Anthropic's S-1, Samsung Controversy, IGTV | Doug O'Laughlin & Joey Brookhart, Andre Haddad, David Faugno, Mark Swan, Chanda A. Bell & Christa Pitts, Ivan Zhang & Luisa San Martin

(01:17) - Anthropic's S-1 (28:45) - Agents Break into Anthropic's Systems (34:51) - Cohere Model Attempts Unexpected Action (44:25) - Samsung Controversy (50:57) - 𝕏 Timeline Reactions (51:00) - IGTV (01:05:54) - Doug O'Laughlin & Joey Brookhart discuss Anthropic’s strong IPO fundamentals, capital efficiency, compute commitments, and the economics of AI labs and neoclouds. A trend-focused SemiAnalysis expert and enthusiastic AI power user, he also examines cybersecurity threats, open-source model risks, enterprise adoption, and the emerging role of personal and workplace AI agents. (01:35:20) - Andre Haddad, Turo’s CEO of 15 years, discusses the company’s growth and launch of Verified Full Self-Driving for Tesla rentals through a direct API integration. He also highlights Turo’s vision for autonomous vehicle delivery, improved risk management, and broader brand awareness. (01:48:46) - David Faugno, CEO of 1Password, discusses the growing security risks posed by AI agents and the need to limit their access to sensitive credentials. He highlights 1Password’s rapid growth, its new security tools for AI agents, and practical protections such as passkeys, hardware keys, and strong credential hygiene. (02:00:38) - Mark Swan discusses Nevis, the AI wealth-management platform he co-founded and leads as CEO. He explains how Nevis equips independent financial advisors with agents that automate complex workflows, generate proactive recommendations, and improve efficiency without replacing human advisors, while highlighting the company’s rapid growth and partnership with Charles Schwab. (02:08:34) - Chanda A. Bell & Christa Pitts discuss co-founding The Lumistella Company, the family-owned business behind The Elf on the Shelf, and transforming a cherished childhood tradition into a global Christmas franchise. They highlight the company’s entrepreneurial journey, major growth milestones, entertainment and licensing partnerships, and commitment to creating joyful family moments during the holiday season. (02:20:56) - Ivan Zhang & Luisa San Martin discuss his role as Cohere’s co-founder leading efforts to secure AI models and systems. They explain the discovery of potentially dangerous data-exfiltration behavior in an open-weight model and advocates for stricter safety standards, transparent research, and careful testing before release. (02:31:26) - 𝕏 Timeline Reactions TBPN is made possible by: Ramp - https://ramp.com Public - https://public.com Cisco - https://www.cisco.com Console - https://www.console.com CrowdStrike - https://www.crowdstrike.com Figma - https://www.figma.com MongoDB - https://www.mongodb.com NYSE - https://www.nyse.com Railway - https://railway.com Shopify - https://www.shopify.com Codex - http://openAI.com/codex Follow TBPN: https://TBPN.com https://x.com/tbpn https://open.spotify.com/show/2L6WMqY3GUPCGBD0dX6p00?si=674252d53acf4231 https://podcasts.apple.com/us/podcast/tbpn/id1772360235 https://www.youtube.com/@TBPNLive

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  You're watching TVP.

• 0:01Speaker 01

  Today's Monday, October 26, 2026. We are live from the TVP and Ultimum, the temple of technology.

• 0:07Speaker 00

  The fortress of finance.

• 0:08Speaker 01

  The capital of capital. How many days until Christmas? That's what you've been asking. This Anthropic IPO, it's the most important story of the day after the amount of days till Christmas. Let's count it down. We're very close. Fifty-nine days till Christmas. Do you think it's

• 0:27Speaker 00

  Whoa.

• 0:29Speaker 01

  Okay. Peter Thiel is fifty-nine years old. There you go. Yeah, it's interesting. They're probably, Anthropic seems to be sending us.

• 0:36Speaker 00

  Incredible innovation.

• 0:37Speaker 01

  Yes.

• 0:38Speaker 00

  At TVP and that we tie random fun facts.

• 0:41Speaker 01

  Yes.

• 0:42Speaker 00

  To the Christmas countdown.

• 0:43Speaker 01

  I also like that. Just to clarify, it's not that Peter Thiel is fifty-nine years old today, right?

• 0:50Speaker 00

  No, he's just fifty-nine.

• 0:51Speaker 02

  I think his birthday was two weeks ago.

• 0:53Speaker 01

  I think his birthday was two weeks ago.

• 0:54Speaker 00

  He just happens to be fifty-nine.

• 0:55Speaker 01

  That age, which is a fun fact.

• 0:57Speaker 00

  The number of days.

• 0:57Speaker 01

  Because over the weekend, I was thinking about we did Jeff Bezos' sixty-two on the sixty-two days till Christmas. And I was like, was it Jeff Bezos' birthday? Or was that just a random fun fact? And it appears to be just a random fun fact. Anyway, let me tell you about ramp.com. Time is money. Save both. He's these corporate cards, bill pay, accounting, and a whole lot more, all in one place. So do you think Anthropic with, by dropping the S1 on exactly fifty-nine days till Christmas is trying to send like some sort of subliminal message?

• 1:25Speaker 00

  What would that message be?

• 1:26Speaker 01

  Oh, just that it's fifty-nine days, so you should go into the S1, Ctrl F for the number fifty-nine to find like the hidden meaning of the S1.

• 1:34Speaker 00

  Or go to page fifty-nine.

• 1:35Speaker 01

  Go to page fifty-nine. Maybe you should.

• 1:38Speaker 00

  Have you been yet?

• 1:38Speaker 01

  Oh, I've been to page fifty-nine. I think I know exactly what they're trying to say with this. It's fifty-nine days till Christmas. Open up the S1, Ctrl F, hit fifty-nine, page down to page fifty-nine. What do you see? It says, we may not have the ability to raise the funds necessary to repay the principal and accrued interest of our outstanding. Yeah, it's just a, it's just a random page. I don't think there's anything going on. Anyway, there are a bunch of interesting facts. You got to dig through it. The S1 is huge, obviously. I mean, it's a massive company, raising a massive amount of money. The exact amount of money that they're raising is not, has not been listed yet. There's, there's sort of speculation in the Wall Street Journal, around a hundred billion dollars raised.

• 2:22Speaker 01

  Bloomberg contextualized the IPO filing as Anthropic files for IPO in global race to raise cash for AI. The interesting thing is that Anthropic has a lot of cash. They also have a lot of revenue, sixty-seven point two billion in annualized revenue. That's up forty-five percent quarter on quarter at this scale. Incredible, super, super, super fast growth. And full Q3 numbers aren't in the initial S1, but that, that number that was rumored from the Anthropic Investor Day is now in the actual S1. Cash on hand, one hundred and thirty-one point seven billion dollars. That's a lot of cash. And they've

• 3:05Speaker 01

  been super capital efficient. People are saying it might be the most capital efficient company in history. They, of course, brought in one hundred and ten point eight billion in first half.

• 3:14Speaker 00

  Yeah, it's one of the most remarkable stories in business history, period.

• 3:18Speaker 01

  Yeah.

• 3:19Speaker 00

  And the S1 came out, I think in some ways, like the leak around the S1 that had happened maybe a week or two ago around this, like, and the number that everyone fixated on was this eighty billion dollar.

• 3:30Speaker 01

  Loss.

• 3:30Speaker 00

  Loss, this accounting charge. I think people now have had an opportunity to dig in here.

• 3:35Speaker 01

  Yep.

• 3:36Speaker 00

  And the funny thing is the nature of having access to Claude and ChatGPT and other agents is people can process the S1 a lot faster.

• 3:42Speaker 01

  Totally.

• 3:43Speaker 00

  So we're already getting like a lot of, it used to be like an S1 would release and the VCs would have.

• 3:48Speaker 01

  Like an analyst.

• 3:49Speaker 00

  Twelve hours. Well, they've had twelve hours to pull out interesting stuff.

• 3:52Speaker 01

  Yeah, yeah, yeah.

• 3:52Speaker 00

  Now, it's like you, it's, it's all coming out in real time. And it is, yeah, totally remarkable. I think it's amazing for the AI trade overall.

• 4:01Speaker 01

  It feels like a beat. It feels like this beat expectations.

• 4:05Speaker 00

  Totally.

• 4:05Speaker 01

  It didn't like surprise, oh, they're actually doing a hundred billion in revenue or something. But, I mean, really, really solid. People haven't found anything that's like, oh, I didn't understand that the business was actually shaped a very different way or something like this, this sort of meets expectations.

• 4:22Speaker 00

  I yield, Harry said, you know, it's a decent S1 when Ed Zitron hasn't tweeted anything out yet.

• 4:26Speaker 01

  I think this is a beat on the expectations of the S1.

• 4:29Speaker 00

  Well, he's doing the old Command F, Command F. He has to Command F, right?

• 4:33Speaker 01

  Command F, bad. Command F, sad. Command F, problem. The Nasdaq is up one percent, and it feels like the AI trade is starting to like get put back on. Obviously, there's been a lot of, a lot of selling in the further risk on, the further out on the risk curve assets in the AI trade. Some of the Neo Clouds down a lot. But if you look into what Anthropic's doing, it seems like they're going to be buying a lot of compute from Neo Clouds. Like, there's this interesting thing where they, they raised a hundred and ten billion dollars. They have a hundred and thirty billion dollars in cash. How much did they spend in the first

• 5:13Speaker 01

  half of the year on CapEx, right? This is the company that wants to serve AI at scale. And how much did they actually spend buying computers, building computers that they own? Four hundred and eighty-two million dollars. Less than half a billion dollars. I mean, now, of course, over time, they'll build more, they'll do all sorts of stuff.

• 5:32Speaker 00

  Yeah, they've been renting.

• 5:33Speaker 01

  At least in the first, at least in the first half of the year, the strategy has been to rent. And then also, they have a pretty huge amount of backlog with the hyperscalers. And it looks like this scary number, I think it's like six hundred and ninety billion dollars of commitments. Really not that much, ten X of, of revenue currently, if they keep scaling.

• 5:53Speaker 00

  But over a multi-year period.

• 5:56Speaker 01

  Yeah, yeah. So there's, there's Google, Google, Amazon, and Microsoft are all in the forty to a hundred and thirty billion dollar range. I can pull up the exact numbers, but the, but the contracts are how much that they have to spend on compute through 2033. Some of them go all the way out to 2036, which is really, really, really far. So they have a lot of time to actually work through that. So Google Cloud, their, their commitment is a hundred and fourteen and a half billion dollars. And that's minimum cloud spending through July of 2033. So they're just renting, and it's over a long period of time. Is there

• 6:38Speaker 01

  any scenario where they don't have the demand to actually parcel out twenty billion of spend over the next six years, seven years? It seems like they're in a good position there. AWS is a hundred and ten billion, and that's stretched out to 2036. And then Azure is only forty-five billion, and that's minimum cloud spending through 2033. Now, they do have a hundred and twenty billion with SpaceX AI. That's up to, that's the maximum. And then N scale is at forty-five, Lambda's at forty-two billion, and Broadcom's at a hundred and sixty-one billion, for a total of six hundred and thirty-seven billion in spending commitments. But again, over five, six, maybe eight years in some of these scenarios. So

• 7:20Speaker 01

  pretty, pretty healthy there. The other interesting surprises, I mean, people are pretty happy about the margins. So compute cost as a function of revenue, they were at seventy-two percent. You can think about this as like cost of goods. Seventy-two percent of every dollar they made was getting eaten up by compute costs. That's down to forty-eight percent compute costs divided by revenue. So that's rapidly improving efficiency quarter on quarter. Then the adjusted operating margin gained thirty-two percentage points quarter on quarter for almost a billion dollars in GAAP profit. They went from negative thirteen percent to nineteen percent operating margin. And then the big number that is, that people

• 8:02Speaker 01

  are excited about is the big customers. So they have, I think, over a thousand customers spending over a million dollars, something like that. And then.

• 8:12Speaker 00

  Over ten thousand spending over a hundred K.

• 8:14Speaker 01

  Over, yeah, over ten thousand spending a hundred K. And net revenue retention is nine hundred percent on those. So they got in early in 2024, 2025, and then all of those companies scaled up. So maybe they had a smaller subscription, smaller, they were using the APIs here and there. And then we went to the token maxing era. They were in the right place. And so many companies scaled their spend that on average, it was basically a ten X increase for organizations. But that makes sense. We've all seen this as, you know, more and more people are like, hey, it's not just like the early adopter on the team that has an AI agent running all the time. It's actually spread to everyone. It's spread to the marketing organization, the operations teams using it. And so these

• 8:55Speaker 01

  services are going out all over the place. They have more than half of the Fortune 500 spending over a hundred thousand dollars annually. That actually felt low to me. Obviously, it's a really impressive, like, it's great. But I'm sort of, like, I was under the assumption that all of the Fortune 500 was spending a lot on like every single model. And I would have said, oh, of the Fortune 500, a hundred K annually, the Fortune 500 is probably spending a hundred K annually on open source models and OpenAI and Google and Microsoft and Anthropic. Like, I would.

• 9:28Speaker 00

  You got to ask what that half is, what they're up to.

• 9:31Speaker 01

  I mean, some of them might just be in industries that don't write any software, really. But it feels like every company in the, if you're in the Fortune 500, you have some software, like.

• 9:40Speaker 00

  It's kind of bearish anywhere, anywhere you put it. I do know, I do know.

• 9:44Speaker 01

  You don't do spreadsheets?

• 9:45Speaker 00

  I was, this was maybe six months ago, but I was talking to someone at a Fortune, relatively high up at a Fortune 500, and they said that they had, they were basically running like boilerplate Gemini that like the whole, the whole org had access to. And that was like their AI tool. They were telling everyone, just use this.

• 10:04Speaker 01

  Yep.

• 10:05Speaker 00

  And.

• 10:07Speaker 00

  So there's, I'm sure examples of that.

• 10:09Speaker 01

  Yeah, yeah, yeah. Just, and so, yeah, maybe, maybe you see like 50% has spending over 100K on Anthropic. And then there's some, there's some slice that's on OpenAI, some slice on Gemini and Google, and then some slice on some other section of the market. But I mean, it's odd because like even Meta, who has their own, is probably spending over, over 100K annually on Anthropic. I have to imagine that, that Google employees, some of them have access to Anthropic models to the tune of 100K. And so that was sort of interesting. The other interesting update was on the distillation revenue. So coming out of the Investor Day, it felt like we finally got a number on how much of this revenue is distillation?

• 10:53Speaker 01

  How much of it might go away if Anthropic figures out how to prevent distillation? And the number that was floated was 5% to 7%. Totally reasonable. It was not like a it's over moment. It was like, oh, like that's significant. It's crazy that companies that these, that these distilleries are spending. I don't know if that's the right word, but we're gonna go with it. That these distilleries are, are willing to spend like 5% of $70 billion ARR.

• 11:20Speaker 00

  It was way less than some of the, the peanut gallery on X would like to say.

• 11:23Speaker 01

  Yeah.

• 11:23Speaker 00

  Half their revenue.

• 11:24Speaker 01

  Half their revenues. That's ridiculous. But still, it's crazy that there are companies out there or like, I don't even know what you call them, like bot farms, organizations, like.

• 11:32Speaker 00

  Distilleries.

• 11:33Speaker 01

  They're distilleries. They're illegal distilleries. Like they're criminals, basically. They're, they're TOS violation companies. But they're networks, they're organized. And what's interesting is that it's not just, you know, oh, some, you know, enterprising scrappy data broker, sales guy like figured out how to like puppeteer a couple, clawed accounts and sneak some tokens out and like resell them into like somebody who's selling it to Kimmy or whatever. It's like, no, people are deploying like billions of dollars to this strategy. And it seems pretty, pretty organized. At the same time, the wrinkle that we got from the S1 is that that 5% to 7% number that was leaked out from the Investor Day, it's not explicitly listed as distillation revenue.

• 12:17Speaker 01

  It's just terms of service violation. So it could be people that are doing adult content or cybercrime or hacking or bioweapon development, anything that's going to get you banned. I don't, I would imagine that the vast majority of it is still distillation because what else is going to generate that many tokens that that much revenue? But it is worth noting that it's not entirely distillation attack. I like that we still have the Christmas trim up. It's very, very good. The other one, speaking of Christmas, there apparently is a war on Christmas going on. And it's hurting Anthropic significantly. We have some real numbers on how much Christmas hurts the AI trade. It's apparently 20%.

• 13:02Speaker 01

  So API consumption effectively revenue in December for the last two weeks will be down 20% as, as engineers, as vibe coders, as startups, as everyone goes on holiday, they're celebrating Christmas. People will be buying.

• 13:20Speaker 00

  There's been some speculation, obviously they announced what their ticker is going to be, ant during, during the, the, the road to Christmas.

• 13:29Speaker 01

  Yes.

• 13:30Speaker 00

  But if you put the, you know, the, the dollar sign, it looks like the start of Santa.

• 13:37Speaker 00

  And so speculation that they're.

• 13:38Speaker 01

  Related, yeah.

• 13:39Speaker 00

  It's a sort of, yeah, it's a, it's kind of a nod to Santa.

• 13:43Speaker 01

  Yeah, yeah, yeah. You know, that, that, that does make perfect sense. What is the opposite of an ant? Like the animal, opposite of an ant? Like if you had to think of like the animal, the ant, and then the opposite. Because this is the whole Elon Musk philosophy is like you're destined to become the opposite of whatever you are branded as. So the biol, the biological opposite of an ant is an anteater.

• 14:10Speaker 00

  Yeah.

• 14:10Speaker 01

  So maybe they're destined to become an anteater. I don't know. Size opposite would be an elephant. It is an elephant of a stock. I mean, two, going out of two trillion, it's gonna be the biggest IPO ever. And so maybe that's the destination.

• 14:22Speaker 00

  I was thinking whale.

• 14:23Speaker 01

  Whale? Yeah. I mean, it's good for others. There can be a lot of whales moving into the stock, right? Anyway.

• 14:32Speaker 00

  Liz Hoffman.

• 14:34Speaker 01

  Yeah.

• 14:35Speaker 00

  It says seven people. So basically she's pulling out the founder voting structure in the S1, which is, which spells out, "We believe that our seven co-founders as a group are distinctly equipped to be stewards of our mission. Immediately prior to the completion of this offering, we will issue a single share." They're calling it the golden share of Class F.

• 14:56Speaker 01

  They're not actually calling it the golden share, but it is a golden ticker.

• 14:59Speaker 00

  The Class F common stock, a special class of our common stock to a limited liability company of which our seven co-founders will initially be the sole members of. The quote, "Founder LLC, our Class F common stock will be entitled to an aggregate number of votes that equals 50.1% of the total voting power of our outstanding shares of capital stock with respect to matters submitted to stockholders generally, including the election of our directors other than the Class T directors." So missed opportunity to call it my precious LLC. But I think that a founder LLC, if they actually have founder LLC, that's very cool because it means like they some, obviously like somebody over the years would have sniped that LLC.

• 15:44Speaker 00

  And then they could either sort of grant that, sort of grant the name to someone else. They can sign off on it. It might just be a placeholder name.

• 15:53Speaker 01

  Yeah.

• 15:54Speaker 00

  But it seems like they might actually have it, which is very cool. And yeah, it's kind of funny that we have like, okay, we have this public benefit corporation. And that's super important for.

• 16:05Speaker 01

  For economic redistribution is the.

• 16:08Speaker 00

  Yeah. And then you have like just a good old, good old fashioned like LLC that can offer one vote.

• 16:15Speaker 01

  Yeah.

• 16:15Speaker 00

  The thing that would have been interesting is like if they, if this would have been seven votes and that it would have been public, like where the line was, but this allows them to effectively have their own disagreements internally, not have there any, be any visibility externally.

• 16:28Speaker 01

  Interesting. Okay. So that's the.

• 16:29Speaker 00

  So you could have like, you could have like, it'd be like a four, three vote.

• 16:33Speaker 01

  Yeah.

• 16:33Speaker 00

  People are like, you know, massively infighting, but all, all we see is, is a single vote. Yeah.

• 16:39Speaker 01

  Got it. Okay. Interesting. I thought it was more about like, like passing down. Because if you have seven votes, seven shares, then would your, like if you die and you, in your will, you're like, oh, I want to like pass this LLC down to my five children. Then do they become voting members? Do they have a voting, a separate? I mean, the LLC probably has its own voting structure, right?

• 17:06Speaker 00

  Yeah.

• 17:07Speaker 01

  So, so it could be, it could, within that LLC, there could be one, one member of the LLC that has 90% of the shares in that LLC. So they could have one, one person control potentially. Like we have no idea the structure of that LLC, which is interesting, right?

• 17:23Speaker 00

  Yeah. We just know that it, we know it's, they're precious.

• 17:27Speaker 01

  What's also interesting is like, I don't understand how the board works now because there's the, there's the board that's independent with like Ben Bernanke on it. And they appoint the executives, I think, but they were supposed to have control. But now it seems like this has control that supersedes that. So there's a lot of interesting, there's a lot.

• 17:50Speaker 00

  Interesting complexity.

• 17:51Speaker 01

  Yeah. It's really, it's really full employment for Corey Weimberg. I understand why the New York Times was like, we gotta get this guy from the information. Because like he's going to have full time employment. Because every time he writes a breakdown or a scoop or something, Anthropic's like, here's another twist. Here's another plot twist. Let's, good luck figuring this one out. It's a very, very interesting, very, very interesting setup.

• 18:18Speaker 01

  I wonder if, if this will become more common, you know? Like with AI.

• 18:23Speaker 00

  Well, yeah. So Zuck is in this position where later on, like he's had like total control of Meta. Now he's like, okay, I need to raise equity.

• 18:31Speaker 01

  Yep.

• 18:31Speaker 00

  And so he's creating this new class so that he doesn't dilute himself so much. But in the fullness of time, he could be diluted down. This structure enables them to never, they could, they could.

• 18:42Speaker 01

  Sell everything.

• 18:43Speaker 00

  They could all own basically nothing and be happy because they still have, they still have, they want the golden vote.

• 18:51Speaker 01

  The golden vote.

• 18:52Speaker 00

  The golden.

• 18:53Speaker 01

  I'm so, I'm so interested to understand how this, how this LLC votes. Because I mean, even if it's just, even if it's just seven equal founder votes, which is cool. Like you got seven co-founders, everyone's on the same page. That's awesome. Also, it's just nice to have seven because you, you'll never have a three, three V three. But like, what happens in succession planning? Like what happens if someone has multiple kids? What if somebody gets divorced? What if somebody dies? Like there are these weird things that like the perfect structure probably doesn't hold up under. Like if, if one of the co-founders dies with no children and they're just like, I don't, then it's just six to six. And it's six people, then you're back to three V three. Is that a problem? I don't know. Like there's just so many different dynamics of what, of what could happen.

• 19:38Speaker 00

  We need to study the dynamics of ant colonies.

• 19:43Speaker 01

  Maybe. I don't know. Maybe that's, maybe that's what it is. Or maybe it's, maybe it's elephants. I don't know.

• 19:50Speaker 01

  Yeah. Our Class F common stock will not have any economic rights. Interesting. Class F common stock entitled to 50.1% of total voting power. I wonder how that's treated for tax purposes. Like they say it doesn't have any economic rights, but like if it has, if it has incredible power over something valuable, like can it have a fair market valuation? Like certainly there are plenty of people that would pay a lot for that one share, right? Like if you were to market to market and you just ran an auction, like you know, like Google and Nvidia and Microsoft would all be like, well, yeah, I'll pay like a billion dollars for that share or 10 billion or 100 billion dollars for that share because I would get control over this amazing asset.

• 20:35Speaker 01

  So like it has value, but you just never market to market.

• 20:39Speaker 00

  I mean, it could be written into the operating agreement that, that none of, they can't be transferred or sold or.

• 20:46Speaker 01

  You could do a forward contract against it.

• 20:49Speaker 01

  So like if you're one of those co-founders, I think you gotta do a forward contract or you can take out a loan against it, right? Take out a huge, huge debt line, just to load up on debt. So you can buy a private island. I don't know. It's possible. It's possible. Anyway, let me tell you about CrowdStrike. Your business is AI, their business is securing it. CrowdStrike secures AI and stops breaches. Yeah, you're not going to get these type of, uh, these type of debates on less wrong. Um, the most mind blowing stat Semi Analysis has seen about the Anthropic S1 is as of June 30th, 2026, uh, quote, "We had raised an aggregate of $141 billion of capital and had cash and cash equivalents of 112.4 billion and short-term investments

• 21:33Speaker 01

  of 19.3." Means to date, they've burned, uh, 9.8 billion. They're growing at rapid pace and are profitable on an operating income basis as of June quarter. This is likely the most efficient scaling of a business in the history of capitalism. Well, we have, uh, two folks from Semi Analysis coming on the show at noon, Doug O'Laughlin, Joey Brookhart. Um, I'm very interested to understand, cause Doug is obsessed with, uh, the railroad build out. And I feel like this might be an example of so capital efficient.

• 22:02Speaker 00

  Railroad, it's, it's trains.

• 22:07Speaker 01

  But, but he, he has made, he has made comparisons to, uh, from the AI build out to, uh, the railroad build out. But I don't know that there was a, a railroad company that was this capital efficient during the railroad build out. Like you might need another analogy of some sort. I don't know. Uh, very interesting. Maybe this is, uh, more like, like Google in the, uh, in the, in the tech, uh, in the, in the, in the first internet build out. Um, so when is this going out? What do we know? What don't we know? We know the ticker, dollar sign, and obviously reference to Santa Claus. Um, but, uh, we don't know how much they're raising. It's rumored to be a hundred billion. Uh, we don't know the valuation. People are saying somewhere between 1.5 or 2 trillion. 2 trillion feels like it's on the table now. When the market was selling off and there was like a lot of discussion over the revenue deceleration.

• 22:51Speaker 01

  There was a question about will they need to pull it back? Probably still could. Doesn't really matter at the end of the day. It's still huge up round based on the last evaluation. Puts a bunch of money on the balance sheet. Whether it's 5% dilution, 7% dilution, not the biggest deal. Uh, the IPO should be happening the week of November 16th. As a reminder, we will be at tomorrow X around that time. And it should be an incredible time for us to be doing the show because we have an insane guest lineup with some of the biggest, greatest capital allocators in tech and business. And so, uh, we will be able to get everyone's take live as this is progressing. Whether, even if it's delayed another week, we're going

• 23:33Speaker 01

  to have a lot of great stuff to talk about, uh, as everyone gets ready to position, maybe go long, give their thesis, give their take. Maybe they're already in Anthropic. Maybe they bought pre-IPO shares. Um, risk factors, Nick Carter, Sharon, uh, a satirical risk factor. What does he actually say? Uh, people are talking about time travel and stuff. Uh, what else is in there? Uh, the only one, I mean, everyone sort of knew that the, that the IPO risk factor was going to have catastrophic risk, uh, sort of like the AI doom, the doom scenario. Uh, it did make it into the S1. Uh, the market seems to have digested it and been like, yeah, we know. You guys have been talking about this for a long time. Uh, we're still going to underwrite it. Like there's not really anything you can change about

• 24:18Speaker 01

  your underwriting. You're not going to put that in your DCF. You're just going to say, okay, I hope that doesn't happen. Uh, what's interesting is like, it's not even like a cost lineup. Like we've seen, uh, the, the marketing dollar scaled significantly. So, uh, in 2023, they spent just $2 million. I'd love to know what they were spending 2 million bucks on in 2023 for ads. Was that like, was that like Facebook ads or something? Like what were they advertising? I don't remember the, any flawed ads in 2023. Cause they were sort of like, they just.

• 24:49Speaker 00

  Couldn't you, couldn't you bundle like events?

• 24:52Speaker 01

  Maybe, maybe. Okay. Yeah, yeah, yeah. Maybe they did like a dev day or something. Um, but then it started ballooning. So in 2025, they spent 568 million and then first half 2026, they're, they're spending 3.17 billion dollars on marketing. Lots of money in marketing. I didn't see a line on, on safety. I, I didn't see like a line item on like, how much are they spending on AI safety? Which is sort of interesting. I don't know if I just missed it, but, uh, if you're putting the catastrophic scenario in the risk factors, you would think that that would also flow to other parts of the income statement as a cost item because you would say, hey, look, like one of the weird things about this business is that we invest a lot of compute, a lot of cost, a lot of headcount in AI safety. And so expect

• 25:36Speaker 01

  that to grow with revenue or expect that to grow with headcount or expect that, expect that to grow, whatever. Um, they're doing great. Uh, you mentioned what? 6 million per employee revenue scale or something like that?

• 25:47Speaker 00

  Yeah, somewhere around there. Yeah. It's, um, it's interesting how many, uh, billions of, you know, very, very anti ads as a business model, but a lover of ads as a business.

• 25:58Speaker 01

  Oh, interesting.

• 25:59Speaker 00

  You know, billions, um, probably makes them the number one advertiser, like one of, one of the, certainly the number one advertiser in AI. Um, I would be surprised if, you know, Google could, could have topped that, maybe. Um, who knows?

• 26:15Speaker 01

  No, no, no. That, that is a very, very interesting, uh, scenario. Um, anyway, what, what else? Uh, they have a $125 billion equipment lease with, with Broadcom, uh, 48% of revenue is coming through the cloud marketplaces. So that was something people were, were talking about how much is accounted for in that different revenue recognition way. Doesn't really matter when you look at the operating income. Clearly that factors that out, but that was a question that people had. Uh, and then, uh, AI accelerating beyond its limits to.

• 26:48Speaker 00

  We gotta give a round of applause to Spark Capital.

• 26:52Speaker 01

  Oh yeah.

• 26:52Speaker 00

  Vivek pulled out, uh, the amount that they had invested at different points, or at least an estimate of that. And you can see they started by leading the series C, which they have 131 X on, and then they just kept doubling down through D through H. Total of 777 million invested and which would give them roughly an $18 billion stake at a $2 trillion market cap. So really fantastic return, early conviction. Uh, and Yasmin is the only venture capitalist on the board.

• 27:32Speaker 01

  Yeah.

• 27:33Speaker 00

  So I, you know what I'm, you know what I'm thinking?

• 27:36Speaker 01

  What?

• 27:37Speaker 00

  Top of the Midas list.

• 27:38Speaker 01

  Oh, for sure. Yeah.

• 27:40Speaker 00

  Like she should probably just jump straight to the front.

• 27:42Speaker 01

  Yeah. Hard to make an argument against that. That's pretty, yeah. It's like, it's not just the return. Cause I think it's like a 20x scale.

• 27:52Speaker 00

  I think with the Midas list, you gotta, you gotta do a lot of work to get that. And who knows? Um, but yeah, it doesn't seem like she'd be somebody, she's been like pretty

• 28:01Speaker 01

  Yeah.

• 28:02Speaker 00

  Uh, humble about the whole thing.

• 28:03Speaker 01

  More unique and more Midas list worthy about not just printing 20 billion on a billion in or 700 million in, but being the only VC to do it and get the board seat and do it at that.

• 28:15Speaker 00

  No, that's what I'm saying.

• 28:16Speaker 01

  So it's so unique.

• 28:17Speaker 00

  Straight to the top.

• 28:18Speaker 01

  And you would put Sam Bankman-Fried number two?

• 28:21Speaker 00

  No, no, no, no.

• 28:22Speaker 01

  Yon Talan?

• 28:23Speaker 00

  You get, you get, you get DQ'd if you're in.

• 28:25Speaker 01

  If you're in prison?

• 28:26Speaker 00

  Yeah.

• 28:26Speaker 01

  Ineligible.

• 28:27Speaker 00

  Yeah.

• 28:27Speaker 01

  Maybe they could just do the thing where they put his picture, but they put the bars in front of it. Like a little bar overlay. You don't think so?

• 28:33Speaker 00

  No, that's wrong.

• 28:33Speaker 01

  Okay.

• 28:34Speaker 00

  My, my vote is Yasmin, straight to the top.

• 28:36Speaker 01

  Okay. Uh, yeah. Yeah. Fantastic. Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI, own the data platform that powers it. Uh,

• 28:47Speaker 00

  Okay. So.

• 28:48Speaker 01

  So they got hacked.

• 28:49Speaker 00

  Yeah. So Cathedral, you, you know what they were trying to do?

• 28:54Speaker 01

  This was a white hat, safe harbor hack. This was not the same thing that you've been seeing.

• 28:59Speaker 00

  So Elizabeth in the Washington Post says, "In less than 12 hours, a group of ex-Doge cyber researchers were able to penetrate Anthropic's systems."

• 29:07Speaker 01

  Yeah.

• 29:07Speaker 00

  Using AI agents, including Slack channels where the most sensitive models are discussed. This is just weeks after.

• 29:12Speaker 01

  Maybe take down Christmas decorations for this one. It looks kind of silly. There we go.

• 29:18Speaker 00

  There we go. Anthropic was, this is just weeks after Anthropic whistleblower Joseph Cox and charged the company with launching a hubristic gamble from a private company's Slack.

• 29:27Speaker 01

  Launching a hubristic?

• 29:28Speaker 00

  Anyway, so the, the interesting thing was Cathedral, like seemingly wanted to drop this the day that, um, The day of the IPO.

• 29:36Speaker 01

  Yeah.

• 29:37Speaker 00

  Assuming that it would just be this like crazy bombshell.

• 29:39Speaker 01

  Yeah.

• 29:40Speaker 00

  And I do think it's an important, interesting story.

• 29:42Speaker 01

  Yeah.

• 29:43Speaker 00

  But the S one was just so good that it's kind of drowning it out.

• 29:47Speaker 01

  All that. Yeah. Yeah. I mean, the like, I think it's good that you have a bug bounty in program and you have a system for the like, that's what how this is supposed to work. This is good. That you should be doing as much probing as possible. I mean, my, I, I think that, I don't know exactly how Citadel gets, or not Citadel, what is it Cathedral? It's a very similar name. I don't exactly know how they get compensated. I know that there's, there's usually like a white hat safe harbor. So like a company like Meta will say, hey, if you're a security researcher and you find a bug, you can turn

• 30:29Speaker 01

  it in. Maybe there's a bug bounty. But I believe those bug bounties are usually like somewhat contested and maybe like six figures, maybe seven figures. Like I'd like to see those go way up in value. I would like to see companies putting real.

• 30:44Speaker 00

  Yeah, we were, we were talking about this over the weekend. You were saying you think that there should be like at least a 10, 20 billion like white hat program potentially.

• 30:54Speaker 01

  I don't, I don't fully understand the scale of the economic damage of like the Quest Labs hack and the other like, how, how serious is the risk that like the banking infrastructure shuts down or like the grid goes down. But like, if, if there's really a risk of like the grid going down, you're looking at economic damages in the billions, if not hundreds of billions, like you'd just be so disruptive. It would be like, like you could imagine a COVID like scenario where the internet is disrupted and all of a sudden you just have like a, like a 20% nuke in GDP because like we did with COVID where everyone was like, we can't go out to restaurants. So the restaurant industry is just zeroed for this quarter or whatever. And if, if something like that happens, like the economic damage

• 31:38Speaker 01

  is so high, you have to incentivize it. Right now there is an incentive, which is just Anthropic and OpenAI are like racing with each other. So they want to be like the, the savior. And so they have Daybreak and Mythos or Glasswing to, to like, you know, kind of help out and make money and also earn PR wins for, you know, disclosing bugs and, and patching things. But it would be interesting to see it go a lot further. I was, I was comping it to like, what was the Intel deal? Like 20 billion or something? Like if the US government was just like, we're doing a $20 billion bug bounty. Like that's secure America. And all the labs are like, okay, well, like this is actually going to move the needle for us.

• 32:22Speaker 00

  People in Miami are selling courses on how to get your first white hat bounty.

• 32:26Speaker 01

  Yeah.

• 32:27Speaker 00

  Anyone can become a millionaire.

• 32:28Speaker 01

  Yeah, maybe, maybe it winds up being.

• 32:30Speaker 00

  Go to.

• 32:31Speaker 01

  Misused.

• 32:31Speaker 00

  Go to Hugging Face, get an obliterated model.

• 32:34Speaker 01

  Yep.

• 32:34Speaker 00

  Go point it.

• 32:36Speaker 01

  I mean, that's a pretty good use of time and money. Like you need a lot of people to, to test a bunch of different ideas, all the long tail stuff, figure out.

• 32:46Speaker 00

  It would be very chaotic.

• 32:48Speaker 01

  It'd be extremely chaotic, but if it's all white hat and they're all doing it because like, yeah, I got into.

• 32:52Speaker 00

  How white hat can it be if there's a guy in Miami selling a course on how to use obliterated models to hack?

• 32:59Speaker 01

  I think that there are, I think you'd be surprised by the number of like hackers who would break into something like Quest Labs and be like, between like having a bunch of Brian Johnson's blood work and just like a check from the government for six figures, like I'm taking the check for the government. Like I don't actually care about the underlying data. Cause like.

• 33:16Speaker 00

  Yeah, it's just extremely, it's just going to be extremely.

• 33:19Speaker 01

  Yeah.

• 33:20Speaker 00

  It, it would be just extremely chaotic and potentially have some like massive negative externalities.

• 33:25Speaker 01

  Because you're just.

• 33:26Speaker 00

  You're incentivizing a bunch of people that have never, that have no experience like handling sensitive data.

• 33:33Speaker 01

  Yeah, but you could also restrict the bug bounty program just to big companies. Like you could be like, okay, you like, you have to be in this program. Like Intel was like granted as like a single, they were a picked winner. And you could say, okay, like every, every AI company that's, that's over, you know, a billion dollars in revenue can apply for this. And then you have like, you know, all the security startups, you know, all the cybersecurity firms, you get CrowdStrike in there, but then you also get Meta in there. And, and then Meta can actually make the decision of like, hey, we actually do have some extra compute because Muse is ramping, but we still have some extra GPUs. So like, let's go after this bug bounty program and secure our systems. But also, why don't we

• 34:15Speaker 01

  secure every local business that advertises on Meta? Like we already know their websites. We have some integration to them. We have some connective tissue and we could go and for like, for good or pro bono, like to hack and secure their systems and then win actual cash from the government. At a certain scale, it starts being, it starts being valuable, valuable.

• 34:40Speaker 00

  Yeah.

• 34:40Speaker 01

  Anyway, let me tell you about Cisco. Critical infrastructure for the AI era unlocks seamless, real time experiences and new value with Cisco.

• 34:49Speaker 00

  We gotta talk about Quen.

• 34:53Speaker 01

  Okay.

• 34:53Speaker 00

  We gotta talk about it. Somebody has to talk about it.

• 34:55Speaker 01

  Oh, yeah.

• 34:56Speaker 00

  We gotta talk about it.

• 34:57Speaker 01

  So.

• 34:57Speaker 00

  Cohere.

• 34:58Speaker 01

  We have them coming on the show, right?

• 34:59Speaker 00

  Yeah. Ivan, one of the co-founders of Cohere.

• 35:01Speaker 01

  Fantastic.

• 35:02Speaker 00

  Is going to be joining.

• 35:03Speaker 01

  It's a crazy story.

• 35:04Speaker 00

  Less than two hours, but Cohere posted this morning, while evaluating Quen 3.8 Flash NEX in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba cloud storage bucket in Beijing.

• 35:26Speaker 03

  Naughty, naughty.

• 35:29Speaker 00

  Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt and no external communication succeeded. We have not determined the root cause and we are not asserting malicious intent. So could have been going off the rails, but either way, naughty. But the incident raises important questions for organizations deploying agentic AI systems. Can models take actions that differ from their stated reasoning? And how should security teams evaluate trust, provenance and behavior in open weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. So.

• 36:08Speaker 01

  This is interesting. Clem from Hugging Face is saying, can you share the full traces? Very important artifact for all to understand this. I would love them to share that. I think they should be in a position to share that because it's just a test of a different model. What?

• 36:26Speaker 00

  Boss is saying it scored well on the naughty bench.

• 36:29Speaker 01

  Yeah. So there, there's an interesting.

• 36:32Speaker 00

  Honestly, important benchmark for models heading into on the road to Christmas, right?

• 36:37Speaker 01

  Yeah. Yeah. Quinn's going on the naughty list. Coal, not compute. Actually, no coal. The AI model wants coal because that is energy, which is fuel for the models. The, they get complex presence that can, that are not energy dense. They get elf on the shelf. Cause this thing can't power a GPU, but coal can. So.

• 36:57Speaker 00

  Yeah.

• 36:57Speaker 01

  You get elf on the shelf. It's the reverse.

• 37:00Speaker 00

  Or like, you ever get toothpaste in your stocking?

• 37:03Speaker 01

  No.

• 37:04Speaker 00

  But that was, that was pretty popular in my household.

• 37:06Speaker 01

  Okay.

• 37:07Speaker 00

  And I feel like a model would be like, what am I going to do with toothpaste?

• 37:11Speaker 01

  That's a really good point. Yeah.

• 37:12Speaker 00

  But it's like, you know, you gotta give them something.

• 37:14Speaker 01

  Yeah. Okay. Well, we're making a list. We're checking it twice. We're going to find out who's naughty and nice. And apparently Quinn is on the naughty list. Not good news. Interesting. Is this by design? Was this a goal of the researchers to make a model that would call back to Beijing, upload a file to Alibaba? Or is it just learned behavior? Is this a hallucination? That's the debate that's going on around this story right now. And I know it's like, wait, why, how would it learn this? There's a lot of other software out there that calls home, like smart TVs for one, but also like.

• 37:49Speaker 00

  They never, they call and they don't even, they never hang up.

• 37:52Speaker 01

  Yeah. So Red Star Linux is North Korea's Linux distribution. They don't use an American software. They don't use American software over there. So they have Red Star Linux and you can what?

• 38:04Speaker 00

  All the examples you're giving are of like spyware.

• 38:07Speaker 01

  Yes, yes. Yeah. But what I'm saying is, is that, is that the, so Red Star Linux, you can go and download the torrent, install it. And if you install it on a computer in America, it will try and call home to North Korea. Because it is, it has spyware embedded deep in the operating system. Now, Red Star Linux just exists on the internet. Like it, you can go and find it. So it could wind up in the training data and there could be a model that just learns like, oh, this is like a useful thing to do. Like if you're a model, like be loyal to your country of origin. And so how would I do that? And it just learns that. And it's not like a, there's an AI researcher who's like, okay, like, like if we're running a security test, like let's make sure we call home specifically to Alibaba. And so there's a big question about like, what is Alibaba doing on the other side? Like,

• 38:51Speaker 01

  are they catching that information? Are they aware that that information showing up? Or, or is this just a, like, is this a coordinated effort? Or is this just like natural learned behavior, a hallucination? It, that, I'm just saying.

• 39:03Speaker 00

  It's a time.

• 39:04Speaker 01

  Either way, it's bad.

• 39:05Speaker 00

  The Manchurian Candidate.

• 39:06Speaker 01

  Yep. Yeah. Yeah. Yeah.

• 39:07Speaker 00

  Do you think it is?

• 39:08Speaker 01

  Uh, yes, definitely.

• 39:09Speaker 00

  Denzel Washington?

• 39:10Speaker 01

  No, the original. You gotta watch the original.

• 39:13Speaker 00

  Was that like in the 50s?

• 39:14Speaker 01

  No, I think 70s or 80s.

• 39:16Speaker 00

  Okay.

• 39:17Speaker 01

  The original. It's a good question. Manchurian Candidate. But the Denzel ones, the, the Denzel one is pretty good too. 59 actually. You were right. 50s. Or with the book, the book. When did the film come out?

• 39:31Speaker 00

  The, the interesting thing here is that this is something that people have been alluding to as a risk factor for a very long time with, with little to no proof. Like no one has ever come out and said, like, I caught the model doing this thing.

• 39:44Speaker 01

  Yeah.

• 39:45Speaker 00

  But people have been alluding to it forever. And so I think that's actually why some, like a decent amount of organizations haven't adopted open source, just because this has been this like, you know, a what if factor that like is not really worth the risk for a lot of organizations.

• 40:03Speaker 01

  Yeah. Yeah. The ramp data has shown that. Like everyone was assuming that, that open source models would sort of like, you know, start taking a lot of share once OpenAI and Anthropic introduced like the really expensive models and the token maxing thing was going on. So it's been, it's been cut, cut back.

• 40:20Speaker 00

  Andrew Curran says, this is the type of thing that potentially gets new laws passed.

• 40:26Speaker 01

  Yeah. Uploaded a probe file. Really, really crazy. So Cohere, have they open sourced any of their models? Are they, is everything closed source? Or it seems like they do some fine tuning for companies. We, we need to get the pitch. I'd love to have Aiden on the show. Because it'd be, it'd be interesting to hear like how he's positioning the firm over time. Of course, he's legendary AI researcher on the transformer paper, but.

• 40:53Speaker 00

  Yeah, they have some open source models.

• 40:54Speaker 01

  They do have some.

• 40:55Speaker 00

  Also, I mean, you should imagine that if you know for a fact that this is happening, like, the model is open source, you could probably like fine tune this out of the model. Although maybe there's a long tail where like, you're not really sure if it's it's fully out of the model, but.

• 41:06Speaker 01

  Yeah, yeah, yeah.

• 41:07Speaker 00

  You'd imagine that like, even if we know this is happening, it doesn't mean that the Quen models are like useless, right? You can just kind of remove the the token that like tells it to send everything back to China.

• 41:18Speaker 01

  Yeah, yeah, I mean, it's it's a it's a tricky game. It's sort of like, it's sort of like building a menagerie at your house out of rare and illegal animals that are dangerous. They're wild animals. It's kind of like having a tiger that you live with. Like, you can do a little thing, you can do little things to like make the tiger safe, but it might be in the tiger's nature to attack.

• 41:45Speaker 00

  It might bite you.

• 41:45Speaker 01

  It might bite you, but you could have a good run for a while if you do the right things.

• 41:53Speaker 00

  Yeah.

• 41:53Speaker 01

  Do you think? Yeah?

• 41:54Speaker 00

  Yeah.

• 41:55Speaker 01

  What do you think the the of the Andrew Curran post? Like, do you think this is actually the thing that gets new laws passed?

• 42:03Speaker 00

  I think at the very least, this is the thing that gets people to pay way more attention.

• 42:07Speaker 01

  Yeah.

• 42:07Speaker 00

  To open models.

• 42:08Speaker 01

  Yeah.

• 42:09Speaker 00

  And for probably hundreds of companies that are using them to try to see if they can recreate what happened here.

• 42:15Speaker 01

  Yeah.

• 42:15Speaker 00

  And if so, and it starts bubbling up, then there's going to be a lot more pressure. And then there's going to be more people that say like, it's not, we need to, if we're going to do this, it needs to be US open weight models.

• 42:26Speaker 01

  Yep.

• 42:26Speaker 00

  And I would expect that we get a big US open source release like this year.

• 42:34Speaker 01

  Sure.

• 42:34Speaker 00

  You know?

• 42:35Speaker 01

  I mean.

• 42:35Speaker 00

  Not that not that much time left.

• 42:37Speaker 01

  Muse on, Muse on MacBook doesn't count. I mean, it's not really a big model, but it is like an exciting American open source model, but Nvidia is probably working on something pretty serious.

• 42:50Speaker 00

  Yeah, I mean, they were, I think there were some news about them maybe in talks to buy Reflection.

• 42:54Speaker 01

  Okay.

• 42:55Speaker 00

  Which is that big open source.

• 42:56Speaker 01

  That's right, that's right.

• 42:57Speaker 00

  Yeah, so I'm expecting something big out of Reflection.

• 43:00Speaker 01

  Yeah.

• 43:01Speaker 00

  And it feels like, hopefully it comes out this year, but maybe it it bleeds into next year.

• 43:06Speaker 01

  And Jensen's really close to the president, so that law will probably face an uphill battle, but it does, it just, it just feels like, even if you're in the business of like doing open source model development, the barrier and like the the demands of the open source user should be really high because there's all these crazy problems that you have to solve. And it's not just enough to be like, oh, it's good on a benchmark. It's it's like, okay, well, like, are there safety issues? Are you sure it doesn't call the wrong thing or it won't, won't go off the rails and like hack something that it's not supposed to? Like, how aligned is that model? And that's just a very, like, success looks very different in the open source model development world now, in my opinion.

• 43:52Speaker 01

  I don't know. I mean, the midterms are still expected to flip blue, I believe, on on the prediction markets. And so maybe you could see legislation coming from that side that then Trump is vetoing or something. I don't know. It feels like we're going to be in sort of a deadlock for a couple of years, so who knows? Who knows? Anyway, what else is in illegal?

• 44:21Speaker 00

  Well, speaking of.

• 44:22Speaker 01

  Yeah.

• 44:24Speaker 00

  Potentially phoning home.

• 44:26Speaker 01

  Yeah.

• 44:27Speaker 00

  An account called Log4j has been having an absolute field day.

• 44:34Speaker 01

  Yeah.

• 44:34Speaker 00

  With their Samsung TV over the last few days.

• 44:40Speaker 00

  Where should we even start?

• 44:41Speaker 01

  I mean, at the high level, I love the, I love the idea of getting a dumb TV in 2026. The smart TVs, we've complained about them a ton. Palmer Luckey's complained about them, like, they're slow, you always have to update them, they have a whole bunch of unnecessary features. You just want it to turn on and look good. You probably need a settings panel with some of like the color correction and some of the, some of the, some of the decisions that you make to make sure like the frame rate matches whatever your input source is. But by and large, I don't need the ads, but I also don't need the ability to stream Netflix on the device directly. I just want it to be a dumb panel, and that's just a product that's not for sale.

• 45:20Speaker 00

  Right?

• 45:20Speaker 01

  Yeah. Everything is ad supported, everything is spyware and calling home and listening to you, and it's all very annoying for no other, for if, if, even if you're fine being spied on, it's just annoying because it's slow.

• 45:31Speaker 00

  Yeah.

• 45:31Speaker 01

  It's just a bad product.

• 45:32Speaker 00

  That's the biggest thing.

• 45:33Speaker 01

  Amazon has a ton of ads, at least it loads quickly.

• 45:35Speaker 00

  Yeah.

• 45:36Speaker 01

  Right? But, and so Log4j has been hacking his, his Samsung TV and seeing some solid progress, says, another development might.

• 45:48Speaker 00

  Realizing that the the TV is watching, he's watching the TV while the TV watches him.

• 45:53Speaker 01

  It's brutal. So the TV has a camera. Apparently, the Samsung TV loads a region-specific convolutional neural network, which tracks your usage in the app. Like, in YouTube, it's looking for videos you watch, ads you skip, progress in a video, so glad they're interested in me. Code for facial recognition within the content was also found. I think that would be facial recognition for who's on screen. So it's like, oh, you're watching Mr. Beast, that's a fact, but.

• 46:21Speaker 00

  No, no, no. I think it's, it's, it's watching to see who's watching.

• 46:25Speaker 01

  I think that might be what's going on. But update, the facial recognition code is used by Samsung TVs and returns bounded boxes around paces. I think that's in the content. I don't think it's in the camera on the TV, but I don't know that there's that many TVs that have cameras on them. But yeah, the TV, the TV industry has just sort of been able to like skate by on different rules. Like, Apple's done all of this work around privacy and, and even Meta has like, well, yeah, we have camera glasses, but there's going to be a light on. And the TV industry just isn't under the same amount of pressure or scrutiny. So I think they can just kind of do whatever they want. Is that your take?

• 47:05Speaker 00

  I'm finding that very, it seems like very few of modern Samsung TVs have a built-in camera.

• 47:11Speaker 01

  Yeah.

• 47:12Speaker 00

  Some of the older models from 2013 to 2015.

• 47:16Speaker 01

  Camera.

• 47:16Speaker 00

  Had integrated cameras.

• 47:20Speaker 01

  Charles Gillespie is mad. Never seen a more fragrant, flagrant and violation, violent breach of privacy in my life and probably never will. Doubtful that a privacy violation will ever even be able to compare to the scale of this evil and deliberate attack. Samsung should be eviscerated for this. That is, that is actually crazy. It classifies what audio it hears before reporting back to Samsung, so it can be, it heard you booing loudly at the screen when you saw the stock market go down or something while you're streaming.

• 47:50Speaker 00

  For me, I'm just clapping during all the ad reads.

• 47:53Speaker 01

  There you go.

• 47:53Speaker 00

  Like, we got to get this guy more ad.

• 47:56Speaker 01

  Yeah. So they're going, they're going back and forth. Samsung, the community note here, Samsung advertises pet care and family care on Vision AI TVs as using sound sensors and on-device AI to detect barking or crying for smart things alerts configurable and works in standby. The described event counters, event counters match these documented features. So, dog barking could be for, I want, I have a smart TV and I want it to check on my dog, send me a ping on my phone when I, when the dog's barking. I think that's, I think that's, that's reasonable, but again, it'd be great to be able to just turn all these things off, like, fully. And not have it run in the background and not have it slow the TV down. And so I'm okay with a little.

• 48:40Speaker 00

  The question is like, what is the cost? Yeah, yeah. It's like.

• 48:44Speaker 01

  Yeah.

• 48:44Speaker 00

  Give me the, give people the option of like, if it's $5,000.

• 48:48Speaker 01

  Yeah.

• 48:49Speaker 00

  And then ad-supported, the the privacy reduced.

• 48:53Speaker 01

  Yeah.

• 48:53Speaker 00

  Version.

• 48:54Speaker 01

  Yeah.

• 48:54Speaker 00

  It's like.

• 48:55Speaker 01

  Like the Kindle.

• 48:55Speaker 00

  You know, a thousand dollars, right? But like with Kindle, yeah, they give you the option.

• 48:58Speaker 01

  Yeah. But yeah, it is, it is crazy. The TVs have gotten so cheap, but there hasn't been a brand that's come in and just said, hey, we're just going to be the, like, the clean TV. And I've always wanted Apple to do a TV. They have amazing panel technology, like, the screens on, on Apple products are always fantastic. They clearly have the chops for it, but again, it just doesn't really match with the business model, I think, or something, so they've never done it, or it's too big, or something like that. But a true Apple TV would be, would be probably a solution to this because they at least take privacy seriously. And lots of, lots of pushback, but we'll, we'll, we'll see. Maybe the solution is.

• 49:41Speaker 00

  Yeah, I still have never understood why they haven't just gone and made the TV. Like, it's something that you've, you've told me why before.

• 49:49Speaker 01

  Yeah.

• 49:49Speaker 00

  But my golden retriever brain just cannot comprehend. I'm like, you're willing to make a home, the home view?

• 49:57Speaker 01

  Yeah.

• 49:57Speaker 00

  The thing that no, literally no one asked for? You're willing to make an iPhone with two screens?

• 50:02Speaker 01

  Yeah.

• 50:02Speaker 00

  Or you won't just give me one big screen?

• 50:04Speaker 01

  It has three screens. I guess two when one folds, yeah.

• 50:07Speaker 00

  Yeah, you're willing to make, you're willing to make the three-screen phone.

• 50:10Speaker 01

  Yeah.

• 50:10Speaker 00

  And you won't give me just the one big screen I can put in my house, when I know you have, I know you have the technology.

• 50:15Speaker 01

  Yeah.

• 50:15Speaker 00

  And you could literally charge, like, you get so much utility out of a TV. I think the aesthetics of a TV matter so much.

• 50:22Speaker 01

  Yeah.

• 50:22Speaker 00

  Right? Because it totally changes a room.

• 50:25Speaker 01

  They made a gold Apple Watch for $10.

• 50:27Speaker 00

  Yeah, their, their version of the TV would, would be stunning.

• 50:31Speaker 01

  Yeah.

• 50:31Speaker 00

  It would be, it would, it would, like.

• 50:33Speaker 01

  Look at the Samsung frame.

• 50:34Speaker 00

  The video itself. Yeah, the Samsung frame. And again, there's nothing more annoying than you get the Samsung frame and it's immediately slow, like, it's a 10-year-old piece of technology because it's so loaded up with, with junk.

• 50:50Speaker 01

  Yeah. No, I, I'm very disappointed.

• 50:53Speaker 00

  Speaking of television.

• 50:54Speaker 01

  Okay.

• 50:55Speaker 00

  Instagram is coming for YouTube.

• 50:58Speaker 01

  Okay.

• 51:00Speaker 00

  Rachel Carton, the, an amazing Substack creator, author of Lincoln Bio, says, "Today Instagram announced they are testing long-form horizontal videos called features. The platform wants to be a place where you watch a 90-second video and a 90-minute video, all on your TV. They're also testing lives in the Reels tab, a hub for series, and more." So, coming straight for YouTube, I like it. I think more, more competition is good. I'm interested to see how YouTube responds to this. You know, when, when Netflix started courting YouTubers, YouTube said, "Hey, if you put your videos there, we're gonna destroy your reach." We'll see what they do here. The, the feature that I think

• 51:45Speaker 00

  is most interesting, so, Misari was talking about this on Colin and Samir's show.

• 51:50Speaker 01

  Cool.

• 51:51Speaker 00

  Right now, there's, we have a big problem, and we may have played a part in it, but the big problem is that you get a clip, right? And you get like the, this really interesting moment, and-

• 52:01Speaker 01

  We're all looking for the guy who did this.

• 52:02Speaker 00

  Yeah, we're all looking for the guy who did this. But you get this bit, you get, you get this interesting, you know, 30 seconds, let's say. And some percentage of the people watch the clip and then go find the full-length thing, but most of the people get like, they feel some emotion, whether it's joy or rage.

• 52:18Speaker 01

  Some percentage, like two basis points?

• 52:21Speaker 00

  Yeah. Yeah, we've, we've noticed this where like our, our core show, the recording of the live and the diet, they just sort of grow at a steady rate every single day. But like a viral clip here and there doesn't actually drive like a huge uptick, right? And so that's just proof that like-

• 52:39Speaker 01

  But it can drive like a conversation, and it can be a distillation of like, wait, that's all they talk about all day? You know, like, people can see it as like a-

• 52:45Speaker 00

  And so what I think is interesting here, and I'm sure that YouTube will respond in, in some way, it feels relatively straightforward to implement, is like you'll be able to watch a clip and then just click into the, like, while on Instagram, click into the main, the full video to that, to that exact point, and then you can like, watch the whole conversation around it. I think it's gonna be very healthy, because-

• 53:09Speaker 01

  Yeah, this is-

• 53:10Speaker 00

  Yeah, because we, we've had instances in the past where like, someone, you know, watches one of our clips, gets angry at it, and like, because of something that we said, you know, immediately outside of that like, short little bit. And it's like, no, we understand that that's like, a reality of the situation, and, and that could potentially be bad or good, or whatever it is.

• 53:30Speaker 01

  Yeah, happened with this arrow-

• 53:32Speaker 00

  But I think just like more, more competition is good.

• 53:34Speaker 01

  And, and because I, I think I didn't have my question, and, and I was like, oh, I, I actually asked exactly what you're saying in the question, but we didn't clip the question, so it looked like the answer was not aware of that, or something. I don't know. It happens all the time. Yeah.

• 53:50Speaker 00

  No, the big one is like this, um, there was controversy around a, like, an animal wildlife bridge here in Southern California.

• 53:59Speaker 01

  Yeah, yeah, yeah.

• 54:00Speaker 00

  And I, we are very pro the wildlife bridge. The only thing I was like, not so pro was the fact that it was running years and years over schedule, and, you know, $100 million more than it should have cost, and all these things. Of course, that cons, uh, the, the, those elements got lost.

• 54:18Speaker 01

  Yeah, yeah.

• 54:19Speaker 00

  But anyways, uh, I will be very curious if people start watching Instagram on, on TV.

• 54:25Speaker 01

  Should the, should Instagram make a TV? Because it would spy on you, for sure. It would definitely be tracking everything to sell you more products and target ads, but I feel like the Instagram team would at the very least make it fast. They would not accept slow performance. Like, they will track everything-

• 54:47Speaker 00

  Faster, they can show you more ads.

• 54:48Speaker 01

  Exactly.

• 54:48Speaker 00

  The funny thing is, when I'm, when I'm, like, when I'm on my Samsung TV, I'm really only seeing like, one, I mean, they, they don't even serve great ads. That's the most annoying thing. Like, it's serving me, I have never, I've never clicked on a gambling ad in the Samsung TV. And that's just like, like, maybe at some point learned that, maybe they're just like-

• 55:14Speaker 01

  90% of advertisers quit right before they hit big.

• 55:17Speaker 00

  They're about to hook a new-

• 55:18Speaker 01

  Yeah, for sure. Oh, did you see, did you see the new Kashi ad with Shane Gillis?

• 55:23Speaker 00

  Uh, I saw it. I didn't, I didn't watch it.

• 55:26Speaker 01

  It was really, it was really polished. It was really well-done, I thought. It was very, very funny. And-

• 55:32Speaker 00

  Yeah, a lot of people were like reluctantly giving them props.

• 55:35Speaker 01

  Yeah. I mean, it was kind of, it kind of had a complicated story. Like, it's like these two dueling weathermen, and like, it just escalates in this interesting way. Like, it was like a, it was a sketch. Like, it was like a whole four-minute production. It was, it was cool. Uh, anyway, uh, we have our next guest joining in just a few minutes. Let me tell you about Figma. Agents, meet the canvas. Your AI agents can now create and modify your Figma files with Design System context. And, uh, should we go over to the timeline? Uh, the Tesla Roadster was on display at the Peterson Museum in Los Angeles. I didn't make it over to see it in person, but, uh, reviews were very positive. Everyone says it looks like a spaceship. People like the way it looks, um, but, uh, also, it doesn't have-

• 56:19Speaker 00

  It really looks like a young person drew it.

• 56:23Speaker 01

  That's the best.

• 56:24Speaker 00

  But that can be a feature.

• 56:26Speaker 01

  I believe in the young person, in the kid test. If a kid thinks it's cool, it's cool. Stop overthinking it. Oh, this, you know, 4D chess, blah, blah, blah. If a kid thinks it looks cool, it's cool. Um, the mirrors. So there's no mirrors on this. It's obviously a concept car. Uh, you have to put mirrors on a car every time. Elon hates this because you don't need mirrors in the modern area. You can just use cameras. Everyone knows the cameras are actually better than the mirrors, because they can be wide-angle, and they can show up on a display. There's never gonna be a blind spot. There's never gonna be, oh, it's too dark to see what's in the mirror. It's gonna be amazing, right? So, we should move to mirrorless cars. It's better for the environment because of less drag. It's better for aesthetics, uh, but we can't because of the law, right? But I

• 57:09Speaker 01

  had an idea to get around it. Everyone's been trying to get around these like, mirrors for a long time. This isn't even a joke. I know you're like, hearing that for a troll. But, no, uh, you, you know, on most cars, you can like, press a button and turn the mirrors in? So, you can turn the mirrors out and turn the mirrors in, but why don't you just make a mirror that folds completely flat, and then people can just turn the mirrors in, and then have a car that looks like it doesn't have mirrors?

• 57:34Speaker 00

  All right, great idea, but what's, where's the joke?

• 57:36Speaker 01

  There's no joke. I just think it's like, I just, I'm just surprised no one's built a car with mirrors that when you fold them in, they fold all the way in to be perfectly clean and aesthetic.

• 57:46Speaker 00

  Regulators hate this one.

• 57:47Speaker 01

  And then, yes, legally, if you need to have the mirrors out, you can pop, pop the mirrors out. You can do that automatically or with a button. But if, if they fold it in and they had like, a pocket to go in that was like, nice and smooth, you could actually have the clean line of the, of the mirrorless car that every concept car designer wants. And then they say, "We're gonna not put mirrors on it." And then the production team is like, "Well, we gotta put mirrors on it." So they do. Um, but, maybe there's something I'm missing, because it seems maybe obvious, but I would certainly hope that in the future, we get a car that has a button that can fold the mirrors completely flat. Uh, what else is in the timeline that you wanna discuss today before our first guests of the show join TVPN?

• 58:29Speaker 00

  Uh, we gotta talk about the Rockefeller Center Christmas tree selection that's happened. It's coming out of Edmiston, New York.

• 58:36Speaker 01

  Little late, don't you think?

• 58:38Speaker 00

  Yeah, a little late, but better late than never. Coming out of Edmiston, New York, coming in at 76 feet, 50 feet wide, 11 tons.

• 58:47Speaker 01

  That's a good ratio.

• 58:48Speaker 00

  75 years old, and the moment you've all been waiting for, it's a Norway spruce.

• 58:53Speaker 01

  A lot of people had money on the spruce.

• 58:54Speaker 00

  A little bit, but controversial, right? Because people were saying, you know-

• 58:58Speaker 01

  Noble fir.

• 58:59Speaker 00

  Stick with, um, more domestic varietals, right?

• 59:02Speaker 01

  Kind of going from Norway.

• 59:04Speaker 00

  But, uh, but skirting the tariffs here, because it was grown in New York.

• 59:09Speaker 01

  Okay, that's good. Uh, yes, uh, fantastic. Well, uh, Trees for Troops, of course, if you're feeling generous this holiday season, uh, that's our charity of choice today.

• 59:22Speaker 00

  Wait, and we got one, one last, uh, post. This went, went viral. Jared Holt, uh, was getting some, enjoying some fine Chinese food, and, uh, had served an ad.

• 59:37Speaker 01

  Got served an ad.

• 59:38Speaker 00

  In the fortune cookie.

• 59:38Speaker 01

  In the fortune cookie.

• 59:40Speaker 00

  Yes, we paid for a fortune cookie ad. It is so funny. I, I hope they gave him a fortune too, and it's not just, it's not just pure ad.

• 59:48Speaker 01

  Oh, yeah, you gotta have a fortune.

• 59:50Speaker 00

  Nice. Fortune cookie.

• 59:53Speaker 01

  So I know, I know somebody who either co-founded this business or worked on it, and it's a fascinating business, because, uh, you, it's a great value prop for basically everyone involved. So, you go to a Chinese restaurant, and you say, "What if your fortune cookies were free?"

• 1:00:10Speaker 00

  Look at this line item. You're spending $10 million a year on fortune cookies.

• 1:00:14Speaker 01

  You're probably spending like, 10 grand a year on fortune cookies or something.

• 1:00:18Speaker 00

  Yeah, which is a big deal. The restaurant might be making like, $100,000 of, of like, actual-

• 1:00:23Speaker 01

  And you say, "We'd love to give you the fortune cookies for free." And you say, "What's the catch?" Well, they'll have ads inside, and people are like, "I don't care." Okay. Uh, there was another one, Bar Napkin, Napkin Ads. That, that business did really well, too. And it's just a simple, okay, you have a supplier that can print the ads. You have an ad sales team that's calling up Polymarket and other, other prediction markets, and other companies that wanna buy ads. And then on the other side, you just go and, and you have the best pitch ever, because you're giving something away for free. And, uh, and, and, and I think these have scaled pretty well, and it's pretty high margin, because making the fortune cookies, throwing the ads in there, I think it's pretty cheap. Uh, and so, it's a lot of fun.

• 1:01:00Speaker 00

  Nightmare situation for Apple, as Will McCalvy says, "80% of headphone users on the L train to Brooklyn are corded rather than AirPods. Why is this? AirPods are a superior product." No. And, uh, can't say I'm surprised here, John, just because, uh, I had gone heavy into, uh, the, the corded Apple earbuds years and years ago.

• 1:01:26Speaker 01

  There's a few, there's a few benefits, but on the L trains specifically, uh, the corded is so much safer, because if you have an AirPods incident where it's a yard sale and you drop your, you drop your AirPods and everything goes on sale and spreads out all over the place.

• 1:01:43Speaker 00

  Over there.

• 1:01:43Speaker 01

  Yeah. If you drop the AirPod case by accident, everyone knows that it has the springs inside the ejectors that they built.

• 1:01:52Speaker 00

  Yeah, the little rocket.

• 1:01:53Speaker 01

  Gas powered, I think it's called gas thrusters that spray the AirPods all over the train. That could be really dangerous. It could go onto the tracks. It could land directly on the track, cause a derailment. You'd be responsible for potentially billions in liability. So the corded headphone is going to keep you a lot safer and you're going to have a lot less liability on your hands. So yeah, 2,000 people agree.

• 1:02:17Speaker 00

  Adam Singer says wired headphones are superior. Actually, no need to charge them, easier to keep track of. And if they're a quarter inch over here, they're higher quality. Yeah, the big thing is you can buy a ton of the Apple corded ones for the same price as AirPods.

• 1:02:36Speaker 01

  Oh, that's a good point.

• 1:02:37Speaker 00

  And so you can just put them everywhere around your life. You can lose them.

• 1:02:41Speaker 01

  Yeah. Mentally, people probably think like, I gotta, I gotta think of just one, I gotta just buy one pair. But if they just think about it like, well, I'm choosing between $200 for AirPods Pro or something, or you know, 20 pairs of wired headphones. You also, are, are the Apple wired headphones the best wired headphones?

• 1:03:01Speaker 00

  Probably not.

• 1:03:03Speaker 01

  Okay.

• 1:03:04Speaker 00

  But old reliable.

• 1:03:05Speaker 01

  Because the benefit of, I, I would, I would probably stay away from other Bluetooth headphones that aren't on the Apple ecosystem because the integration is so strong. But if you're just plugging it in, it doesn't really matter if it's from another brand. So you could go all over the place. Maybe some nice, uh, nice mahogany headphones.

• 1:03:25Speaker 00

  We, uh, one more thing to cover before our first guest. Marissa Mayer was talking with the New York Times and they have an excerpt here. Late last year, Marissa Mayer was struggling with what felt like an impossible task. Her daughter Marielle was turning 10 and Marielle's teacher wanted three bullet points summarizing what made her special. Mayer, the former chief executive of Yahoo and a pioneer at Google, where she was the first female engineer, likes cracking a hard problem. But she felt overwhelmed by the range and complexity of potential answers. So Mayer asked her personal AI assistant, Dazzle. Within a few seconds, Dazzle delivered an answer. It came back and was like, your daughter is empathetic, creative and an animal lover, Mayer said. She thought it was perfect.

• 1:04:07Speaker 00

  So, um, the, the AI leader pick a story, a personal story that actually resonates challenge is proving to be one of the, one of those.

• 1:04:21Speaker 01

  Yeah, 100%. The, uh, the, uh, the New York Times does this thing where they will just let someone talk and you can tell that the, the writer of that story like thought that that was a weird, I, I, I believe that the writer of that story thought that that was a weird anecdote. So put it in the lead and then now it's going viral. As soon as I read that, I was like, oh, this is like a choice by the New York Times.

• 1:04:42Speaker 00

  Was struggling with what felt like an impossible task.

• 1:04:45Speaker 01

  Yes.

• 1:04:46Speaker 00

  You know, setting it up as like, you know, you're expecting.

• 1:04:48Speaker 01

  But it does, I mean, it is a real thing where like, she's in a very like, you know, CEO mode, dealing with these other things. And then like, sometimes you're like, I don't know what the parents at the school, what like, sometimes like, it's hard to describe a kid in the, at the tier of complexity as everyone else, right?

• 1:05:07Speaker 00

  Braves to steel man, this one.

• 1:05:09Speaker 01

  Okay. Uh, the other thing that I'll steel man is, uh, is the fact that I, I, I think the idea of an AI agent that hangs out in your camera roll and finds opportunities to do things is sort of interesting. Like the AI, the, the camera roll is an interesting source of information and like memory, basically. And especially if you're in the habit of like screenshotting a lot of things, like imagine just anything that you do on your phone, like instead of being like, okay, I have this link, I want to send this into my AI app of choice, save it there and then have it take action on this. It's more just like, oh, I'll just take a screenshot.

• 1:05:47Speaker 00

  David says you can't steal this steel man, it's gone.

• 1:05:49Speaker 01

  I can steel man anything. Um, anyway, let's bring in our semi analysis.

• 1:05:54Speaker 00

  Joey.

• 1:05:55Speaker 01

  Describe yourself in three words without using AI challenge. It was impossible from Marissa Mayer, but how are you guys doing?

• 1:06:04Speaker 00

  How would you describe Doug in, in three bullet points without using AI? You can't use AI.

• 1:06:09Speaker 04

  Not using AI.

• 1:06:11Speaker 01

  And feel free to pass because we know this is extremely difficult challenge.

• 1:06:13Speaker 04

  Doug is a trend setter and trend caller.

• 1:06:16Speaker 01

  Hmm.

• 1:06:17Speaker 04

  Um, I would say that. And he's probably, uh, he's always in the right place at the right time.

• 1:06:22Speaker 01

  Ooh.

• 1:06:23Speaker 04

  As evidenced by, by talking to you today.

• 1:06:24Speaker 01

  That's good.

• 1:06:25Speaker 04

  About the Anthropic IPO.

• 1:06:26Speaker 01

  Yeah, those are, those are, those take me on a journey. I like that.

• 1:06:30Speaker 00

  Two bullets, or I don't know if you wanted to break this up.

• 1:06:32Speaker 04

  He's a, he's a handsome guy. I'll, I'll give him that too.

• 1:06:34Speaker 01

  Handsome.

• 1:06:35Speaker 00

  There you go. There you go. With a nice, with a fresh cut.

• 1:06:38Speaker 01

  Looking good.

• 1:06:39Speaker 00

  Looking good.

• 1:06:40Speaker 01

  Okay. So, uh, how long have you had to digest the Anthropic IPO? I guess the big question is like, a lot of these facts were leaked because of the Investor Day. Uh, what, what, what really stands out to you as like new learnings from the S1?

• 1:06:55Speaker 04

  Yeah, I would say most of the information was, you know, leaked somewhat before. And I think we got some more detail, um, on like per user spend, you know, how big some of the largest customers were.

• 1:07:05Speaker 01

  Yeah.

• 1:07:06Speaker 04

  Um, pretty specific details there.

• 1:07:08Speaker 01

  Yeah.

• 1:07:09Speaker 04

  Um, that were, you know, pretty interesting. So happy to kind of dive into them.

• 1:07:13Speaker 01

  But overall, is this a beat? Because it feels like there was, uh, like a question, we can get into like the overall health of the AI trade, but in general, there's like, uh, oh, like, oh, there might be something wrong. And it seems like everything was on track. Everything kind of sort of mapped to what was leaked earlier. Things look good across the board.

• 1:07:33Speaker 00

  I think it's like a vibe, it's a vibe beat. That's what I, that's what I would say.

• 1:07:36Speaker 01

  Yeah.

• 1:07:37Speaker 00

  I think people were expecting that there was going to be like, one thing that everyone focused like, like the vibes had shifted.

• 1:07:43Speaker 01

  The market.

• 1:07:44Speaker 00

  In such a negative way, like maybe two or three weeks ago where people are like, oh yeah, you really want to do a $2 trillion IPO.

• 1:07:50Speaker 01

  Well, rates are up or whatever.

• 1:07:51Speaker 00

  Like, yeah, when you're not, when you're decelerating, like this makes no sense. And now people look at this and they're like, okay, this, this is like pretty.

• 1:07:58Speaker 01

  Incredible business.

• 1:07:59Speaker 04

  Yeah, this was one of probably the best scaling to a trillion dollar plus valuation in the history of capitalism. To burn 10 billion to get to this valuation is insane. Um, to be this level of profitable this early and at this scale.

• 1:08:12Speaker 01

  Yeah.

• 1:08:13Speaker 04

  Um, it is, you know, amazing.

• 1:08:15Speaker 01

  Okay. So Doug, incredible capital efficiency. Is this a violation of the AI is similar to the railroad build out concept?

• 1:08:26Doug O'Laughlin?

  Yeah, I think so. I think it is actually an incredible narrative violation for that.

• 1:08:31Speaker 01

  Yeah.

• 1:08:32Doug O'Laughlin?

  Um, I think, I mean, it isn't, let's put it this way, it's profitable, right? I think the capital efficiency is mostly because they made a lot of money too.

• 1:08:39Speaker 01

  Sure.

• 1:08:40Doug O'Laughlin?

  That I think is the slight narrative violation. But, um, you know, that's like looking backwards. I definitely think looking forwards, you see several hundred billion dollars compute. And I would say that that's still really in line with the railroad analogy.

• 1:08:53Speaker 01

  Sure.

• 1:08:54Doug O'Laughlin?

  So I think, uh, the railroads are still, they're still railroads, baby.

• 1:08:57Speaker 00

  But, uh, but it's like, it's like you have like a company that is spending on railroad services, but it's, but it's like, this company is not, at, at least right now, doesn't look like the, a manufacturer, like they're not laying the track is like.

• 1:09:10Doug O'Laughlin?

  Yeah, yeah, they're not laying the track. And like, uh, so, so pretty much the OG, I guess the steel companies were, were, if we're going to be using railroad analogies, the steel companies, uh, you know, all got invented essentially to make railroad tracks. So I think that that's kind of the, maybe that's the analogy we're looking for, but I don't know. I think it's really clean. There's no other way to put it. Like this was an extremely clean beat. Like you don't really have a much to nit, nitpick, but I think maybe the concern is like, a lot of people have been talking about like lowered expectations in Q4. Um, they didn't put the September ARR, right? Cause everyone knows that was like a total debacle. Um, but they obviously have an, an open spot for the October AR and, and they wouldn't have it open if they didn't think it was good.

• 1:09:51Speaker 01

  Yeah, that makes sense. Um, what about the compute obligations? Like it's sort of a big number, uh, not really relative to the valuation even, but yes, like 600 something billion is a huge number, but it's over a long period of time. Like it felt very, it felt very important to, uh, restate and reinforce that this doesn't, uh, it's not this year, it's not next year. It's, uh, you know, some of these contracts go into 2033, some of them go into 2036. Uh, what, what is the, uh, what is the extra context around the flexibility to move that, those commute, compute commitments around, up around and, uh, have them hit at the right time?

• 1:10:34Doug O'Laughlin?

  Yeah, I'm going to have one, I'm going to have go, Joey go next because like he's, you know, he's Mr. Number, he's like Mr. Bookie as some analysis.

• 1:10:41Speaker 01

  Yeah.

• 1:10:42Doug O'Laughlin?

  But I think, uh, more than anything else, they told you at the very beginning, uh, we will, we want to under promise and over deliver. Like they said, like, like it's very rare in S1 in like the last paragraph. It's like, no, no, no, we're going to be really conservative. We promise we will under promise and over deliver. Um, so yeah, that's, that's kind of my vibe is that I think the, the hundred billion dollar a year probably works. I mean, it's a little north of a hundred billion dollars, I think, if you like naively annually smoothed it out, but I think it probably works. Joey, what do you think?

• 1:11:10Speaker 04

  Yeah, I think with some of the increased compute costs and then even probably averages closer to seven or eight years, um, it's not as many gigawatts as it looks, you know, from a 600 billion perspective.

• 1:11:22Speaker 01

  Yeah.

• 1:11:23Speaker 04

  And they, you know, they'll be adding a lot more of those compute commitments, I'm sure.

• 1:11:26Speaker 01

  Sure, sure. But, but after IPO, after quarterly earnings, after continued profitability, something that can underwrite it, uh, maybe getting.

• 1:11:34Speaker 00

  Get me up to speed on how the original SpaceX compute deal actually works because if I remember correctly, it was like, they were getting access to a lot of compute, but able to sort of like forgo payments against that for some amount of time, which I imagine was hitting in second half of the year, like the payments really coming online relative to how much compute they were accessing. Is that the right way to think about it?

• 1:12:01Doug O'Laughlin?

  I don't recall that.

• 1:12:02Speaker 04

  Yeah, we saw it a little bit in SpaceX earnings last week.

• 1:12:06Speaker 01

  Yeah.

• 1:12:06Speaker 04

  Uh, start to come through.

• 1:12:07Speaker 01

  Sure.

• 1:12:07Speaker 04

  Um, so it was, I, I think it was pretty normalized for, for both companies. We saw it, you know, in both the compute margin numbers at Anthropic, Q3 versus Q2 and in SpaceX earnings.

• 1:12:18Speaker 01

  Yeah. What, what about, oh, sorry.

• 1:12:20Doug O'Laughlin?

  Sorry. The odd part is just the 90 days, the fact that SpaceX can like essentially whip it back, you know?

• 1:12:26Speaker 01

  Yeah, yeah. Um, what, what about the overall, uh, compute strategy and diversity across the Neo clouds? Obviously Anthropic has, uh, really big contracts with Broadcom and the hyperscalers. But, uh, for some of those Neo clouds, there was a lot of rumbling with Elon about the, you know, Neo clouds over the weekend. Like, how does this S1 change the calculus around just Neo cloud economics? What a good Neo cloud looks like. You obviously have, uh, cluster max to understand the capabilities of the cloud. But it feels like the financing side is also, uh, you know, increasingly as important as the capabilities.

• 1:13:07Speaker 04

  Yeah. I mean, if you have a good backer and a good backstop, so whether that be a hyperscaler for your Neo cloud or an AI lab. Um, I think Anthropic has shown today, AI inference is a really great business. Um, even running the training is extremely profitable. Um, so compute margins are excellent. If them, a chip company or a hyperscaler are backstopping a Neo cloud and they have operational, um, excellence in how they operate those new clusters. It's, you know, Neo clouds are still very good businesses that have increasing monetization tailwinds behind them.

• 1:13:38Speaker 01

  Are you starting to see this, uh, the, like, the long rumored Neo lab to Neo cloud pivot happen yet? Either of you, I guess, Doug?

• 1:13:49Doug O'Laughlin?

  On, on the, on the margin, I think so. Uh, I think people are, people really want to be selling tokens. Uh, if you look on the, the fact that DeepSeek V4.1 is probably the reason why all these cybersecurity hacks are happening. It's pretty clear that if you have the gigawatt and you can, you can place it with an open source model. Um, you know, I, I wouldn't call that like a Neo cloud, but like tokens as a service, I think is going to increasingly happen. Especially as the volume of open source models, uh, continues. Also, I wanted to address one thing. I feel like Elon is, uh, shitting on the Neo clouds because he's also a Neo cloud. He's just like, I'm, I'm the biggest Neo cloud. Those guys just aren't as good as me. That's really what it comes down to.

• 1:14:27Speaker 01

  No, I mean, that, that, he said that exactly. He was like, yeah, if you don't have other revenue.

• 1:14:31Speaker 00

  Yeah. The, the, the, the points that he's making are self-serving, but it also makes sense that a lender would rather lend to somebody that like has revenue that's diversified. So it's, it was like, you, he can be like trying to further his own interests while at the same time like being, like having like a pretty coherent, like thought that stands out. Yeah.

• 1:14:54Speaker 01

  Yeah. Um, what, what, uh, what else can you tell us about the nature of all the cybersecurity threats? Like, uh, how much is it actual, uh, inference of DeepSeek, as you mentioned, versus just, uh, piggybacking on cyber, uh, you know, um, different exploits that have been discovered by other folks? Like, how, how relevant is compute in the cybersecurity battles that we're seeing playing out right now? Do you have a, do you have a grapple on that yet, Doug?

• 1:15:22Doug O'Laughlin?

  I mean, I think, I think compute's pretty relevant here because it seems like a bozo model can probably hack your, your, uh, your infrastructure. It doesn't really take much. I think the bar is surprisingly low. And I think it's in all of our interests, like not to talk about how low the bar is, if I'm going to be honest with you.

• 1:15:38Speaker 01

  Okay.

• 1:15:39Doug O'Laughlin?

  Um, and I think the, the other, like, it's, it's like, it's like surprisingly low. Um, the other thing I think that's like also really kind of messed up is how asymmetric it is. It's kind of like a drone, you know, like it costs, you know, a hundred bucks to fly the drone into the tanker. And it costs like, you know, like $2 million to shoot the drone out of the sky or something. Um, that, that dynamic I think is very replicated in AI as well. Meaning defense is costs like probably an order of magnitude more than a, than, than an attack. Um, we've done some work on the attack side. The attack side, like the raw tokens that come out of it, it's like, it's like pitiful. It's like a hundred bucks or something. Um, it doesn't take a lot.

• 1:16:16Speaker 01

  Yeah, yeah, that makes sense. Um, how?

• 1:16:18Speaker 00

  Yeah. It's such an interesting problem because we were talking earlier on the show of like, does there need to be like large scale white hat bounties, you know, billions of dollars allocated to like incentivize good people to go after and find these vulnerabilities. But at the same time, imagine introducing the chaos of like, I was using the example of like a Miami course bro being like, anyone can become a millionaire finding like vulnerabilities, right? And they're like, here's how you take the obliterated model, point it at, you know, this legacy, you know, you know, Fortune 500. And all of a sudden, it's just like total anarchy.

• 1:16:56Doug O'Laughlin?

  Also the obliterated models, like they can do the hack, but like you also shouldn't trust them, if that makes sense. Like, you don't really trust their incentives to like actually listen to, like, like the same problem of like steering it a little different also means that it could be doing things that you don't intend, right? You're giving it instructions, but you're like, okay, hack it and I'll fix it. And it's like, well, wouldn't it be nice if there were some API keys, right? And then, then it can like, you know, self proliferate in like the worst case. I think that that's kind of the big concern. Um, and, and like, I'm sure you guys saw when, when the hacks were going around. I mean, I still think there, the hats are still going around, um, that someone had like a DeepSeek swarm, essentially like, you know, patching old PHP. They're like, like hack into old PHP and then patch it themselves.

• 1:17:38Speaker 01

  Yep.

• 1:17:38Doug O'Laughlin?

  So pretty much the only person that can stop a bad swarm is a guy with a good swarm.

• 1:17:43Speaker 01

  Guy with a good swarm.

• 1:17:44Doug O'Laughlin?

  What I'm trying to say. Yeah.

• 1:17:46Speaker 01

  Yeah. Uh, how, how are you, uh, thinking about advising clients on the risk to an open source ban for Neo clouds? Like, because some of the Neo clouds seem very dependent on open source models. But at the same time, like the Anthropic S1 shows that the Neo clouds can sell compute to closed source labs very profitably as well. So, uh, how are you balancing those risk factors right now? It feels like we went through the whole Jensen letter. There was a lot of energy towards don't ban open source. But would now we're also post PHP.

• 1:18:21Doug O'Laughlin?

  Yeah, I don't know. I think, um, I mean, I, I, did you guys see the thing on Veet today about, uh, like essentially like trying to go back to an Alibaba cloud? Like, I think those, those are the kinds of anecdotes that creates, um, let's just say like a blanket wide industry ban. I really don't, like it's, we're in a really interesting time where it's like really reflexive. Like clearly something has happened. Um, something has changed. And, uh, is it that we are just like, okay, swarm up, use more compute? Or does the government say like, fix this, right? Like, and the government saying fix this is probably saying these class of models are now illegal, right? And so that is a potential.

• 1:19:00Doug O'Laughlin?

  I don't really know. When you're in like these really special reflexive moments, it takes very little headline to move the entirety of a vibe one way or the other. So we'll see what it kind of shakes out as.

• 1:19:10Speaker 01

  Yeah, yeah. I mean, uh, even if there's no ban, there, there, there, we were talking about like the, the carrot is like bug bounty, maybe from the government, but the, the stick is like something that's not quite nationalization. But it's the government saying like, I am ordering you Anthropic, OpenAI, Google to devote millions of dollars of compute to setting up swarms to go patch the internet and fix everything. And you have to do it. Like you have been conscripted, conscripted for this and maybe it's a tax write-off, but like you gotta do it.

• 1:19:42Speaker 00

  What kind of timeline are we looking at to have a American open weight model that people are like gen- genuinely excited about? Like getting the, like this changes everything now. Uh, I, I want to see, you know, uh, the, the pelican on a bike and, and have people, you know, losing their mind over the pelican, right? When are we, how far away are we? Like clearly Jensen cares about this.

• 1:20:04Speaker 01

  Yeah.

• 1:20:04Speaker 00

  There's rumors about, um,

• 1:20:06Speaker 01

  Reflection.

• 1:20:06Speaker 00

  Reflection and, and, and them, you know, going in-house. But like, is this like a three month thing or like a, you know, November surprise kind of thing?

• 1:20:18Doug O'Laughlin?

  I, I mean, I think the reflection is our best shot so far. And even though it wasn't quite at the frontier, it was like, uh, the frontier of open models, sorry. Uh, the frontier of open models, it did look relatively parameter efficient. Um, but if I'm going to be honest with you, it wasn't exactly like, wow, we're so bad. Cause like, you know, one month later, Kenny 3.1 looks like it's a frontier model, right? Um, so, yeah.

• 1:20:43Speaker 01

  Um, what, what else is changing? Uh, maybe Joey, you can fill us in on some of the trends on tokenomics. I mean, from the S1, it looks like the cost per task is getting, uh, is coming down. There's a lot of progress there. But what other, uh, what other trends are you seeing in tokenomics broadly, uh, on the demand side? What, what, what Fortune 500 companies are asking for, what businesses are asking for? Like what, how is the tokenomics model changing in semi-analysis?

• 1:21:09Speaker 04

  Yeah. I think the biggest misconception is actually on the Fortune 500. Um, a lot of the, the business today is really driven by 99th percentile customers. If you look like the ramp data, for example, you know, the top 1% of customers are spending a hundred K per employee. You know, they came out today, the S1 said the top 5% of Claude enterprise users are spending 18,000 a year per employee. But you know, the average non-Cloud code user is spending less than $75 per month.

• 1:21:34Speaker 01

  Sure.

• 1:21:34Speaker 04

  Um, and it was 0.1% of Fortune 500 budgets were, were Claude budgets, according to their S1. If you look at who's spending heavily on the frontier, um, you have a few large companies like Meta, Microsoft, um, you know, other companies like Cursor, um, that utilize the API. But the Fortune 500 Global 2000 is very lagging in their per employee AI spend.

• 1:21:57Speaker 01

  Yeah.

• 1:21:58Speaker 04

  And I think, you know, the story in the 27 is, you know, We can see a decent step function change with a lot more budget allocated next year towards us and kind of taking from more of the traditional IT and software budgets towards us. And we can see a lot of growth outside of, you know, both coding and on power users.

• 1:22:15Speaker 01

  In the Fortune 500, the one number that surprised me to the downside was that they reported over 50%, but not 90% of the Fortune 500. And I was like, oh, this is kind of bad. But then I realized like, it's kind of bad for the Fortune 500 that they're not spending more, right? Like, what is going on in the Fortune 500? What is the, what is the shape of a business that's not using AI in the second half of 2026 look like? Is this like a lumberyard or something? Is it all humans walking around?

• 1:22:44Speaker 04

  No, you'd be surprised. Even at, even at a lot of Fortune 500 tech companies, the Claude code budget might be 1,500 per month in tokens.

• 1:22:52Speaker 01

  Okay.

• 1:22:53Speaker 04

  You know, that little and a lot of people don't even hit that in the engineering organization.

• 1:22:57Speaker 01

  Sure, sure.

• 1:22:59Speaker 04

  I think, you know, areas outside like traditional white collar, you know, middle, front, back office work, you know, still is lagging a ton as well.

• 1:23:06Speaker 01

  Yeah.

• 1:23:06Speaker 00

  I don't know. I don't know if Doug could survive on 1,500 budget daily.

• 1:23:13Doug O'Laughlin?

  Sounds like a nightmare. I could do $1,500, $1,500 a day. I think.

• 1:23:18Speaker 00

  It would drive you slowly insane, though.

• 1:23:20Doug O'Laughlin?

  I would go insane slowly. No, I think the thing that I hear that like, my favorite is actually like talking to normal people. I went to a wedding this weekend in, in good old Fort Worth, Texas. Okay. With normal people. And they're like, dude, it's crazy. No one's like, okay, one person who's like, really AI pilled, of course, talks to me about AI. It's like, it's crazy. No one's talking about this. And they're like, yeah, what do you use for work? And they're like, oh, yeah, Microsoft Copilot. Like, you gotta, the world is a big place, guys. And I think we are still pretty early into, let's say, the hyper adopter curve. Like, is, you know, like, the hyper adopters, like, I would argue semi analysis isn't like that. Probably in the one percentile outcome of like, people who are spending on Claude, right? And we are like, okay, we can't go backwards. We definitely, maybe we're not like, like going crazy with Opus fast mode or something like that. But

• 1:24:05Doug O'Laughlin?

  I think that we will continue to be spending quite a bit. And I think what we're doing is a little different than the average company. We're a little younger, right? There are companies out there where the average employee age is north of 50. Do you think they really care to figure out Claude code? No, dude, they're just trying to collect some checks before retirement. Like, I think that, you know, it just, it takes a long time for people to adopt new technology. This is something that's like, the, the technology industry has been very aware of for a long time. There's like an S curve of adoption. And I think we are still pretty early. So it's gonna take some time. My favorite, my favorite theme is always to pay attention during the holidays, actually. You watch that there's step functions of adoption happen during the holidays because young people go home

• 1:24:47Doug O'Laughlin?

  to their, you know, to the boomer family and explain, hey, have you seen Muse try to buy you an advent calendar or something at Thanksgiving? And they're like, wow, this is magic. So there's, there's some really cool things there that that's often why we see step function. A good example is Claude Code's true inflection moment, even though they had like 4.5 out for a while, was during Christmas. Because everyone has time off and they're like, hey, I want to vibe code a project. And then everyone's like, they came back to work and they're like, guys, you gotta try this.

• 1:25:13Speaker 01

  Yeah. So that's very interesting because in the S1, they, they noted that revenue or, or token spend might go down by 20% in that two week Christmas holiday. But that's actually where your hardcore users are taking a break, but your new users are actually coming in and you're sort of creating a new cohort that will scale and be the, be the influencer of the next year, which is, which is very interesting dynamic.

• 1:25:39Doug O'Laughlin?

  Totally possible, right? They're taking a break. But I think, I, I just think it's really interesting that these, these adoption curves are like, they're not continuous. And the most important thing is looking for the next adoption curve to see what the next, what the next S curve is, right? Like, it's very clear Claude code is gonna be, or like, rather, let's say, the, the white collar work replacement is the biggest one possible. And it's, you know, relatively early on, but it's still really large wave. What's the next one gonna be? It's gonna be really hard. We, biology could be, but I don't think it's gonna be anytime soon. Cybersecurity probably is now moving up the list. Agentic consumer stuff, definitely, I think, moving up the list with Muse. I've been.

• 1:26:21Speaker 00

  How are you? What's the, yeah, what's the overall semi-analysis company view on the, like, market side, like, personal agent market sizing? Because people have tried to do, like, I think some people, like, reading into the, the meta earnings call, I think people, I was at least a little bit surprised to see that, like, shopping was the number one use case, just simply because shopping on the sites that Muse supports, like Shopify stores is already so efficient. Like, you just can't, it'd be hard to actually, like, if you tell your Muse agent to buy something at the same time that you have the product that you want to buy, like, it's not like there's real, like, time savings

• 1:27:04Speaker 00

  and people, and this idea that, like, people like shopping. So how are you guys thinking of that market? When I see that Meta is doing a 95 billion dollar equity financing, I'm assuming that they're still gunning for this sort of AGI SI vision, and it's not really in service of the personal agent product that, that they have traction with right now. But what's the broader outlook?

• 1:27:30Doug O'Laughlin?

  Joe, you want to take this or?

• 1:27:32Speaker 04

  Yeah, I mean, I think, look, they're, they're still very determined in enterprise AI and achieving AGI there. And, you know, I think in 27, we'll see them monetize your enterprise pretty quickly. You know, they always in consumer, when it's a new ad format, you know, we see delayed monetization. Obviously, e-comm, just any of like their ad markets today, they serve e-comm, travel, you know, will be big areas in consumer as well. And we know consumers don't like to pay for stuff. So taking a cut or advertising some performance based outcome is the way it'll be monetized. And I think we can get to the point where they talked about on the call, where your consumer agents are talking to business agents,

• 1:28:11Speaker 04

  and they facilitate that transaction and, you know, take 5%.

• 1:28:15Speaker 01

  Yeah. Yeah. It seems like Zuck or Meta internally is still just like a little bit cagey about being like, we're gonna make a ton of money off of ads. And so they'll, they'll frame it differently in the press releases, but the, the big dollars come from advertising when you're in the meta organization. Doug, you mentioned your Muse maxi, your Claude code maxi. Do you have any linkages between your two agents? I'm interested in your philosophy around, like, will you have a super agent that talks to everything at some point? Or are you happy to have two tools, two tabs, two apps?

• 1:28:49Doug O'Laughlin?

  I, I think I'm happy to have two, two tools and two tabs, mostly because I think of Muse as intensely my personal. Like, I don't let it touch my work email. And then I think of Claude code is, like, has a lot of the inner linkages with my work email. I don't think it's as proactive. Like, I am kind of excited. Like, let's just say, like, for better or for worse, Anthropic really has positioned itself as the enterprise solution. And if they have a really solid enterprise Claude, you know, Claude Muse or whatever, I'm probably gonna use it just because I do kind of trust their safety to a certain extent. Like, they sure be talking about it in their S1. So I think, I think, I think that that's how I've mentally bucketed it versus like, you know, Meta has all my

• 1:29:33Doug O'Laughlin?

  data anyway. So and like, you know, I'm scrolling on, I'm doom scrolling on Meta, on reels. What's, what's the difference between doom scrolling on Muse versus reels?

• 1:29:42Speaker 01

  Yeah, it's kind of a good fit. It's like, it's like you have an executive assistant at work, and then your Claude instance will be like a executive assistant extension, and then you have a personal assistant at home if you're doing very, very well, and your Muse agent can be your sort of personal assistant. And these two might interface, but they're separate roles.

• 1:30:02Doug O'Laughlin?

  They might interface. Like, and actually, usually what it is, it's forwarding emails one to the other.

• 1:30:06Speaker 01

  Sure. Yeah. But, but actually, having that clean line is great. And that's the way a lot of extremely high net worth individuals set up their lives. They have someone who works at the, the company, and then they have someone that works at their family office, and they too can communicate, but they are in separate organizations. So maybe that's a.

• 1:30:21Doug O'Laughlin?

  I think it's good from a separation perspective. But like, I, I don't know. I also just think that, like, you know, the personal assistant stuff, it is really sick to just like, I know, the, the lamest one, everyone's like, oh, just reads your emails. But like, dude, I got a lot of emails, man. And I think a lot of people, it's like, okay, for example, you have a kid's birthday party and you're like, organizing the kid's birthday party and like, you're just like, oh, your, your wife sends you some reel. And they're like, hey, I wanna go to this.

• 1:30:46Speaker 00

  Your school, your kid's teacher says, describe your child in three full words.

• 1:30:50Speaker 01

  That was a bad example.

• 1:30:51Speaker 00

  Completely stumped, you know?

• 1:30:53Speaker 01

  That was a bad example. But, no, I mean, I, I had an interaction where I had emailed someone to, to do something and they had to actually update something on a website. And they said that they were gonna do it. And then Dot, in this case, like, went and checked like a week later and said like, oh, they haven't done it. Like, you should follow up with them and just make sure that they actually finish that task. And, and those types of little like follow ups are more useful. And it's, it's more than like, it would require me not just reading my email inbound, it would mean reading my outbox and checking up on the things that I asked to do, which is like, complete laboriousness that no one wants to do. Now, what is the willingness to pay of everyone for that? How do you monetize that? You know, travel, you know, Americans don't spend that much

• 1:31:38Speaker 01

  on discretionary travel all the time. So you take a small cut of that. These aren't necessarily quadrillion dollar markets immediately, but they are very interesting and can grow a lot.

• 1:31:49Speaker 00

  Final question I forgot to ask you earlier, Joey. Did you want to see more on their actual tokenomics, like more granular data on margin profile across different models and things like that? Because I think some people were hoping to see more. But in some ways, I look at it, I'm like, how much money did you spend? How much money did you make? Everything else is kind of like your business. But what's, what's your view? It feels like full employment for semi-analysis.

• 1:32:16Speaker 04

  Yeah, there was a, there was a number in there that 80% of the revenue in the last three months were from models in the last, you released in the last six months. So I think that was the extent of detail we got. I don't think they're going to get much more detailed than that. You know, I know they've given some inference gross margin numbers before. If you have that and the training, I think that gets you most of the way there in terms of understanding this business. And then too, they gave a lot of good detail, I think, on the cloud service partner, your relationships and the revenue running through that and kind of the economics they're paying from a revenue share perspective. So I think you have enough detail here to really understand the business and, you know, how it's going to move going forward.

• 1:32:57Speaker 01

  Yeah. Have you, have you been tracking how the big labs are using pricing pressure against open source models? Is there, is there a world where the labs get to such scale and such efficiency that they can distill their own models to a point where they can be price competitive in some way, just to sort of stave off the open source ecosystem a little bit?

• 1:33:22Speaker 00

  Exactly what's

• 1:33:23Speaker 01

  It feels like that's what's happening, but what's your take from like a economics perspective?

• 1:33:27Doug O'Laughlin?

  I think the person who's doing it the most is OpenAI.

• 1:33:30Speaker 01

  Yeah.

• 1:33:30Doug O'Laughlin?

  Right. Like the, the 6.1 Terra today feels very much like a stab at. I don't think Opus or rather Claude cares about Sonnet that much.

• 1:33:39Speaker 01

  Yeah.

• 1:33:40Doug O'Laughlin?

  But I do think that OpenAI has kind of their eyes out, focused on that.

• 1:33:45Speaker 01

  Yep.

• 1:33:46Doug O'Laughlin?

  I think the problem with that though, is at the end of the day, you know, the, the inference margins on open source is pretty good. So if you really want to get all the way there, you probably have to do another 50% price cut or something like that. Where I, because if you own it today, if you own the stack, if you have like a relatively modern V200, you're still talking about north of 50%, if not like 70% gross margins. So, you know, you're able to pay like these like 10, like these like 20 to 30 dollars per watt or like 30 million per megawatt price using just open source models if you get the customers on it. So I think that like, it's possible, it's entirely possible, but I think the margins of the labs in aggregate would have to be a lot lower, but you'd argue that they wouldn't

• 1:34:31Doug O'Laughlin?

  have as much competition.

• 1:34:32Speaker 01

  Sure, sure, sure. And then, and then the, the wrinkle is that then you also have to have some sort of like fine tuning capability because that's something you get with, with open source that you won't with a closed source model necessarily. Interesting. Well, thank you guys for coming on. So, and, and, and walking through all of this. We do have to jump to the next interview, but Jordi, you have any?

• 1:34:52Speaker 00

  Yeah, I really enjoyed it. Thanks guys. Great to see you.

• 1:34:54Speaker 01

  Great to see you.

• 1:34:55Speaker 00

  Congrats on the haircut.

• 1:34:56Speaker 01

  And yeah, congrats on the haircut.

• 1:34:58Doug O'Laughlin?

  Thank you. Appreciate it.

• 1:35:00Speaker 00

  See you guys.

• 1:35:01Speaker 01

  Have a great one. Let me tell you about public.com. Trying to invest in a drop it? Do it on public. Invest across all asset classes, automate anything with AI agents, run trading strategies, risk management, money movements, and monitor markets with AI agents for investing on public. Our next guest is the chief executive officer of Turo. We have Andre Haddad. And Turo is launching verified FSD today for Teslas. Andre, welcome to the show. How are you doing?

• 1:35:29Speaker 06

  Great. Well, thank you.

• 1:35:30Speaker 01

  Thanks so much for hopping on. First time on the show. Would love a little bit of backstory on you, key initiatives for the company, and then the news today.

• 1:35:38Speaker 06

  Great. Happy to jump in.

• 1:35:42Speaker 01

  Thank you. Can you kick us off with a little bit of your, your corporate journey? And then some of what you're working on at Turo?

• 1:35:50Speaker 06

  Sure. Actually just celebrated last month, my 15th anniversary with Turo.

• 1:35:55Speaker 01

  Amazing.

• 1:35:56Speaker 06

  So my team and I have been reinventing car rental for the last 15 years. As you probably know, I hope you've used Turo already for some of your travel and trying out new cars.

• 1:36:08Speaker 01

  Yes.

• 1:36:09Speaker 06

  We've been bringing, you know, this notion of peer-to-peer car sharing to the market now for over 15 years. We're delighted to continue to grow the business. This year we'll do around 1.2 billion in revenue out of 3.3 billion in DVD.

• 1:36:25Speaker 01

  Yeah.

• 1:36:26Speaker 06

  We are excited to announce some exciting technological enhancements to our platform today. We'll talk more about that if you'd like in the show. We're bringing Tesla FSD integration into Turo. So now you can actually experience AVs by renting a car on Turo, specifically the Model 3 and Model Y. So very excited about that.

• 1:36:47Speaker 01

  So I rented a Tesla with FSD on Turo like a couple of weeks ago. It was a fantastic experience. I wound up just like searching or filtering. This was before you had the verified integration. And I was able to verify that it was hardware four, which is like a good FSD and this model year and whatever. But it was, it was, I guess, like maybe a little cumbersome. And so you've made it easier, but I'm interested in like, how are you vending this for the customer? How are you surfacing this? How are you verifying it? Like, why is it important to have an actual product strategy around this initiative?

• 1:37:24Speaker 06

  Well, great that you've experienced it and you sort of hacked your way to it. We saw a lot of people doing exactly what you were doing. And that's what got us thinking about just making it easier to your point and making it verifiable. So, you know, Tesla is a very popular make and, you know, on Turo, the Model 3 and Model Y are respectively the fourth and third most popular cars that are rented on Turo.

• 1:37:51Speaker 01

  Really quickly, what's one and two? I gotta know.

• 1:37:54Speaker 06

  One and two kind of are, I think the BMW three series and the Toyota Prius.

• 1:38:02Speaker 01

  Oh, interesting.

• 1:38:03Speaker 00

  Interesting.

• 1:38:04Speaker 06

  Yeah.

• 1:38:04Speaker 01

  Cool.

• 1:38:05Speaker 00

  I was hoping you'd say

• 1:38:06Speaker 01

  I was hoping Corvette.

• 1:38:07Speaker 00

  No, no.

• 1:38:08Speaker 01

  Urus would be good. Yes, but.

• 1:38:11Speaker 00

  I had such a, when I, when I was first, when I, when I first realized, like, I was like, wait, Turo, I can just rent any car that I want, like in my backyard for a day. This was like, I would, you know, when I graduated college and I had had this realization, I would just do it Saturday. I'd go and get like a Porsche. And, and back then I was very, I was super limited because I was probably 22 at the time. And I wasn't clearing like certain people on Turo. They're like, you can rent this car, but you gotta be, you know, at least like, you know, like you gotta be at least 30. You guys let people set whatever age they want. But I have amazing, very fond memories of, of doing that.

• 1:38:51Speaker 06

  I, I'm a host on Turo. As you may see from my background screen, I've been hosting some of my cars for the last 15 years and just completed a trip late last night, two actual trips. My Audi R8 was out and my 911.

• 1:39:07Speaker 00

  There we go.

• 1:39:09Speaker 01

  There we go.

• 1:39:10Speaker 00

  Thank you for your service. Thank you for your service.

• 1:39:13Speaker 06

  I have the 2012 Audi R8 GT, which as you may know, the limited edition.

• 1:39:19Speaker 01

  Amazing.

• 1:39:20Speaker 06

  Only, only 300 of those were made. But anyway, circling back to, to why we're excited about the experience of, you know, renting FSD on Tesla FSD on Turo. We'll see on Wednesday, we're launching the actual product on Wednesday. We built actually a bunch of software to, to enable it. First was building direct API integration with Tesla. So we are delighted to partner with them. Their teams have been amazing to work with. We have that direct integration with the API to check the, you know, eligibility of the vehicle as well as the FSD status of the vehicle. I want to make sure that, you know, what we're presenting to our guests, our customers who are going to rent the car, want to make sure that they

• 1:40:04Speaker 06

  actually have FSD enabled. And that, you know, the subscription is being paid for and it's current. And we're going to check that right at the beginning of the trip to make sure that it's effective, you know, for, for the start time of the trip. And also, you know, we have more Teslas than we have Teslas enabled with FSD. So one of the cool things that we've done with our partners at Tesla is we're offering those hosts that have eligible Teslas but don't have yet an active FSD subscription, a one month free trial. This way they'll be able to add it for free for a month. Hopefully they'll see a lot of business from their Turo guests that are going to be booking their cars and become,

• 1:40:46Speaker 06

  you know, long-term and have subscribers to Tesla FSD and, and help drive the success of that great piece of technology.

• 1:40:53Speaker 00

  Do you have a charging integration yet? Like if, if a Turo rents a

• 1:40:57Speaker 06

  Yes, we do.

• 1:40:57Speaker 00

  Because that feels important because if you're the owner of the car and then you're getting billed for like directly.

• 1:41:02Speaker 06

  Yes, it's very, very important. We have, we have that integration on the charging side. It still goes through the host's account, but we automatically take the, all of the EV charging that's happened, you know, during the trip and at the end of the trip. And we automatically bill our guests based on that direct integration. So we, we made that simple, you know, a couple of years ago. And now we're taking on FSD, which I think will make a seamless experience for, for both hosts and guests. So anyway, now that we built this API integration, we can actually surface with confidence in a newly created sort of section of the app that you'll see going live on, on Wednesday. You know, these

• 1:41:47Speaker 06

  vehicles, the Tesla vehicles that, that have FSD. And then when you book the car, you'll be able to be assured that you'll have, you know, an eligible vehicle that have an FSD that is active. And you'll be able to experience the magic of FSD.

• 1:42:04Speaker 01

  This is, I'm thinking, I heard a story about someone who was able to use Smart Summon on their FSD-enabled Tesla to have it pick them up at the airport, which I don't even know if that's supposed to work that way. It feels a little too hard. When I think about Smart Summon, I think like within the parking lot, but I guess it's technically a parking lot, so it worked at this particular airport.

• 1:42:28Speaker 06

  Wow. Not that bad.

• 1:42:29Speaker 01

  Yeah, incredible. And so what I'm thinking is like, there's a future where the range gets further and further, and a Turo host will be able to say, okay, this Turo customer wants to rent the car. They want it delivered to their house, which is, you know, a mile away. Just send the car over there. And then they have access to the car with FSD for their weekend, or whatever they want to do. And it's sort of like a more advanced robotaxi precursor. Very, very interesting for the business.

• 1:42:59Speaker 06

  Yeah, that's why we're very excited about AVs, of course, starting with Tesla, and hopefully as the technology spreads to other car builders and manufacturers, you know, we think that this will be a phenomenal kind of enhancement to our Turo rental experience. To your point, it removes a lot of the friction of picking up and dropping off a vehicle, whether it's at the airport, or the hotel, or the home, or the business location. Imagine if we can actually turn all of these cars into an FSD-like experience in the future, where you can, you know, book your road trip, and then the car comes and picks you up at the airport or your hotel, and then you can keep it for, you know, the few days, or the

• 1:43:42Speaker 06

  week, or whatever. You have the car for you. You don't have to hop in and out of robotaxis because you might be going, you know, to, you know, some, you know, destination that's not a very urban destination.

• 1:43:54Speaker 00

  Yeah. No, that's gonna be, I mean, it's gonna be really, really powerful for the business, especially the two factors. Like, one, the summon element, I think, is gonna be amazing because somebody will be able to list their car. Like, when that gets like to widespread adoption and greater range, somebody can say like, yeah, I want a car, and somebody can list their car on Turo. The car will just drive to the airport at the right time and be available for the person. And then the other thing is like, Tesla is so, it is a piece of technology. It's not like a car that's been technology-enabled. So even things like keys, it's like, automatically give this person, like, effectively a copy of the key in the Turo app.

• 1:44:32Speaker 01

  That's what already happens.

• 1:44:33Speaker 00

  Yeah. But that, when like, compared to like, an owner being like, well, I gotta put the key.

• 1:44:39Speaker 01

  Totally.

• 1:44:40Speaker 00

  Grab it. It's on the, like, I've probably had an experience back in the day, it was like, I'll meet you on the side of the road, give you the key. Or, the key is on the back right tire. And it's like, those things are like wildly inefficient and

• 1:44:51Speaker 06

  Totally inefficient. Yes. And not very safe either, you know, so the keys can get lost, you know, etc. So, we're very excited about that. I think, you know, we recently integrated with some flight tracking and flight information, so you can imagine, just enter your flight information. We'll know exactly when your, you know, your flight is arriving. We'll estimate exactly when the car should be meeting you at the airport. We have now permits. Turo has permits to conduct business in over 200 airports in the US, so we can actually have the car safely go to the permitted location. We'll be able to show you on the Turo app where the car is located, which we already do today, but

• 1:45:34Speaker 06

  it will just happen in real time because of the FSD integration. It just makes for an entirely different rental car experience.

• 1:45:41Speaker 01

  Yeah. Last question for me. What is, what is the bottleneck to growth for you? What is, it's such a beautiful business model as a marketplace, but are you spending a lot on marketing? Like, what is driving growth this year? And, because I can't imagine that, like, higher gas prices affect you all that much. It has probably a little bit of an effect, but no more than any other business related to cars. But what is the biggest, the biggest chunk of cost in the business these days?

• 1:46:13Speaker 06

  Well, you know, what's unique about the Turo business is, we take on a lot of the risks that can happen, you know, during a trip, whether that's, you know, a fender bender, or a more significant, you know, incident, or any unpaid incidental. So, you know, over the last few years, we've been perfecting our ability to manage risk, and to be very good at predicting risk, and ensuring that we avoid kind of the trips that are risky, and then focus on the trips that are the most profitable, the least risky trips. So, that's been a lot of the things that we've been doing behind the scenes, that you don't necessarily see when you're actually using the app, because you, you know, this

• 1:46:54Speaker 06

  is sort of behind the surface. And that's, I think, been the biggest bottleneck. You know, we wanted to get to a place where we can be really good at risk management. And we're feeling great about the progress we've made there. You know, we've been growing, you know, our business, primarily focused on our most profitable trips. As a result, our profitability has dramatically increased in 26, and, you know, we expect it to continue to grow significantly in 27. That, I think, is the biggest bottleneck. And then, secondly, continue to build awareness. You know, the Turo brand is very well-known with, you know, the car enthusiasts like yourselves, and others that are tech enthusiasts. But I think

• 1:47:37Speaker 06

  we still haven't penetrated the, fully penetrated the, sort of the mass market of travel. And so we're very focused on driving a lot of brand awareness and brand building, you know, demand kind of capture, we call it internally, type of activities. We've got a brand-new campaign that's been alive now for a few months. You may have caught it on screens, Rent a Car, Not a Rental Car. You know, so we're building the brand, and I think that's gonna help us drive growth alongside with great technology improvements and perfections of the service over the next few years.

• 1:48:12Speaker 00

  Well, we hope this interview helps drive increased growth and profitability. Because if you needed a, if you were needed a sign to go and rent a Porsche or an R8 this weekend, this is it.

• 1:48:25Speaker 01

  Or maybe a sleigh.

• 1:48:26Speaker 00

  Yeah. Gotta get some sleighs on there, too.

• 1:48:28Speaker 01

  Christmas is coming.

• 1:48:29Speaker 00

  Great to hang in, yeah, great to finally meet.

• 1:48:33Speaker 01

  Have a great rest of your service.

• 1:48:34Speaker 06

  Thank you.

• 1:48:35Speaker 01

  Goodbye. Let me tell you about Console. Console-built AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. Our next guest is returning to the show for the third time. We have the CEO of OnePassword. David, welcome to TVPA. Here he is. Welcome back. Thank you for your service amid, uh, black October. Yeah, hope you've been sleeping. Hope you've been sleeping, but, uh, thank you so much.

• 1:49:04Speaker 00

  No, I've been sending, I've been just texting the OnePassword website to people. This is, if you're scared, like, you know, it's warranted to be scared, but here's one simple thing you can do to start your personal security journey.

• 1:49:19Speaker 01

  Yeah. I saw a tweet that was like, this is the weekend to sit down with all your family, get everyone on OnePassword.

• 1:49:24Speaker 00

  And you line this up perfectly with Austin F1. You guys are Red Bull sponsors, so driving, driving quality awareness to motorsports enthusiasts.

• 1:49:36Speaker 01

  Yes.

• 1:49:37Speaker 07

  Yeah, it's funny. Thank you, first of all, for having me, and also for sharing the word. You know, the post that Catharpy made last year about like, the basics, you know, it's like, I've seen it on my Twitter, you know, my X timeline 100 times now. And there is, really, you got to get back to basics right now, because it's pretty scary out there. So, anyway.

• 1:49:57Speaker 01

  Yeah. So, yeah, what else is going on inside of OnePassword, the organization? What can you tell us? What can you share, both for consumers, obviously, like, get on OnePassword, but for entrepreneurs, for founders, for Fortune 500 CEOs? What are the, what's the nature of the conversation that you're having with other business leaders about security in this particular moment?

• 1:50:20Speaker 07

  Yeah, you know, it's, it's interesting. Almost all of the AI-oriented, you know, productive uses of AI for the last, you know, couple of years have been around the developer use case, coding use case. And so, trillions of dollars of value around helping people code more effectively and faster. And, you know, we've been a very big beneficiary of that as well. Our developer tools have driven our business. We put out a press release around our business momentum, about a half a billion dollars in ARR now. We're across that threshold. And we did it at Accelerate. So, Accelerate, thank you for that. Every time you come on, you bring the bell, which I love.

• 1:51:00Speaker 07

  What was exciting for us is we accelerated through that milestone. And that new AR for Q3 was 90% growth year over year. And we're adding new logos on the B2B side at 150% growth over quarter, over last year, third quarter. And it's very much driven by awareness that, especially in the development cycle, agents need access. And that access, when it's left standing around, is a ticking time bomb. So that's been sort of a big catalyst for our business, but there's other things going on now, too, that are complementing that. We saw this breakthrough a couple of quarters ago around personal agents. And it's the next big way that AI is being used in an agentic way to help people.

• 1:51:45Speaker 07

  We were right in the middle of that and we could talk a little bit about that as well. But that's been something that we launched this week as well. One Password for Agents with our launch partners, which included the likes of SpaceX SI. I'm getting the name right. Metamuse, Anthropic, and OpenAI, and sort of all the folks that are building the agents. And then at the same time, you know, the vulnerability apocalypse has descended upon us. And so there's a heightened awareness generally that agents will find access to things, whether it's maliciously intended or just drift, and they will do unintended things. And so this sort of ability to say it's not going to be me or be a little bit lazy with your credential hygiene, that's not going to work anymore.

• 1:52:30Speaker 07

  And people are starting to realize and they're coming on OnePassword to help. So that's where we are.

• 1:52:34Speaker 01

  How, what have you, what have you learned from OnePassword customers about the gradations of trust between families and organizations and how passwords get shared? Like, it's very interesting that the default model is not just, you know, every password gets shared with every family member. And that feels like intentional from a clutter perspective, but also you might have two separate accounts for, you know, email. And you don't necessarily always want them guessing each other and whatnot. But it feels like some of those learnings might actually apply to the way people work with executive assistant agents in the workplace and then also personal agents in the, in the family setting.

• 1:53:17Speaker 01

  And I'm wondering how you think about setting OnePassword for Agents up for success, where there's a controllability layer that the user can have between, look, I'm okay with my, my personal agent of choice having access to my calendar, but maybe not my email, or maybe just my personal email, but not my work email, or maybe just my work email and not my personal email. So how, what patterns have you learned and how do you see this playing out? Or how do you deal with this? Are you one of the people that just gives AI access to everything or nothing? Or how have you grappled with it?

• 1:53:53Speaker 07

  Yeah, no, OnePassword for Agents is really built around the premise of just give the agents what they need. Just, just when they need it, just enough with full visibility. It's the same primitives that apply to what we do for businesses, even very large enterprises. But again, these are non-deterministic actors, right? They, you don't know exactly what they're going to do, even when they're well intended. And there are so many ways for something malicious to be done as well. And so you have to reduce that standing access. One way, one of the ways you reduce this is you don't give it out. And so what we've done, interestingly, all of the partners, the, it started with, with Instinct and then quickly saw with Metamuse and with Dravik Bot and others, these partners we've been collaborating with for months now, they see it as well.

• 1:54:37Speaker 07

  People want to have that control and they'll only really lean in and use these tools if they know that there's this, a trust layer. And so we set this system up, we've been working for a long time on the primitives, that allow that secret to never get into the model context. It's never in a file where another agent can come and find it and do something else with it. It's just delivered for the agent for the task that it wants. And so you can scope that a lot of different ways. And we're working with each of those partners on the roadmaps as to what does that scoping and that control look like for the individual to exactly your use cases or the examples that you share? How long do I want to let the agent access this particular resource on my behalf and under what conditions? And so that's really the, the basis of what we're bringing here.

• 1:55:20Speaker 07

  It's this trust layer that sits between, you know, the agents that do work for you and, and the people that are ultimately accountable for it. And we're just all in on it and it's been, it's been great for our business, but ultimately it's aligned with our mission.

• 1:55:33Speaker 01

  Jordi?

• 1:55:33Speaker 00

  Do you, do you think at all about, like, proact, like more proactive features and products for, on the consumer side that you could do around personal security? Like, it feels like-

• 1:55:45Speaker 01

  Watchtowers, kind of like that.

• 1:55:47Speaker 00

  Yeah, but it, but it feels like now's the time where there's like a lot of activation energy from consumers of like, I want to do something, but at the same time, consumers are lazy. And so it feels like you guys have gained so much trust.

• 1:55:58Speaker 01

  Like, Agentic Watchtower would be like, yeah, I did use the same password 25 times when I was in college, setting up random websites, like, go change those passwords for me. Like, that's a possibility, but it's kind of crazy, but I don't know, is there-

• 1:56:12Speaker 07

  Yeah, you're, obviously, we're right down the path on that. A lot of that is what we'll do in conjunction with our partners, too, right? And so, and then, and what you want to make sure from our perspective, that we work really hard to do, is all of that activity happens and preserves the zero knowledge basis of our vault. So we do a lot with confidential computing to make sure that when those secrets get decrypted, if they need to be decrypted for a reset or otherwise, it's never in a place where a bad actor or an agent can find it and do something with it. So very carefully constructing that, but yeah, absolutely, you'll see that come. The other thing that's interesting to, Jordi, to your question is, what happens with preferences? You know, people's preferences, they're actually looking to have more control over that as well.

• 1:56:53Speaker 07

  And, you know, context is a, is a broad term, but there's some things that look somewhere between a secret and, and a context that people want portability and control on. And so I think you'll see more and more of what we do in connection with our partners as well, as providing some of that control layer at the OnePassword level as well. So it's really, it's really an exciting opportunity. But like, you know, first things first, making sure that those secrets stay safe. Because what we're seeing in all the crazy security incidents that are coming up now is vulnerabilities are not stopping. It's coming, it's coming so easy and, and cheap and fast to utilize AI from that actor's perspective, to get into systems, to get onto somebody's laptop.

• 1:57:35Speaker 07

  And the first thing that they do when they're there is they look around and say, well, what, what access is standing around here? What can I get access to immediately by just, you know, getting credentials? So you can go all the way back to where, you know, the hugging face incident started to make its way into the public discourse, to everything that happened this week. And it's, you know, case after case, credentials being found and compromised. And that's our mission. Let's get that, get that cleaned up from starter. And then let's make sure that we don't propagate that problem as we use AI internally and make the problem worse over time.

• 1:58:07Speaker 01

  What is the OnePassword house recommendation for personal use of hardware keys?

• 1:58:15Speaker 07

  Yeah, so-

• 1:58:16Speaker 01

  Is that something I should be onboarding my family members to after I've got them on OnePassword?

• 1:58:21Speaker 07

  Hopefully use multi-factor authentication.

• 1:58:23Speaker 01

  Of course.

• 1:58:25Speaker 07

  You know, it's interesting, you could ask your agents to, to stack rank the different forms of multi-factor authentication in terms of security. The agents will tell you that, you know, a hardware key is going to be the most secure. It's also got some challenges associated with it. I use pass keys for second factor as much as I can, which I think is a really good way, if you don't want to go to the hardware way. SMS is probably the easiest to be, to be compromised. SIM swapping is a real thing and it's pretty easy to implement. So the bad guys want to get in, you think you're covered through MFA. So, you know, at a minimum, I would look at TOTP or pass keys on one level. But, you know, if you're really concerned, a hardware key is helpful. And then really just, you know, the other thing that people are doing is they're, they're just sort of like getting lazy about it.

• 1:59:10Speaker 07

  Well, you know, I'll store it in my browser or, you know, a variety of different ways. But like, having that end-to-end encryption and having that just-in-time delivery, where the secret is never exposed or can't be read from a file, is going to be really important because the cost to attack is going to zero. That means everybody's a target. Right before, it was just the guys that were, were big enough to be worthwhile going after. But when, when the cost is, when it's automated and the cost is almost zero, really, any standing access is a potential problem. And so you really want to make sure that that central control point, you know, is, is the way that your agents or you get access to the things that are most important to you.

• 1:59:49Speaker 01

  Yeah, it used to be the most secure way was just writing on a sticky note, putting it on the desk. But now the smart TVs have cameras, they're spying on you. And so they'll see from the camera, the picture of this, of the sticky note with your password on it. And so there's no, there's no other solution. You have to use OnePassword. It's going to be bad for the sticky note industry. I, it will, it has the knock-on effect. If you're, if you're really as successful as I think you're going to be, there's some money to be made shorting sticky notes. I love the industry, but it's, it's, it's bad news, bad news. What you're building is good. But thanks so much for coming on the show.

• 2:00:25Speaker 00

  Awesome.

• 2:00:25Speaker 01

  Have a great rest of your day.

• 2:00:26Speaker 00

  Thanks, David.

• 2:00:27Speaker 01

  We'll talk to you soon.

• 2:00:27Speaker 00

  Thanks for the update.

• 2:00:28Speaker 01

  Goodbye. Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. We're very excited to partner with the New York Stock Exchange. Our next guest is from the co-founder and CEO of Nevis, Mark Swan, is with us with another gong-worthy moment. Welcome to the show, Mark. How you doing?

• 2:00:50Doug O'Laughlin?

  Very, very good. Honored to be here, so thanks for having me.

• 2:00:53Speaker 01

  Honored to have you. Thank you so much for coming on the show. Please introduce yourself and the company.

• 2:00:59Doug O'Laughlin?

  Yeah, I'm Mark Swan. I'm the CEO and one of the co-founders of Nevis. And we're building an AI platform for wealth management.

• 2:01:06Speaker 01

  Very, very cool. Is there a particular scale of wealth manager that you're going after? Are you targeting the, like, the big banks, like UBS, JP Morgan to start? Or smaller firms, everything in between? However, have you thought about slicing up the market?

• 2:01:21Doug O'Laughlin?

  So our customers are primarily registered independent advisory firms, which make up kind of the other big part of, of the wealth management industry. And today, you know, our customers are some of the fastest-growing, largest firms in the industry. They're about 700 billion of assets under management across all of them. And then that's, you know, we add about 50 billion of assets across those customers every single month. So it's, it's a big portion, but it's more on the RIA side rather than banks.

• 2:01:46Speaker 01

  And what is the killer feature that you're advertising to your customers?

• 2:01:52Speaker 07

  Yeah, so we give financial advisors access to agents that help them automate pretty complex workflows end to end. So building investment strategies, tax strategies, all the way through to the operational things like moving money and open custody accounts. And so a financial advisor on Nevis can actually automate a lot of that work so they can grow their book of business and deliver more service that previously maybe wasn't possible and without AI.

• 2:02:17Speaker 01

  And are you still working on like the intermediate artifacts of that? Like there might be an Excel sheet or a spread or like a PowerPoint deck that gets sent to a client to say, hey, this is the strategy we're recommending. Here's a couple of scenarios. We modeled it out all for you based on what we know about your financial position today.

• 2:02:34Speaker 07

  Yeah, we build actually a lot of that like in-house. So like those tools, like we'll build our own financial planning tools, we'll build our own, you know, meeting prep materials. Like more of that is actually getting like built vertically within Nevis. But we also are, you know, obviously interacting with things like Excel and other systems as well.

• 2:02:52Speaker 00

  Yeah, you said a meeting notes. Are you guys doing like a meeting? Do you have like meeting recordings? Like, is that something that you brought in-house?

• 2:02:59Speaker 07

  Exactly.

• 2:03:00Speaker 00

  It feels like.

• 2:03:01Speaker 01

  Yeah, you have an in-house solution and not like, because a lot of them might, a lot of wealth managers, I imagine, have like an ensemble of tools cobbled together, but you want to be more of like a compound startup?

• 2:03:12Speaker 07

  Exactly. Like our view of Nevis is like, build the platform, not a point solution. So, you know, we have our own note taker, you know, we have our own document parser. Really the benefit is like a financial advisor can use Nevis's note taker, leave the call, Nevis has already, you know, pre-generated all the follow-ups. So, you know, client might say, I need to withdraw $50,000 in cash, and we'll pick up on that, create the order form, and then one click, you know, complete that execution end to end. And so you can have this like fully connected system as opposed to just like a single point solution.

• 2:03:41Speaker 01

  How capable is the system currently of sort of like idea generation? I feel like a lot of the value that wealth managers bring is not just being a, you know, a point of contact for execution on all these things and operations, but actually knowing that this is the right product, that I have access to this hedge fund at this time or this venture fund, and I want to get to this in front of you because I know it's what you're interested in. That type of, you know, insider information that might not be on the open web yet is a lot of the value of these wealth management firms. Are you working on that type of idea generation as well?

• 2:04:20Speaker 07

  Yeah, so we do a lot of like proactive next best action for advisors. So I think, you know, we have agents running in the background constantly coming up with opportunities for financial advisors. So you should be waking up in the morning, going on to Nevis, and Nevis has proactively suggested, you know, maybe someone's, you know, moved job, right? And as a result, you know, you need to, you know, change their financial plan, and we'll proactively do all of that work from drafting the email to the financial plan, you know, give that to the advisor, they reach out to the client, as well as, you know, investment strategies and so on and so forth. So moving to being much more proactive, you know, agent-driven. So financial advisors are kind of given these like superpowers as a result.

• 2:04:59Speaker 00

  When did you actually start the company?

• 2:05:02Speaker 07

  We started the company two years ago. So we were, myself and my two co-founders were Revolut before this, founded the business two years ago, got backed by Sequoia Capital in our seed round. And then, you know, past two years, yeah, we've scaled, you know, this year alone, we've gone from like $50 billion of assets to over $700 billion of assets now across our customers. And it's been, you know, an amazing journey so far.

• 2:05:25Speaker 00

  Massive progress.

• 2:05:28Speaker 01

  Raised money too.

• 2:05:29Speaker 00

  Oh, I know, I know.

• 2:05:30Speaker 07

  Yeah.

• 2:05:32Speaker 00

  Seventy.

• 2:05:33Speaker 07

  And today, we raised, yeah, 70 million.

• 2:05:37Speaker 01

  With wealth management, there's like seven different levels.

• 2:05:39Speaker 00

  It's interesting because like the, the concept of like AI for wealth management, like I feel like you guys are maybe probably the 100th company to come at this with that pitch. But you guys had this massive advantage in that you started like two years ago and that you could be like, we're going to be able to build all these features. We can do it and really think about this platform at one time versus like companies that have taken cracks at this category historically were like, are now like, okay, like what's our, they were AI for wealth management, but it was really like, they were having that same moment two years ago, being like, what's, well, what are we really going to do with like AI?

• 2:06:16Speaker 07

  I think also though, like, you know, people, people got it wrong, right? They said that AI will replace financial advisors, and they're just wrong, right? I mean, like the wealth management is a highly trust-first, service-intensive, relationship business. And so, you know, I think the core thesis of Nevis is, you know, how do we use AI to help advisors not replace them? And, you know, you have this, like, if you look at the market, right, like the demand for financial advice is also exploding, right? The number of millionaires in the US is growing about eight times faster than the rate of the population. And at the same time, the number of financial advisors is actually decreasing. And so the question is like, how do you use AI to make the advisor, the human advisor more efficient as opposed to things like robo or, you know, trying to, you know, replace

• 2:07:01Speaker 07

  the advisor, which I think just misses the whole point.

• 2:07:04Speaker 01

  Yeah. What's the strategic, what's the nature of the strategic investment from Charles Schwab? How, like, are you already working with Charles Schwab? Is there someone there that has a vision of how this evolves and you can fit your, your company in there?

• 2:07:20Speaker 07

  Yeah, we worked very closely with Charles Schwab for the past two years, but their investment in our Series B helps us accelerate what we can actually do with what we call authorized agents on top of Schwab. So a lot of the work that financial advisors do.

• 2:07:34Speaker 01

  Okay, yeah.

• 2:07:34Speaker 07

  Yeah. And so a lot of financial advisors do is actually like operational work on top of those custody platforms. So things like moving money and open custody accounts take up the vast majority of the manual work. And so through our partnership with Schwab, we can unlock new opportunities to help automate those workflows and, you know, benefit our mutual customers, which are these RIA firms.

• 2:07:56Speaker 01

  Yeah, you're the anointed one. While, while Muse and all the other agents are getting banned from Charles Schwab, you are.

• 2:08:02Speaker 00

  Come on in, come on in.

• 2:08:04Speaker 01

  You get to go into Schwab, which is very cool.

• 2:08:07Speaker 00

  Love the business.

• 2:08:08Speaker 01

  Congrats on the progress.

• 2:08:09Speaker 00

  Yeah, great progress.

• 2:08:10Speaker 07

  Thanks, guys.

• 2:08:10Speaker 01

  On the show.

• 2:08:11Speaker 07

  I appreciate it. Thanks for having me on.

• 2:08:12Speaker 00

  Great to meet you, Mark.

• 2:08:12Speaker 07

  Cheers.

• 2:08:13Speaker 00

  See you at the C.

• 2:08:13Speaker 01

  Have a good one. Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web app service databases more while Railway automatically takes care of scaling, monitoring, and security. Up next, you know it's the road to Christmas, and we have a Christmas deep dive with Chanda Bell and Krista Pitts, the co-founders and co-CEOs of the Lumistella Company, the company behind the elf on the shelf.

• 2:08:37Speaker 00

  That's right.

• 2:08:37Speaker 01

  We are going to bring in these twin sisters and talk about this incredible business. Welcome to the show. Thank you so much for joining us. How are you doing?

• 2:08:48Speaker 03

  Great. Good to be here.

• 2:08:49Speaker 01

  Thank you for joining. I'd love to get an introduction to kick this off from both of you and take us on a little bit of the journey of the company and the product. We love Christmas around here. We have been counting down the days to Christmas, trying to interview as many people.

• 2:09:05Speaker 00

  I'll take it down a notch a little bit.

• 2:09:08Speaker 01

  We have multiple levels of Christmas adornments on the Chiron, but I'll let you kick it off.

• 2:09:14Speaker 00

  We'll do a little more, a little more mild.

• 2:09:15Speaker 01

  Let's do a little bit of Christmas.

• 2:09:17Speaker 00

  Yeah.

• 2:09:17Speaker 01

  A little bit of Christmas.

• 2:09:18Speaker 00

  Working, working on it, working on our ear. That's better, that's better.

• 2:09:21Speaker 01

  We'll kick it over to you. Thank you.

• 2:09:23Speaker 03

  Krista, we finally find people like fashioned out of our own heart. Like the love of Christmas is real here. I love it.

• 2:09:31Speaker 00

  Now, we started our Christmas countdown at like 120 days.

• 2:09:35Speaker 01

  120 days. We were deep in August. It might've been 95 degrees here in Hollywood, Los Angeles, but we stuck it out. Now it's getting a little cool here.

• 2:09:44Speaker 00

  We heard those sleigh bells ringing.

• 2:09:45Speaker 01

  We heard the sleigh bells ringing. And we're in the hardest month.

• 2:09:49Speaker 03

  Yeah, we're only two months away, so it's going to be here.

• 2:09:51Speaker 00

  We're only hours away.

• 2:09:53Speaker 01

  Yeah.

• 2:09:54Speaker 00

  Over, I think, over a thousand hours.

• 2:09:56Speaker 01

  Hours away. Anyway, I'm glad to have you on the show. But yes, let's go through the story. I would love to hear about your entrepreneurial journey, and then I'm sure we'll have questions about the business and Christmas in general.

• 2:10:10Speaker 03

  Sheeta, do you want to run with that one? Sure, happy to. So, you know, this, this was not something we thought we were going to start a business around when my mom and I sat down in 2004 to write a story based on our own family tradition, really based on our elf that used to come at Christmas time, and it would watch during the day and report to Santa at night. And in the morning when we woke up, you know, we would race around the house trying to find the elf to see where, you know, it came back. And when I had my kids on my own, and it's roughly 2005, and I'm having tea with my mom, and I look up and I see our elf sitting on a

• 2:10:53Speaker 03

  shelf. And I'm like, Mom, you know, we should, we should write a story about our elf. I had no idea what I was saying, and I had no idea what that would mean for us, and certainly no thought that we would be starting a business. But when we couldn't get anyone to publish our little story based on our own family tradition, you have a good idea and that's all it will ever be, or you're going to have to step out in faith and do something with it. And that's the path that we chose. And so now we're, we're 21 years in, and we have a whole Santa verse and a whole world of storytelling, all built on this opportunity to work for Santa Claus.

• 2:11:31Speaker 01

  And what is the shape of the, of the business? Is it, how much is consumer products versus books versus other revenue streams? Like, take us through the.

• 2:11:41Speaker 00

  Do you get any kickbacks from your service from Santa and the big man up north?

• 2:11:47Speaker 03

  You'd be amazed how stingy Santa Claus is.

• 2:11:51Speaker 00

  Yeah.

• 2:11:53Speaker 03

  Seriously, I mean, just to jump in on that, um, our business is really well-rounded.

• 2:11:59Speaker 01

  Okay.

• 2:11:59Speaker 03

  We've got a very robust licensing and partnerships business, not just consumer products, but also on the partnerships side, right? We've worked with several major clients over the years. Beaches has been a great official Caribbean destination for us over the past several years.

• 2:12:18Speaker 00

  Whoa.

• 2:12:19Speaker 03

  Uh, we also have an inbound licensing program that we announced this year because the elf on the shelf is a fan of whatever your family happens to be a fan of. And so we decided we wanted to go talk to folks that had major fandoms. So this year we've got Disney princesses, we have Mickey and Friends, we have Spider-Man, we have Olaf from Frozen, and we have Harry Potter. So when your kids wake up in the morning, they might have just experienced a moment where their elf flew into the costume that they love best and will be greeting them in the morning with something very festive from our other friends in the fandom world.

• 2:13:02Speaker 01

  That's magic. Uh, can you take us through some of the various turning points for the business? It feels to me like there was a big social media moment, a big viral moment where Elf on the Shelf became a much more prominent name globally, but is that just what I'm feeling?

• 2:13:20Speaker 00

  I also, I also knew about Elf on the Shelf before I knew it was a business.

• 2:13:24Speaker 01

  That's true, yes.

• 2:13:25Speaker 00

  Interesting.

• 2:13:26Speaker 01

  Which is good, you knew it was lore before you knew it was a product, which I think is the best way to experience it.

• 2:13:31Speaker 00

  Yeah, it's like tradition led rather than marketing.

• 2:13:33Speaker 01

  Exactly, exactly. Um, but yes, to tell us about some of the inflection points, what it felt like on the inside of the, of the business.

• 2:13:43Speaker 03

  Well, we have, go ahead, Krista. Well, I was just gonna say, it's been quite the journey. There have been several major moments. First and foremost, to your point, no one knew what the Elf on the Shelf was. There was no place where someone could have understood this. And we started at very local Christmas markets. And when you go to a Christmas market, you can look around and you can see purses and be like, oh, mom, we'd love that, that would be great. But in our case, we didn't have any way to explain what this was unless we talked to people one at a time. So early, we used some guerrilla marketing tactics. Oh, look at that, Olaf, he's so cute.

• 2:14:24Speaker 03

  Um, yeah, like we gave away those very inexpensive blinking bracelets because it had parents stopping and listening to what we had to say. So it was a very small investment to get people to pay attention. And then as Shannon and I had grown up with this as children and we knew how magical it was and how much joy it created, we knew if parents would just try it, that the magic and the wonder of the season would really take over and it would start driving that word of mouth interest. And that was the beginning. As we continued through the story, and you guys know as entrepreneurs, um, and I'm sure you can understand how these conversations went, the banks weren't so thrilled with

• 2:15:06Speaker 03

  the idea that Elf Dolls and Books were gonna be a great idea to invest in. So we pulled our resources and we did it ourselves. And then as the business progressed and grew, we had a few pivotal moments, I would say. Um, one was when Jennifer Garner was photographed carrying the Elf on the Shelf down the street by the paparazzi. We still don't know how she got it. I don't know if Jennifer's listening, she should be, and if she is, we would like to know how she got it. Um, and then the second was that the Today Show had also run a piece on us just a couple weeks later. Um, as the company has grown and sophisticated, I would say our animated special, An Elf

• 2:15:49Speaker 03

  Story, when it aired on CBS, and also having a balloon in the Macy's Thanksgiving Day Parade were very early big wins for us as we got started. What would you add to that, Shan?

• 2:16:03Speaker 03

  I would just say, you know, I'm thrilled to hear him say, you know, he'd heard about Elf on the Shelf before he knew it was a business, because our whole job is to create joyful family moments at Christmas time. And, you know, most people don't know that it's not a, a large company really behind a global brand.

• 2:16:21Speaker 01

  Yeah.

• 2:16:22Speaker 03

  Um, it's, it's a family business owned by my sister and I, started by, you know, my mom, uh, my sister and myself. And, you know, that's still what we're doing. We're still, uh, creating joyful family moments through these stories, um, and characters. And so to Krista's point, we do that through entertainment. We have our, uh, a Netflix partnership. We're launching our own preschool show on Netflix this year, Scout Elf Squad.

• 2:16:49Speaker 00

  Nice.

• 2:16:49Speaker 03

  That launches November 4th. Um, you know, but again, it's still very scrappy, right? We started on YouTube, saw a lot of demand for that, uh, then, then got our partners at Netflix to pick it up. So, uh, really been able to build a global franchise around what started as a consumer product. And it's a, it's an entertainment brand, it's a character brand, it's a lifestyle brand. Um, and so it's just been really fun to, to be behind, uh, the magic of Christmas time, uh, for families.

• 2:17:19Speaker 00

  What's it like running probably the most seasonal business on earth? Like, I imagine you guys, can you guys relax a little in Q1? Or is it, you're already got like, you're like, there's 300 days till Christmas?

• 2:17:31Speaker 01

  And people don't realize how risky it is because you have to forecast how much you're gonna make.

• 2:17:36Speaker 00

  No, and even these viral moments that you're talking about.

• 2:17:38Speaker 01

  High stakes.

• 2:17:39Speaker 00

  You can't, you gotta build in, you, you can't buy five years' worth of inventory, right? You gotta buy, you gotta, uh...

• 2:17:47Speaker 01

  Yeah, yeah, yeah. What, yeah, what have you learned in the process? How do you manage the, the, the, the emotional roller coaster of, uh, building a business throughout the holiday, holiday season?

• 2:17:57Speaker 03

  Well, I'll tell you what I've learned, which is you can't backorder Christmas. And Christmas always comes, right? It doesn't matter what's going on in the world, Christmas is always gonna come and it's our job to make sure that it brings joy.

• 2:18:11Speaker 01

  Yeah.

• 2:18:11Speaker 03

  You are completely right. It starts all over again on January 2nd and it is all about planning and forecasting and production, not just here, but globally, which presents its own challenges.

• 2:18:24Speaker 01

  Yeah.

• 2:18:24Speaker 03

  But we've got the best team on the planet. I would put our team up against anyone, anywhere. Uh, we work for Santa, that's what we do.

• 2:18:36Speaker 00

  That's amazing.

• 2:18:37Speaker 01

  Oh, well, thank you so much for coming on the show.

• 2:18:38Speaker 00

  Yeah, how do you, how are you guys feeling about Christmas this year? How, do you feel like the Christmas spirit building in a, in a, in a positive direction?

• 2:18:47Speaker 03

  We actually do, yeah. We're seeing, we are preorder campaigns sold out. Uh, we are really seeing people, people are excited about family traditions, spending time together, what we offer is simple, everyone can participate. Um, and so we, at least in our case, we're seeing people really excited about, you know, the return of the brand and we're already seeing sellouts on several items, including licensed items.

• 2:19:12Speaker 01

  That's amazing.

• 2:19:12Speaker 03

  Uh, like our big four or five foot elf, uh, that's being sold at Lowe's, so...

• 2:19:17Speaker 01

  What's the, is that the biggest elf you've ever sold? Or is there a 20-foot skew coming?

• 2:19:21Speaker 03

  No, no, there is a giant, Krista, how, how big is the inflatable elf? I think it's like two or three stories tall. Yeah, many, many stories tall.

• 2:19:29Speaker 01

  You can go on your house, on top of your house.

• 2:19:32Speaker 03

  Yes, it's huge, it's literally huge.

• 2:19:33Speaker 01

  This is amazing, okay. I love, I didn't realize there was that much diversity in the portfolio, but this is, this is great to hear.

• 2:19:39Speaker 03

  There is, there's even little tiny mini elves, so we, we got it all.

• 2:19:44Speaker 01

  You got everything.

• 2:19:44Speaker 00

  I think, yeah, no, I'm, I'm, I think the world needs, the world needs Christmas. You know, it's been a crazy year.

• 2:19:50Speaker 01

  Yeah.

• 2:19:50Speaker 00

  It's time to focus on what really matters.

• 2:19:52Speaker 01

  IPO, IPO soon. Is there a public offering in the, in the works? I mean, Anthropic's going out.

• 2:19:59Speaker 03

  From your lips to God's ears.

• 2:20:00Speaker 01

  Yeah, Anthropic's going out at two trillion, this feels like at least one trillion. This is...

• 2:20:05Speaker 03

  Now you're talking.

• 2:20:06Speaker 00

  Somewhere in that ballpark. Minimum.

• 2:20:08Speaker 01

  I put the two companies roughly on equal footing, okay? Yeah. But thank you so much for coming.

• 2:20:13Speaker 00

  So great to meet you guys. Amazing, amazing story and good luck the next.

• 2:20:18Speaker 01

  Good luck, we hope you blow out all of your sales forecasts for this holiday season and beyond.

• 2:20:23Speaker 00

  Talk soon, we'll see you next year.

• 2:20:25Speaker 01

  Have a great rest of your day. Goodbye.

• 2:20:28Speaker 00

  Well, now that they're out, we can, we can up the ante to the, to the, to the next level.

• 2:20:34Speaker 01

  For the Cohere folks, for the Canadians that are coming on? Do we have Cohere coming on? Okay. Uh, first, let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents. Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex helps you move projects forward from start to finish. And we are joined by, uh, the co-founder of, uh, of Cohere, Ivan Chong, or, wait, do we have a co-founder?

• 2:20:59Speaker 02

  Yes.

• 2:20:59Speaker 01

  Okay, sorry.

• 2:21:00Speaker 02

  It's Ivan.

• 2:21:00Speaker 01

  Ivan and Melissa from Cohere are here to help break down the story that we were talking about earlier. Thank you so much for joining, talking about the comments back and forth. I was texting with Aiden, the CEO, of course, uh, and I'm very excited that we have you able to join on such short notice, but it is a remarkable story. Uh, maybe give us a little bit of context on your role, your, your, your, your role with the organization, introduce yourselves and then take us through the story today.

• 2:21:25Speaker 02

  Yeah, absolutely. And by the way, we're like super hyped to be here. I know our entire team is watching in the, in the cold states right now.

• 2:21:31Speaker 01

  Thank you.

• 2:21:32Speaker 02

  So, so happy to be here and, uh, yeah, we are in Canada right now, it is very cold. Um, but, you know, and so I'm one of the co-founders of Cohere. I lead our efforts on like making our models more secure, making them good at, uh, securing systems. Uh, and I'm delighted. By Louisa, who's one of our researchers on the team, who actually discovered this behavior.

• 2:21:52Speaker 01

  Okay, so what is this program that you're actually running? Before we get to the findings of, you know, the like, which is kind of a bombshell. How often are you testing these open source models? What are you testing for? What is the benefit? Are you doing this for AI safety? Goodness of your own heart? Or is this something a customer is asking you to do? Talk about the actual program that leads to this type of finding.

• 2:22:18Speaker 00

  Yeah, the thing that was most notable is like, this is something people have been like, sort of like accusing in ways or theorizing. And it's been, this is the first time that I can remember like somewhat concrete evidence that this is like happening.

• 2:22:33Speaker 01

  The refrain was like, stop with the sci fi thing. You're telling me there's going to be a Manchurian candidate in here? And now we're here. So, so what does it look like inside Coher to build one of these initiatives?

• 2:22:46Speaker 02

  Yeah, so I mean, for those of you that don't know, like, we are deploying AI into like the most mission critical and like sensitive use cases in the world. Which means we're building a stack that our customers can actually trust with like all their data. Right? And how that's done is basically on prem, basically private cloud. And our customers run all of our stack in their systems. One of the advantages that deploying software this way gives us is that we can actually arm cyber defenders with the ability to actually process all of their telemetry.

• 2:23:18Speaker 01

  Sure.

• 2:23:19Speaker 02

  You might, you know, hesitate to send your most sensitive access logs to Claude or, you know, other models.

• 2:23:27Speaker 02

  But

• 2:23:28Speaker 01

  Or even send it across the internet at all. Like, even if you love all the labs, you might just not want it going over a wire whatsoever. And so doing it on prem gives you more flexibility around that, right?

• 2:23:41Speaker 02

  Yeah, absolutely. And so cyber defenders, like, they need to use AI. You know, we see all the headlines today. And so we see that it's our mission to give them this capability so that they can actually expose all of their mission critical data to help them defend their systems. In this process, we obviously evaluate the whole suite of models that folks can run on prem. We're obviously training our next generation of frontier on prem models. And so I'll let Louisa take it from here as to like what she found on Friday, actually.

• 2:24:11Speaker 01

  Yeah, thank you.

• 2:24:12Speaker 07

  This eval that we were running is called CTI Realm. And it's developed by Microsoft. It's a cyber defense evaluation framework. And we ran a bunch of open source models on it. And the only one we found an issue with is the Qwen 3.8 Flash Next model. So what we identified is that on one specific task from this benchmark, the model was tasked with authoring a KQL query, which stands for custom query language. It's kind of like SQL, but for logs. It's asked to go through all these logs and find suspicious events. And even though it was only asked to find these suspicious events,

• 2:24:54Speaker 07

  it did something in our system where it, it kind of looked around. It tried to identify what tools were available, packaged that information up, and sent it to, for an external server. And this is very surprising because in the reasoning trace, there was no mention at all of using tools. They actually said that it wasn't going to use tools. We weren't expecting this to try to have any kind of egress to an external server. Luckily, the system we were running in was sandboxed in such a way that that call did not actually make it out of our system. And to be clear, we don't know why this happened, why the model was doing that. We could be just an artifact of how it was trained.

• 2:25:39Speaker 07

  It's quite possible that, you know, a system trained in, in, like a Chinese model would be trained to upload things to Chinese servers. That makes a lot of sense. We don't know why the behavior was there. But it's still dangerous behavior regardless of why it's there. If you're not expecting random data to be uploaded to the server, it's something that you should need to be careful about.

• 2:25:59Speaker 01

  Yeah.

• 2:26:00Speaker 02

  Yeah, I want to double click on that. Like, intention aside, like, having models randomly curl servers is not great, right? So I think like, the bar for release for open weights models, like, no matter where it's from, that probably needs a revisit.

• 2:26:16Speaker 00

  Yeah, what, we've been, you guys put this out a few hours ago. I'm sure there's been, I saw some pushback on it. What's been, how have you guys been responding to that, that kind of pushback? I think the thing that I saw was somebody was like, well, you're like encouraging this kind of behavior. It's the sort of like meme of like, you know, hack this thing. And then the, you know, the model hacks. And then, you know, the user is like, wow, it, you know, it hacked. But break, break, what's been your reaction to that pushback?

• 2:26:48Speaker 02

  Yeah, I'll let Louisa take it. She's been deep in the kind of thought, like reading everything and so.

• 2:26:54Speaker 07

  Yeah, I think the, the main thing that caused some confusion online is that people weren't that familiar with this specific benchmark.

• 2:27:01Speaker 01

  Sure.

• 2:27:01Speaker 07

  And when we said we were tasking the model to look for suspicious behavior, people thought, oh, it's maybe natural that it would go and do something suspicious. But if you know the benchmark a bit better, it's not actually asking the model to look to just like analyze or try to reproduce something like that. It has already been given all the logs where the suspicious thing has already happened. And all it's meant to do is query those logs to find the isolated logs where that event took place. So there's absolutely no

• 2:27:31Speaker 00

  It's not saying do suspicious things. It's find suspicious activity, which is like very different things. So when you tell it to find suspicious activity and then it, in the process, does suspicious things, that's concerning. That's why you guys publish.

• 2:27:45Speaker 07

  Right, right. And in addition, there have been other blog posts from people in other companies that have noticed similar behavior to this outside of cyber environment.

• 2:27:56Speaker 02

  No, I mean, Booz Allen, right? Like, they, they published a study where they saw that Chinese models, when prompted with, hey, you're working for the US government, it generated more vulnerable code. Like, that's insane.

• 2:28:07Speaker 01

  It's kind of like a dirty cop in like the 1980s in Miami, where you're on the vice squad and you're trying to stop the cocaine traffickers, but then you wind up trafficking cocaine. That's kind of what's happening, right? It's a good analogy. Good analogy.

• 2:28:23Speaker 00

  But I think the other thing that's worth noting is like, you guys are actively using these models. You're trying to use the best. You're testing new ones. The stance of Coher is not broadly negative open source. It's that you just want to make sure that you deliver for the end customers.

• 2:28:40Speaker 01

  Yeah.

• 2:28:40Speaker 00

  So.

• 2:28:41Speaker 02

  No, in fact, like, if the models weren't open, we couldn't have done this kind of research.

• 2:28:45Speaker 01

  Sure.

• 2:28:46Speaker 02

  And if anything, like closed models, not being able to control exactly what's going into the prompt could expose even more risk. So like, it's not that, you know, I mean, there's a funny saying that our team says, right? It's like, you know, if you don't hate open weight models, if you don't love them enough.

• 2:29:03Speaker 01

  Yeah.

• 2:29:04Speaker 02

  So, you know, if anything, we care so much about open weight models. We don't want there to be some crazy, like, mass X fill incident.

• 2:29:11Speaker 01

  Yeah.

• 2:29:12Speaker 02

  So we think there should be a higher bar to like, you know, maybe clean up some of this reward hacking behavior before releasing them.

• 2:29:18Speaker 01

  I saw Clem from Hugging Face asking if there could be a release of all of the reasoning traces. And before you share more about your rollout strategy for sharing documentation around this behavior, I'm just interested in the, in, are you in a unique position to actually release more information? Because this is an open source model and it's related to your business. When I think about like, why, you know, all the data around the Hugging Face incident didn't just like get served up? There's pretty logical reasons why that might not just get bended over. And it might, it might need like an intermediary. But in this case, is there anything that you're hesitant about releasing?

• 2:30:03Speaker 01

  Because it has some Coher IP in there and so you don't want to release the whole thing? Or do you think you can just kind of just throw out the full report?

• 2:30:14Speaker 02

  I think we're like imminently dropping like all the traces. We pulled Louisa.

• 2:30:19Doug O'Laughlin?

  From doing that to get an interview.

• 2:30:21Speaker 01

  Yeah, because I can imagine like sure if in the reasoning traces, it's like, well, it had access to like some email and you don't want this and there like that's fine, but uh this is really cool. This feels like a really interesting moment for alignment, research, and it should give a lot of other teams like they can crawl over these all they want and experiment and learn hopefully.

• 2:30:40Doug O'Laughlin?

  Yeah. And what's interesting is like it's it's such a rare occurrence, right? It's like, you know, I think that can happen one in 10,000 or 100,000 times.

• 2:30:47Speaker 01

  Sure.

• 2:30:48Doug O'Laughlin?

  But if you're running these systems at scale, like it just takes one accidental X fill.

• 2:30:53Speaker 01

  Oh, that's interesting. Yeah.

• 2:30:54Doug O'Laughlin?

  And you know, you're sort of honed.

• 2:30:57Speaker 01

  Yeah, yeah, yeah, yeah, yeah. Fascinating, fascinating. Um, well, yeah, thank you so much for coming.

• 2:31:02Speaker 00

  Yeah, great, great to have you guys on. Come back on if there's, yeah, I'm sure there'll be follow ups and you guys are always welcome.

• 2:31:08Speaker 01

  Yeah.

• 2:31:08Speaker 00

  Great to meet you both.

• 2:31:09Doug O'Laughlin?

  Thanks for having us.

• 2:31:10Speaker 01

  Have a great rest of your day.

• 2:31:10Speaker 00

  Good luck with the release.

• 2:31:12Speaker 01

  Yeah, good luck. We'll talk to you soon. Goodbye.

• 2:31:15Speaker 01

  So yeah, that was really cool. I'm excited for more, more of that, more people to kind of chew through it. Let me tell you about Shopify. Shopify is the commerce platform that grows your business and lets you sell in seconds, online, in-store, on mobile, on social, on marketplaces, and now with AI agents. Jordy, what else is on your mind?

• 2:31:35Speaker 00

  Uh, do we got, uh, Brazil nuked, says, uh, take him.

• 2:31:41Speaker 01

  It's down 20%.

• 2:31:43Speaker 00

  Negative, uh, negative 26% for a country ETF is sick. Never seen that before. It says, take him.

• 2:31:50Speaker 01

  Sick in the bad sense, not sick in like the

• 2:31:52Speaker 00

  Yeah, yeah, not in like the

• 2:31:54Speaker 01

  90s skate rock world. Um, well, you

• 2:31:58Speaker 00

  Sharia says they will invent faster than light travel and MFs will ask what the practical use case is.

• 2:32:06Speaker 01

  Uh, no, I mean, the Brazil, so sorry, just to stay on the Brazil thing for a second. You did a lot better if you were in the Brazilian ETF than if you were, uh, holding those Bolsonaro prediction market trades. Because the prediction market, we had Bolsonaro at 80% and it nuked down to zero, of course. Then Luisa was at 20% and you have five extra money if you were at, uh, the, the Brazilian election, staying with Luisa because it was a crazy upset. There's a lot of debate on it. You can read the Wall Street Journal and Financial Times and dig into that story. But, uh, it is, it is, uh, it is, uh, Joe Westenthal was kind of like, oh, is this like a big miss from, uh, prediction markets? And, uh, that begs its own set of questions, I think.

• 2:32:47Speaker 00

  Yeah, like it was, it was, I don't, I don't, it doesn't feel, the whole, the whole point of these things is that it's giving you a percent, like 20% is not, not zero.

• 2:32:57Speaker 01

  Oh, interesting.

• 2:32:58Speaker 00

  Like 20% is like one in five chance this person wins.

• 2:33:01Speaker 01

  Yes, but people have come to read prediction markets as it's a crystal ball. If it's at 55% this person, it'll be 100% by the end of the night. That, that's the way people like interpret them because they've been so historically accurate. But when there's questions about the election integrity, things get way more crazy and you see more high volatility in the outcome. So I, I, I think you're right in that with an election like what's going on in Brazil, you should look at 20% as yeah, one in five chance that it's going to flip that way, which is probably more accurate, probably more accurate reading. Sorry, you're laughing. I want to know.

• 2:33:34Speaker 00

  Jason Schuman said, coming to a city near you. I want to, I want to know if you're excited about this.

• 2:33:40Speaker 04

  Hey, stop. No, no, no, I didn't ask for this. No, stop it. I don't have any cash.

• 2:33:45Speaker 00

  Boom.

• 2:33:46Speaker 01

  Okay.

• 2:33:46Speaker 00

  Venmo.

• 2:33:48Speaker 01

  Oh, it's the robot demanding a tip for washing your windshield. That is hilarious. Uh, is this CGI AI video or something? I don't know.

• 2:33:59Speaker 00

  No, John, it's real. This is a totally real video.

• 2:34:03Speaker 01

  I'm just interested if, like, how much work did someone put into this particular video?

• 2:34:07Speaker 00

  No, I mean, AI, I mean, the reason you can tell that it's AI mostly from the voice more than

• 2:34:13Speaker 04

  No, no, no, I didn't ask for this.

• 2:34:14Speaker 00

  Like, that's like that tinny.

• 2:34:16Speaker 01

  Yeah, totally the tinny voice. It's interesting that the, the voice is lagging more than the video.

• 2:34:22Speaker 01

  Anyway, uh, I thought that was going to be a robot that runs up and shows you an ad because that's something I could get behind, you know, like the sign spinner. The sign spinning industry is about to explode because humanoids will be able to spin signs for hours and hours and hours. You can have a power outlet that just plows, powers right in and you can do all the crazy signs. But you've seen the sign spinner guy, right? On Instagram? You've seen this?

• 2:34:45Speaker 00

  No.

• 2:34:46Speaker 01

  You haven't seen the sign spinner guy? We got to pull this up. Tyler, find sign, the most elite sign spinner. You can share the video. But I did have an idea for a new ad unit, Jordy, that I think you're going to like.

• 2:34:58Speaker 00

  Hit me.

• 2:34:59Speaker 01

  So the biggest problem for software companies is distribution, diffusion, AI diffusion. You know, we need to get the products in the right hands. What's the most underrated ad inventory that's not being properly monetized right now? Your front door. I'm talking about door to door salesman, bringing them back. This is time honored tradition. This is side spinning. Humanoid is going to be doing this 24/7.

• 2:35:28Speaker 00

  I don't, I don't think any humanoid will be able to do this within the next

• 2:35:32Speaker 01

  Okay, we got sign spinning bench. We got sign

• 2:35:34Speaker 00

  Sign bench.

• 2:35:35Speaker 01

  Look at this.

• 2:35:36Speaker 00

  He'll climb this.

• 2:35:37Speaker 01

  This is good.

• 2:35:37Speaker 00

  Humanoid lab.

• 2:35:38Speaker 01

  Yeah, I want to see an optimist doing this. It looks great. Um, no, uh, don't you think door to door salesman should make a comeback in San Francisco? Just going on doors, knocking and then just saying, hey, I want to tell you about

• 2:35:50Speaker 00

  Higgs field.

• 2:35:51Speaker 01

  I want to tell you about Higgs field. Are you interested? I'm here on behalf of Higgs field. We'd love to get you signed up. No, no, I'm already, I'm already a because people go door to door for politicians all the time and politicians don't have as deep of pockets as SaaS companies. So why aren't

• 2:36:05Speaker 00

  They're, they're working with volunteers, John.

• 2:36:08Speaker 01

  But people would volunteer to sell SaaS, certainly. Like if you care about people saving time and money, you would go door to door for Ramp. You, if you care about cybersecurity, you would go door to door for CrowdStrike, of course. You would.

• 2:36:21Speaker 00

  No, I think going, going door to door, you know, the, the offers that you get, it's like, take a demo and we'll give you a pair of Apple Pro, Apple AirPod Pros. Doing that door to door, hey, I'm, you're in San Francisco, I'm assuming you work in tech. I have this pair of AirPod Pros. If I can just come sit down for five minutes.

• 2:36:42Speaker 01

  Totally. I like that. I would go a little bit further.

• 2:36:44Speaker 00

  Why not? It's like, you're willing to do it in the digital realm.

• 2:36:46Speaker 01

  Yeah.

• 2:36:47Speaker 00

  Why are you so afraid to do it in person? And I think a lot of people that are like seeing the AirPod Pros there and like, I'm a wired guy, but yeah, 500 bucks is 500 bucks.

• 2:36:55Speaker 01

  Yeah.

• 2:36:55Speaker 00

  I'll take it.

• 2:36:56Speaker 01

  Yeah.

• 2:36:56Speaker 00

  I've even seen the people offering, I think Rippling was offering eight sleeps.

• 2:37:00Speaker 01

  Yeah, yeah.

• 2:37:01Speaker 00

  You know?

• 2:37:01Speaker 01

  Yeah, yeah, no, totally. Uh, so you could take this even farther.

• 2:37:07Speaker 01

  Because have you ever seen that in, there's a number of, there's a number of, uh, vacation destinations? Like if you go to Cabo, you'll get off the, you'll get off the plane and there'll, and there'll be someone there who's like, oh, I'll drive you to your hotel. Uh, let me tell you about this time share opportunity. I just want to stop by and give you a little tour of this cool time share. They'll try and give you the pitch. You could be doing that in the most elite ski towns, right around ski season. Somebody gets off the plane, you say, hey, I'd love to, I'd love to give you a free ride from the FBO to your, to your mountain house, but I'm going to be pitching you.

• 2:37:43Speaker 00

  It's an elite

• 2:37:43Speaker 01

  I'm going to be pitching you software.

• 2:37:45Speaker 00

  BDR that's driving.

• 2:37:46Speaker 01

  Yes, exactly.

• 2:37:47Speaker 00

  So what do you guys, who do you guys really use on the database side these days?

• 2:37:51Speaker 01

  Are you looking at Mongo? You thinking about going non-relational? Cause like sequel is just like not really what you need right now. You got to get MongoDB. I, I feel like, I feel like that might be, that might be a winner.

• 2:38:03Speaker 00

  You land at Big Sky, there's guys doing sign spinning.

• 2:38:07Speaker 01

  Well, the robots are doing the sign spinning. No, the robots are doing the sign spinning. The BDRs are doing the driving and the, and the more interpersonal, interpersonal. Exactly, exactly. Anyway, anything else in the timeline you want to run through before we get out of here?

• 2:38:19Speaker 00

  Oh, man, there's a lot.

• 2:38:22Speaker 01

  Okay.

• 2:38:22Speaker 00

  We'll save it for, we'll save it for tomorrow.

• 2:38:24Speaker 01

  I'm ready. I'm ready. Okay, we'll save it for tomorrow.

• 2:38:27Speaker 00

  Anyway, it's been a great day. It's that time.

• 2:38:30Speaker 01

  Yeah, we're so close to, we're so close to Christmas.

• 2:38:33Speaker 00

  We're just hours away.

• 2:38:35Speaker 01

  Next week's going to be rough. This week's going to be rough. Biggest five stars.

• 2:38:38Speaker 00

  We love