year. So that hundred may have been
- 47:27accurate, you know, 910 eleven months ago.
- 47:30Luckily for James and Eric, they both already look super smart
- 47:34and they say Anthropic was. Unrelated, much smaller revenue
- 47:38than it actually has, yes. Max luckily gets to look super
- 47:42dumb when he said Anthropic was very overrated because I would
- 47:47say 10X you're on your revenue growth absolutely justifies a
- 47:50crazy 1/3 of open AI valuation. Interestingly, I stand by my
- 47:55take that their consumer app growth is not that impressive.
- 47:58I think it was 20% in the last quarter because it looks like
- 48:01basically all the revenue comes from their API and their sort of
- 48:05developer tools. All right, So that's a minor
- 48:07update on our overrated, underrated episode, keeping
- 48:11ourselves honest, and we'll let you know if there's any new
- 48:15information that comes to light after these episodes.
- 48:17And yeah, reach out to us. You know, you can what?
- 48:20e-mail Eric at newcomer.co, probably that's me personally.
- 48:23And send me your takes Comment. We want more comments.
- 48:27Yeah, comment on the post with your own opinions.
- 48:29And certainly happy to. We have very smart listeners, so
- 48:33would love to hear where we got it wrong.
- 48:36Reach out anytime. Thanks so much.