Davos Reactions, Waymo’s Safety Case, Czinger 21C Hypercar Tour | Martin Shkreli, Bradley Tusk, Lukas Czinger, Andy McCune, Michael Broukhim, Aaron Katz

26:00Waymo has been doing it for decades at this point. They have insane CapEx and insane OpEx. There's humans in the loop a lot of times. And it's just like the most advanced team. It seems like it's the most advanced technology stack. There's a lot of advantages. And that's actually showing up in the data. And so her main dispute with Zipper is over what data she's including. He's including in making this claim that, hey, we don't know if robo taxis are actually safer than human drivers. And she gives this great analogy that I really, really enjoyed. So she says, imagine someone writes a piece and they say, we don't know if airplanes are safe.

26:39Some people say that crashes are extremely rare and others say that crashes happen every week. And when you investigate this claim further, you learn that what's going on is that commercial aviation crashes are extremely rare. This is true. While general aviation crashes, small personal planes, including ones that you can build in your own garage, are quite common. And so, yes, if you lump that together, you're going to get this muddy picture of aviation. But you shouldn't use that to sort of fearmonger about getting on a 747, because 747s just so rarely crash. But yeah, if you build your own Cessna and you're modding it and you're taking it through the canyons, you've got to watch out.

27:24Ask any recreational pilot if they've ever gotten into any hairy incidents. Totally. They'll tell you like, oh, yeah, well, this time I was flying. Yeah. And the wind came on super strong. Yeah. And I landed and I was basically, you know, 90 degrees. And a lot of times that counts as a crash. Like, I mean, Harrison Ford has like crash landed on a golf course. And I believe no one was injured. I think this happened multiple times, actually. By Harrison Ford? Harrison Ford. Yeah, he flies his own planes. But he flies like old planes and stuff. Like he's like a true enthusiast. Like he's, he's, he's, he's like a plane guy.

27:57How many simple flying.com? How many plane crashes has Harrison Ford had in the last 25 years? The actor has been involved in several incidents involving emergency landings, rescues and runway incursions. Exactly. Like it happens. It sounds crazy. It is crazy. I would never want to be in any sort of plane crash, but there's just way more. There's just way more planes flying around when you include those Cessnas. And it's not really fair to put that on United Airlines or American Airlines. 13th of February, 2017, Santa Ana, California. Single engine aviote Husky flown by Harrison Ford was cleared to land on runway 20L.

28:38Ford's aircraft reportedly landed on a parallel taxiway over flying American Airlines 737. It was holding short of runway 20L. 5th of March, 2015, Santa Monica. Ford was a pilot and sole occupant of Ryan ST3KR Recruit, a two-seat open cockpit aircraft that was used extensively. Open cockpit. Open cockpit. Think about that. You're flying and the wind's in your hair. Like, that's not a normal experience. That's not coach on American Airlines. The aircraft was used extensively as a training aircraft by the U.S. military in World War II. According to a preliminary report, Ford reported a loss of engine power shortly after taking off and was attempting to return to runway three.

29:21So he ended up crashing into a field. Ford chose to land on a nearby golf course, clipping the top of a tree before landing. If you just took Harrison Ford, he's probably more dangerous than American Airlines. June 2000, a landing in Lincoln, Nebraska. Ford's plane departed the runway due to a gust of wind. The aircraft sustained minor damage. Oh, no. 23rd of October, 1999. Wait, there's a fourth one? I thought I was exaggerating. Ford was on a training flight in a Bell 206 helicopter when he and the instructor made an emergency landing in a dry riverbed near Santa Clarita, California. That's so crazy.

29:57Although the helicopter rolled over on its left side, neither Ford or the instructor were injured. Yeah. That's all that I can find. Okay, I have another plane story. I mean, imagine four major incidents make it into the record. He's got to have at least 10 more. Yeah. He's just a sketchy situation. No one wrote a piece about it. Let's not talk about that. Let's not talk about that. Let's agree. Hey, calls this PR person. Hey, let's clean this up. How about you take a picture of me outside Erwin? Call the Paps. I'm willing to get coffee with somebody. Give me an Erwin Brandon smoothie. We need a new story.

30:29We need a different story. I want to be on the front of the New York Post. I'll take the bad picture on the beach of me falling off or something. Anyway, Cognition, they're the makers of Devon, the AI software engineer. Crush your backlog with your personal AI engineering team. So I actually have a friend who's been on the show. I'm not going to dox him, but we have two friends. Both of them fly. One of them is extremely serious, has IFR rating. So there's VFR, which is visual flight rules. That's like you take the plane up. It has to be clear because if you get into clouds, you're not trained for that.

30:58You can't fly at night. Exactly. IFR is you're flying on the instruments, instrument flight rules. and so we have one friend who's like super by the book, super safe and then we have another buddy who's a little bit wilder, VFR and he will rent like the cheaper planes and one time like the dial lights went out and he was landing at night, shouldn't be flying at night at all just like botched the schedule and didn't check sunset and he has to use his phone flashlight while he's landing to look at the dials as he's coming in on this rickety like old plane into these flying. People get crazy out there.

31:36They get confident. It is very dangerous. But the, just like if you're, if you're getting into personal aviation, flying your own planes, the, the, the jump in safety from VFR to IFR is significant. Like the, the IFR guys like crash way, way less. So highly recommend that if you are getting into flying your own planes. Anyway, Restream, one live stream, 30 plus destinations. If you want to multi-stream, go to Restream.com. So Waymo is funny to write about because there's still a class of people that are writing the stories about like, hey, Waymo is actually safe. And I'm like, I completely buy it.

32:12I'm in. Roll the Waymo's out everywhere. I've been in them. I feel safe in them. I've seen the data. I think they're safe. I'm just like, my eyes glaze over when someone's like, I'm going to blow your brain out. I'm going to blow your mind. Like, Waymo's are safe. Robot are safer. I'm like, yeah, I'm in. I'm totally in. But there's still this debate. Most people in tech are debating about just Waymo's unit economics and like, can they beat Tesla? What happens there? And so typically, I don't really wade into the safety thing because I think it's very safe. But this Piper piece, this piece by Kelsey Piper, did make me wonder about what happens if you further segment the human driver population.

32:51So we need to segment robo-taxis into Waymo and everyone else because Waymo is so good. But how does Waymo compare against the best human drivers? And I wasn't thinking like the Lewis Hamiltons of the world. I'm sure Lewis Hamilton could be superhuman at driving if he needed to. I wanted to think about it more as like if I'm an insurance agent and I'm underwriting a class, a group of people, what would I be going for? And are they better than Waymo? And you can't look at fatal crashes per 100 million miles because there have been so few fatal crashes in Waymos and there's just not enough data yet to look at that.

33:26So what you can look at is any injury crashes per million miles and the Kelsey Piper piece is really good it goes into a bunch of different methodologies for data collection looking at Incidents where Waymo's airbags release because that's a very binary moment where there's a question about like well What's an injury like if somebody like got motion sick and then they had to go to the hospital Like does that count as an injury or like what if they were getting out of the car and then they fall like does that count? So there's all these things where it's like does it count? So anyway, the best data that we have on Waymo is any injury crashes per million miles.

33:59For the average human, it's about four any injury crashes. So any injury that counts, four per million miles. Waymo is way better, way better. 0.75, so about five times safer than humans broadly. But there's one very specific case where humans, I believe, are safer. Now, there's debate about the data. I just had Claude crunch it all together and try and pull a bunch of stuff together. But I think this holds. So a married 60-year-old woman who's college-educated and is driving a large luxury SUV in Massachusetts on a Tuesday morning is closer to 0.5 injury crashes per million miles. Is a Tuesday morning part real?

34:41Yes. So the most dangerous time to drive, and I'll flip it around. The most dangerous time to drive is midnight on a Saturday because that's when people are at the bars. They leave. They're drinking. And so it's dark. It's the weekend. There's a lot more people on the road. They're driving much faster. They're getting places they might be inebriated or under the influence of something. And so Tuesday morning is the safest time to drive. And so basically I looked at all the government IHTSA data. It's the Transportation Authority. I'm getting the acronym wrong. But you look at all the different segments, and then you add all those together, and you get kind of like the synthetic safest driver cohort.

35:22It's a little loose. But it does seem like if you looked at that cohort, they're still beating Waymo. Humans still got it. At least, like, kind of. It's pretty close. It's honestly neck and neck. But so none of this should take away anything. None of this should take away. Feels like a Nathan Fielder episode in the making of, like, studying the women in Massachusetts. Yeah, like what do they do? Saved drivers on the planet. Yeah. Just interviewing all of them. Married 60-year-old college-educated women. Just like the exact wrong conclusion. Oh, yeah, yeah, yeah. We need to get everyone married and college-educated.

35:56No, the conclusion is everyone needs to move to Massachusetts. Yes. Well, it's the married 60-year-old women that need to be the taxi drivers everywhere in the world. Yeah. They need to be. It's like instead of Waymo. Yeah, he starts like a competitor to Waymo. Ride hailing. That's just them. Yeah, that's just them. Yeah, but it only works in Massachusetts. So clearly Waymos are very safe. Rolling them out nationally would be a huge safety upgrade for our society broadly. But it's worth noting that the Massachusetts married women still got it. But if you want to flip this around and find the hypothetically riskiest driver profile, it's very funny.

36:33It's an 18-year-old single male with a DUI driving a Dodge Challenger at midnight on a Saturday in rural Mississippi. That's as risky as you can get. And that's like an order of magnitude or 40 times as dangerous as the Massachusetts driver. Yeah, but the interesting takeaway is that you said earlier off the show is like Waymo is actually not a superhuman level driver. Yes, yes. Because in order to be that, you actually have to be better than the best human, right? Exactly, exactly. So we need a word for not above average and in the top cohort of humans, but not better than any human ever. Because a lot of AI tools are getting there.

37:20Well, I guess, okay, so I actually looked up the definition. So superhuman means having powers, abilities, or qualities beyond normal human capacity. So it's potentially these women in Massachusetts at their superhuman as well as Waymo. So maybe we do need to give Waymo a little more credit. Well, if you're looking for superhuman payroll, head over to Gusto, the unified platform for payroll benefits and HR built to evolve with modern, small, and medium-sized businesses. It's the backbone of this show. It is. Max Hodak had a post about Waymo. He said, interesting benefit of Waymo. It always matches instantly since it knows exactly what cars are available and that they will always accept.

38:01And it can also provide accurate wait times before requesting. And so lots of love for Waymo. Interesting to see where they'll roll out. The really interesting thing is that we've had a number of people on the show advocate that rolling out Waymo is a societal good. It will reduce the number of traffic utilities. That's all true. But if you really want to accelerate the impact that Waymo will have on safety, you've got to deploy it in Mississippi before you bring it to Massachusetts. And I just feel like Waymo is coming to Boston before it comes to Vicksburg. But I don't know. It was fun. There's other love for Waymo on the timeline.

38:36First, let me tell you about Okta. Okta helps you assign every AI agent a trusted identity. So you get the power of AI without the risk. Secure every agent. Secure any agent. Mike Volpe. He said, I have to say I love that Waymo operates from San Francisco to the South Bay. So great. Some folks say the freeway driving is so-so. I hadn't heard that. uh the antidote to that is i get uh into an uber and and do the same drive it's literally life threatening i was driving i was driving uh with paul the other day and i heard uh he was telling me about the different modes that tesla has i've never owned a tesla so i wasn't familiar but did you know that tesla actually has something called mad max mode the autopilot where it'll drive it It'll drive it roughly around 85 on the freeways.

39:27And if you're in the fast lane and you come up to somebody, it will merge over into the lane to the right, make a pass, and go back to 85. That's insane. Somehow it feels slightly illegal, but I like it. Okay. MongoDB. Choose a database built for flexibility and scale with the best-in-class embedding models and re-rankers. MongoDB has what you need to build what's next. We have an update in the Warner Brothers Netflix deal. Netflix, they're going all cash. All cash deal. Netflix had said to Paramount, hey, what we want is an all cash deal. David Ellison and the Paramount crew, they brought Warner an all cash deal.

40:13And Netflix or Warner was still saying, hey, we're going with Netflix. Now Netflix has upgraded. So we can read through a little bit of this from the Wall Street Journal. So Warner Brothers Discovery and Netflix said Tuesday they struck a new all-cash deal for Netflix to buy Warner Brothers Studios and HBO Max. Warner also released financial details on the cable network's business. It plans to spin off the all-cash deal of$27.75 per share replaces Netflix's previous cash-in-stock deal. The sweetened offer comes as rival bidder Paramount continues pushing its all-cash offer for all of Warner Discovery.

40:46The value of Netflix's offer remains$72 billion. Warner and Netflix said they expect the new structure to enable Warner shareholders to vote on the deal by April. The change could also help sway some shareholders who might be weighing its bid against Paramount's. Netflix co-CEO Greg Peters said in a statement that revised agreement demonstrates our commitment to the transaction and accelerates the process for Warner shareholders. The new Netflix agreement reduces by$260 million the amount of Warner debt being placed on Discovery Global, the company that will house cable channels, including CNN, TNT, and Food Network.

41:21There's also another, the next kind of thing that I was going to get to is interesting. CNN is doing$600 million in profit on revenue of$1.8 billion. What are we doing? They are truly newsmaxing. Quick sidestay. for CNN. Congratulations to everyone at CNN. Not bad at all. Absolute dog. Of course, a ton of debt strapped to the broader group. And Peter Kafka is saying that private equity should pick it up. It's a shrinking business that throws off cash. It's perfect for private equity. Let's give it up for private equity. I know Peter's a big fan. I'm a big fan. somebody somebody uh you know i was saying yesterday that maybe maybe ai could save hollywood yes somebody would uh i knew i knew that was gonna piss a lot of people off for a lot of reasons i wasn't really clocking what you were saying either and so we should talk about it more but continue well one of the critiques was like you know uh interesting to say that la is dying as you're building a media business from la yep and my response is that we are in a huge complex we We're recording the show from a huge complex.

42:31And every other building in the complex is empty pretty much every single day. Like there's no, there's nothing going. When I said LA is like dying, I was specifically talking about the context of the entertainment. Yeah, I mean there's that Wall Street Journal report that just shows sound stages were typically leased 90 % of the time. And it's just way off peak down to like 60%. And so there is like a broad trend. And then just in general, like, I don't think we're necessarily at a phase in our life where we're like, okay, tabula rasa, where's the best place to build this business? Let's go find that.

43:14It's like we live here. Yeah, we didn't. We're not like, okay, we should go to Atlanta or something or go to New York. Yeah, we certainly didn't say like, let's start a media company because we live in LA. It's like the current struggles that LA is having are not going to hold us back. If anything, it's going to be easier because we can get space and hire people. And 10 years ago, this space would have been like two, three times the price to rent. The bigger question is, you know, there are these blips in industry towns that have happened many times. New York, famously in the 70s and 80s, went through a really rough patch.

43:50San Francisco, during COVID, everyone was moving out and there was a real vibe of like, oh, is San Francisco going to be less relevant in tech? And then it came roaring back with the AI labs. And I think most people don't debate that cities go through rough patches and then can come back. And so we certainly hope that LA comes back. It would be fantastic. But there are a lot of changes that need to be made. And this has been said by a number of famous people in Hollywood and Hollywood actors who have talked about different tax regimes, different incentives, different changes structurally to the way content is made and distributed.

44:28There's a lot of different elements that are contributing to L.A. having a rough go. At the same time, it's a fantastic city. It's where I was born and raised. I love it. I'm not going anywhere. You're going to have to pull me out of here. He's not leaving. He's not leaving. Gemini 3 Pro, Google's most intelligent model yet. State-of-the-art reasoning. Next Level Vibe, Cody, and deep multimodal understanding. In its Tuesday filing, Warner projected$17 billion of revenue in 2026, adjusted, I won't explain EBITDA, but$5.4 billion of EBITDA. Those figures are expected to decline to$15.6 and$3.8 billion, respectively, in 2030.

45:07The only individual network projections were CNN, which I just shared. They're expecting to increase revenue to$2.2 billion in 2030 and expecting to continue to grow EBIT as well. Well, Netflix seems to be doing everything right. As much as I enjoyed the David Ellison profile and learning more about how seriously he takes his business, it does feel like Netflix is just willing to pull out all the stops to get this across the finish line. And if they were able to match Paramount on the cash deal nature and all the different consideration, deal with the regulatory hurdles, it might take a long time, but it does still feel like they will land on it.

45:52Yeah. And again, I mean, it just feels like I'm so curious to see how this works out because it felt like the Ellisons were just putting on this absolute master class and then this one and then kind of ran into a brick wall with this, right? You could argue they overpaid by a ton for the UFC, which made sense if you had this. Again, like if you have all these assets under one roof, it is going to be a really compelling consumer product, consumer subscription. But currently kind of unclear who this kind of platform is really going to be for. Yeah. It was also surprising to me because Netflix for 20 years has been pure play tech company, right?

46:42And many of the previous tech companies, when they come out, like they haven't done a lot of acquisitions of legacy assets that I'm aware of maybe. But I'm just thinking like social media sort of like disrupted the news business. They didn't – Meta didn't go out and acquire the New York Times. They could have done some sort of hostile takeover. They have the money, but that never really figured into the business. And they wound up just working together synergistically. And now the New York Times has been growing across social media very effectively. And they've adapted their business model. Both businesses are doing well.

47:14They're sort of disruptive to each other. But ultimately, they never decided to buy the legacy player. So it was always something that was never on my bingo card of, oh, Netflix is going to buy a TV channel or a movie studio or a traditional Hollywood player. That just didn't seem like something that was going to be logically happening anytime soon. And yet here we are. And it makes a lot of sense when you think about the assets and the IP specifically and the longevity that that will bring when it comes, when you get Batman and Superman on Netflix, that's going to be exciting. Anyway, Sentry.

47:52Sentry shows developers what's broken and helps them fix it fast. That's That's why 150 ,000 organizations use it to keep their apps working. We got a deal. We got a pact. It's not just a deal, it's a pact. OpenAI and ServiceNow, they struck a deal to put AI agents into business software. ServiceNow, one of the greatest software companies in history, one of the few to reach revenues at their scale. The line from the Wall Street Journal is, the AI model maker and business software provider, They signed a three-year pact. I like the word pact. That underscores how AI - The partnership economy is over.

48:29Yes, you got to be pact. Press release economy is over. Yeah. This is the pact economy. No more deal guys, no more deal titans, deal makers, pact makers, pact guys. That's what's next. This is going to underscore how AI agents are increasingly being embedded in corporate software. You heard Dario talk about where he's potentially going up against other labs on corporate B2B deals. He said OpenAI, he said DeepMind, he didn't really say, he said no China labs have ever competed. Well, ServiceNow went with OpenAI on this one. So they signed a three year deal that will integrate the AI models, ChatGPT's models, into ServiceNow's business software.

49:07The deal depends on customers using OpenAI's models within ServiceNow and also includes a revenue commitment from ServiceNow to OpenAI. Enterprises want OpenAI's intelligence applied directly into ServiceNow workflow, said Brad Lightcap, OpenAI's chief operating officer. Look ahead. Customers are especially interested in agentic and multimodal experiences so they can work with AI like a true teammate inside of ServiceNow. We've seen a lot of other companies add. We've talked to Mark Benioff about Slackbot. We've talked to Satya Nadella about Copilot. Like if you have one of these massive software suites, enterprise software suites with just so much surface area, You're going to want to vend LLMs and agents into every little nook and cranny, and you probably don't want to build your own lab because it's going to be incredibly expensive when OpenAI will go and marshal the capital and build the data centers and train the models, and then you can just focus on the implementation and delivering value for your customers.

50:07So Salesforce has already folded AI agents into its flagship sales and marketing tools. SAP and Workday have been similarly pitching businesses on the idea that their embedded AI agents are essentially getting the value out of AI. The deal comes as OpenAI looks to build its enterprise business, both by selling direct access to its models and AI agents and through business software collaborations like with ServiceNow. It bases competition not just from rival AI model maker Anthropic, but also Google and Microsoft suite of enterprise AI tools. For ServiceNow. I want to find a company that is using the agents at ServiceNow or Salesforce or a company like that and saying, I can't live without these agents.

50:54I love these agents. We got to find one person. Yeah. I wouldn't be surprised if it's just like really sort of like maybe a year or two behind on the actual implementation. But it's just like, oh, I get like daily summaries of what's going on in the system. And it's a little bit nicer and neater. And yeah, I could probably extract all the data and write an agent that does that. but you know it was just one click of a button to add it to my installation and I did it and it's adding a little bit of value I don't think you're gonna see anyone who's like it's transformational it just works on its own I don't do anything it's more it's much more incremental than that it's it's it's still in like an autocomplete phase little helper tools there it's Clippy Clippy's going into every piece of enterprise software and so people are having fun service now increased revenue 22 % in the third quarter as AI demand from both existing customers and new customers grew, particularly in its customer relationship management business.

51:59The software company will report its fourth quarter earnings and full 2025 financial results on January 28th. We've got to do a deep dive that day. Anyway, Applovin', profitable advertising made easy with Axon.ai. Get access to over 1 billion daily active users and grow your business today.

52:20What SaaS founders should know about entering the Japanese market? Yeah, I found this on Hacker News, and I just thought it was interesting because we've talked to a few folks about expanding into Japan. It's not been on the top of the roadmap for many founders that we talked to. But I thought this was interesting, and I thought we could go through just a little bit of it, and then you can read the article if you want. So this is from embedworkflow.com. In one sentence, what's the difference between selling SaaS in Japan versus other countries? And the answer is the process is different. You just can't treat sales and go to market as a whole in Japan the same way you would in North America or Europe.

52:58The SaaS sales process in Japan is structurally different from the United States and other Western markets, primarily in terms of pace, decision-making style, and buyer behavior. It's not that Japanese companies are unwilling to try new software, but rather that they prefer to do significant research before engaging directly with a vendor. So many successful SaaS companies in Japan structure their websites with two primary calls to action. So one is just for downloading product documentation. And then the second is for booking a demo or starting a free trial. And most Japanese leads enter through the documentation path first.

53:38In America, we just want to get on the phone. We want to head straight to the instructions. I'm not reading the instructions. No way. I don't have time to understand how this thing works. I need to go straight to a steak dinner. Steak dinner first. Then I'll figure out what your company does. But let's talk about golf first. Let's talk about a track day first. And then you can give me the pitch for your company. That would honestly, the steak dinner button on a SaaS website, it's like you can try the product now or there's a button that's just like a steak emoji. And if you click that, you can get steak today on demand with somebody on the team.

54:13Yes. No matter where you are. No matter where you are. It just agentically books a steak restaurant. I'm tired of all these people that are like, oh, oh, you know, just take a demo and we'll give you some AirPods. Spend the money on steak. Take the AirPods budget and put a steak button on your website. Steak button. And have the steak flow. The steak flow. Steakholders. Steakholders need steak. So in the US, the typical flow is that a prospect visits a website, books a demo on their own, or drops off and is followed up aggressively by sales. A conversion happens quickly. A proof of concept or trial begins and the deal either closes or the buyer moves to a competitor.

54:52The cycle is fast and transactional. Buyers are comfortable testing tools, replacing them, and switching vendors if something doesn't work. In Japan, the process is slower and more deliberate. The company will usually download product materials first. The vendor then reaches out and begins a longer engagement process focused on education and relationship building rather than pushing for an immediate demo or trial. The goal is to support the buyer's internal evaluation process. The lead will take the information back to their organization and begin internal discussion about whether and how to proceed.

55:25So I just thought that was pretty interesting. How do you gain trust from the Japanese market? Part of it is showing your commitment to the Japanese market. where simple things matter. Is your product localized? Is your documentation localized? Do you have a support person that I can speak Japanese with? That's one layer. And the other layer, which is harder to potentially navigate at first, is the social proof aspect. So you need to get some social proof. Yeah, I'd be so curious to know how a company like Notion has done in Japan. It's product-led. They do have enterprise as well. But again, it's a product that just kind of scales naturally.

56:03But yeah, for so many different kind of like vertical SaaS players, it would feel like unless the Japanese startup ecosystem is asleep at the wheel and not paying attention to what's working outside, like can you imagine as a vertical SaaS player going and trying to compete in Japan against a local, like a talented local team? It feels like it would be very hard. It feels like you just get smoked. Yeah. Anyway, Vanta, automate compliance and security. Rudy. Vanta is the leading AI trust management platform. Max Hodak is on a tear. We got to do this one before we bring in our guest. Fun fact, rather than asking Claude to double check its own work, you can lie to it and say Codex went romping through the code and left bugs and it should find them.

56:44I like the idea that somewhere in the weights, all the drama of the AI lab battles are being encoded and creating bitter, deep rivalries. It's, yeah, it's very funny. I saw something else about Claude just rejecting a prompt whole cloth. Did you see this? Somebody was like, rewrite this entire app in jQuery. And Claude was just like, no, that's a bad decision, bad use of resources. I'm not doing it. I don't know if that's real, but I thought it was very, very funny. Anyway, we have our first guest of the show, Martin Shkreli in the Restream Waiting Room. We will bring him in the TBP Ultra. Martin, how are you doing?

57:25Great to see you. What's up, guys? Good to see you. Happy New Year. How is your year going so far? Great. I'm a new dad, so that's... Congratulations. They're pretty wild, yeah. Congratulations. That's fantastic. Is it... Has it been wild, or is it... Where does it rank up the list of challenges? To me, parenting is – it does feel natural. And so some people have this idea of like, oh, it's so insane and it's so complicated and I have to read 20 books and other people is like kind of follow – just follow kind of your own nature and just like take care of your child. It is straightforward. I'm following my nature.

58:09I mean I'm following my nature. The kid's got to produce at some point. He started to come into work like once a month. There we go. You know, comes in, goops, gag ads, you know, entertains the troops here. Perfect. But at some point, you know, some trades, something, do something. Yep. You know. Put some trades on. Yeah, you got to find some alpha. Have some conviction. Yeah. How big is, what's the environment like in your office? You guys have remote component? Are you anti-remote? No remote. Remote is evil. you know i think that if you're going to war together you can't go to war remotely you know it's uh you know my people fought the ottoman empire and you know one one man you know famous general you know stopped the ottoman empire largely by himself and uh at least that's a legend you know and if we're gonna take on blueberg or whoever we're fighting um you know i just don't think you could do it with some guy sitting in you know at home with this what's the current What's the current battle that you guys are fighting, and how is the war going overall?

59:13The war's going great. We just closed a seed round. We raised five. Whoa. All right.

59:21Congratulations. We raised two like a year ago to get the ball rolling, or a year and a half ago. We raised five as a seed round, and then we'll probably do like a 10, 20, 30, something like that this year as an A. we have a lot of revenue. We're almost profitable. The interesting thing about financial software is traders want to try everything. It's the greatest business in the world. You do have to have domain expertise, which nobody has. I think that's a lot of fun. Do you mean nobody in tech has? Yeah. Yeah.

1:00:03Our customers are brilliant. I think people who are in finance have financial expertise, or at least many of them believe they do. Obviously, to some extent, it's like a casino, and you're sort of arming them with tools. A lot of the tools haven't changed. I do have this thing I want to show you guys when I'm ready. I think we're going to call it like SIT or SITREP or Sitch. It's going to be like the situation monitoring suite. That's just like Bloomberg has these like fake systems that are pretty awful, but there's something where you can see the map and you can kind of see different things happening like earthquakes or whatever.

1:00:39But take that and actually make it really good. I think people would truly want to monitor situations. I think that's a new... It's a new hobby. I mean, I think it's a new sector. Yeah, it's a new category. And it's a prosumer category because people, I think, actually enjoy it so much that it's like camping where I think people will be willing to spend money. Oh, totally. So think about the revenue of Bloomberg with 8 billion, 10 billion, whatever. A satellite constellation, maybe it's 100 mil. And imagine if you're sitting at Citadel or Millennium or some hedge fund or Fidelity or something and you're literally tasking a satellite.

1:01:17Like, no, actually I want a satellite to point here. You know, and what margin difference would it make? You could probably pass through, you could probably be a creative diva to have your own constellations, like satellite constellation. And then things like drones or maybe, you know, Bloomberg right now, I think they have like ships and shipping routes and flights and like kind of fairly boring stuff. But you can make it 100 times better. So that'll be fun when it's ready. And again, this is all inspired. This whole sector started with the Bezos kind of. That's right. It really did. What's your take on prediction markets?

1:01:51Do you enjoy them? Do you think that they're valuable uses of information? Do you think it all just winds up being sports betting? Like where does all this go? And we will be trading. On what you say. Do you guys get paid? No, no, no. I haven't seen it. No, no, no. Yeah, I'm super. I have 1 ,000 X long on you saying biotech at some point during this show. All$3 on it. Yeah, exactly, exactly. No, no, but yeah. What do you think on prediction markets? Valuable, useful? Do you use them? Do you trade? Yeah, it's great. It's a great question. I mean, I think it's probably like social media, like the first social media companies, if you like go back into like that Steve Levy book about Facebook, which I think is just fantastic, like almost documenting, you know, every part of it.

1:02:41There were a couple of very random, even before Friendster, you know, a couple of these really proto social media companies and the sector exploded. I think that's where we're at. I don't think, I mean, I wish the companies the best of luck, except for Halsey, but the two companies, but I do think that, you know, markets have been around for, you know, hundreds of years, if not longer in ice and SIBO and these other companies are probably traded. They don't sit there and twirl their thumbs and say, gee, I, you know, I hope, I hope some creative new market comes along. They make financial products all the time that fail, you know they they innovate they add new things and then they actually take them down because they don't catch fire on wall street so if you want to be like that micro kind of uh you know micro markets sort of pioneer it's interesting but it's sort of still like you get into scale problems with like who actually arbitrates some of these bets obviously sports is easy and some of these other things but you know how much are you just financializing for financializing sake And is that sustainable when you actually need people with tons of money to fight the arbitrage gradient?

1:03:49You know, if you're betting. So in full disclosure, I've been running the spot on Polymarket that makes a market in the Bitcoin 15 minute markets, which are somewhat inefficient. And nobody needs to trade this, right? There's no actual benefit to trading this other than you can take two to the end bets. So you get around, I guess, 96 bets a day. And if you somehow, you know, the money velocity of getting bloggers right, you can compound insanely well. Of course, it's random. But, you know, people hope springs eternal that you could somehow get the rhythm of the market every 15 minutes. And you can, I think somebody posted on Polymarket that they took a very small amount of money and a very large amount of money.

1:04:32So, you know, it is it is kind of a I think it's a bit of a waste of ingenuity and engineering. And we're working on trying to sort of make a prediction market that might be more relevant, say, private companies. You know, right now, the binary outcomes of prediction markets have forced you into this box of like, will open AI go public this year? Well, SpaceX did a trillion valuation. Why can't it just be dollar for dollar? Why not like this? This is just a market that pays you and dollars per, you know, swap. And I'm sure that might be being worked on by other companies. But I think that's an economically rational thing.

1:05:06So the spirit is there. We should have our own markets. We should be able to deploy whatever we want. But actually make it make sense. And I think they're still trying to get that up. What is the state of tracking large, late-stage, private companies' valuations? We have a ticker at the bottom with MAG7, big tech companies with market caps. I would love to add Anthropic, OpenAI. I, but I saw some debate over, you know, how you actually get the data, how reliable it is. Yeah. Explain. I got into it. I got into it at Lonsdale. I unfollowed him. I unfollowed him for five minutes because of it. I was so mad.

1:05:42Brutal. Because. Worst five minutes of his life. Worst five minutes. I might have made a few days, but eventually I relented. Capitulated. I think, yeah, I think that what Lonzo is saying is right in that if he gets invited to a deal, per se, Ramp or whoever, you know, he was complaining that my Ramp number was too low. And, OK, fine. That's fine. But it's the same thing as insider info in the stock market. So you have insider info. Great. You know Ramp's going to trade up 50 percent higher than my number. He's like, your number's wrong. What's no different from me saying, well, you know, the NVIDIA number's wrong, but I know they're going to have a blowout quarter.

1:06:22Oh, yeah. Well, how am I supposed to know that? My market only reflects the people trading on it, not the news you know before everyone else knows it, which is in private, as you know, guys know very well, in private world, that's to be expected. Otherwise, you won't part with your money. But in the public world, you're not supposed to have that information edge. So he's so used to having the information edge, because I'm sure he gets invited to many, many great deals that, you know, he knows, oh, the price is about to change. And so SpaceX is a really funny example. On our platform, SpaceX price has tripled because, you know, Elon sort of dropped this bomb that like, oh, we're going to do a, you know, do a buyback at this crazy price.

1:07:04Obviously not crazy to him, but to the marketplace, they're like, what? WTF? It's like more than doubled in the last few months. um you know well and and a lot of the stuff late stage if somebody you know there's so much demand and there's so little supply that if somebody comes to you and they're like hey i've got spacex at one and a half trillion and somebody's like do i i'm really if i believe elon can get this thing out at one and a half trillion sure i'm gonna be flat over until they get out or whatever, but at least I'm in, right? So I don't know. People will start to be able to justify almost any price up until that point.

1:07:46Sure. And I know Matt Grimm is on this warpath against these Yeah, SPV hustlers. Yeah, yeah. So I mean, I feel like the only solution is we have to have a private, we have to be able to trade just like Polymarket, private companies. And there's a big battle brewing behind the scenes over this. We'll see what comes of it. Speaking of Elon, speaking of XAI, SpaceX, walk me through all of the internal secrets that you learned from the Infamous podcast. First of all, if your most junior employee went on a cinematic podcast for 90 minutes and just told all about Godel and everything, how would you react?

1:08:30How would you react? I think, you know, my first week in hedge funds, I thought this was a good idea. My sense for asking the boss just like kicked in at the last second. Wait, so you were going to call up Bloomberg and say, I got to tell you, we're working on some exciting stuff. I got to tell you about some of our trading strategies. This specific market is wildly mispriced and we're going to make a killing. We're printing. I was long the stock and the company told me, oh, we're doing this newspaper piece. Do you mind giving a quote for it since you're an investor? And I told our CFO, I said, I'm going to be doing this just so you know.

1:09:08He's like, no, you will not be doing that. OK, so what did you actually learn from the XAI deep dive? I think it's less about learnings and more about like just the gumption of like the way you send your questions. I can understand like, look, you're an engineer or you should be able to go on a podcast. Questions come up about XAI. You probably say, look, I can't really talk about it. You might drop a few things. But, you know, he was really like people asked the interviewer asked him, how is Optimus going to roll out? And he was like, let me tell you. That's Elon's job. You know, it's not yours.

1:09:41And so the real reason I kind of came after this guy in a not so, like kind of no bleep way is he had, I looked, I looked up this guy. I have a lot of friends at XA. I was like, who is this guy? And he had blocked me. I said, oh, no, no, no. This is a big thing. so I could amplify you know that you did this and again I mean I think the guy was fired you know at least the second he went on the show I doubt that I had any part to do with it but I definitely didn't mind amplifying the insane kind of decision he made and you know if you block me that's I mean Ro Khanna Ro Khanna no he came at me for absolutely no reason I said you know I replied to to beat like 80 million other people.

1:10:25Like, this is a bad move. And it wasn't a day out. I was like, look, I think that the problem is that it's not the billionaires. It's the billionaires and their teams, right? So if Peter leaves California, if Palmer leaves California, if Elon leaves California, it's not just them who leave. It's just that it works for them. And it's every florist that works for those guys. I'm trying to find florists. It's every house florist. Yeah, because you have the highly productive, high agency team members that are like, I want to be in the same building as my boss every single day because I don't want to go to war remotely, like you said.

1:11:03And if I'm going to be a serious player at this company, I need to have proximity to the person. I mean, yeah, the average billionaire family office has got to be 50 employees, their families. Like, it gets pretty big when you think about the footprint. I mean, just the companies they're running, right? I mean, just the companies they're running. There has to be a ton of new employment in Texas because of this, and I'm sure it will only continue. Our friend Jordan was just saying that he's like, Ro Khanna is just rage-baiting you guys, and you just completely played into it. And I think that's – I'm sure Ro will use this to set up his next play, but it doesn't take away from the fact that you're going to leave.

1:11:45I don't care if you're like if you're just using this as a as a stepping stone to like, you know, go to the national stage being like I stood up against the California billionaires that weren't playing their fair share. Well, like I live in California and I we still have to live with the consequences of like your political ambitions. Right. And pushing something that that's so obviously, you know, I said, like, what do you want California to become like an agrarian society, you know, where it's just like agriculture and vineyards and tourism? Like there's a lot of like every other state will be happy to take take these jobs and and, you know, host these companies.

1:12:24And I don't think we should just hand them all over. What's really crazy is my partner's at one of the big AI labs. And, you know, if she leaves California and then sells her stock, California will kind of come after you. And someone just posted this really almost like fascist letter that the California tax department sent them. I don't know if you guys saw this. Yeah, I saw it. I saw it. Yeah, yeah, it was insane. Basically, the letter was like, we want to understand the narrative of why you left California. It was like, spin us a yarn. Tell us a story on how you decided to, on how you had the audacity to leave this state.

1:13:01Yeah. I'm shocked by it. Yeah. Go ahead. On XAI, did the podcast make you more bullish on the strategy at XAI specifically? were there things that you thought were particularly exciting about what was, what was laid out or things that you thought were maybe misunderstood or mispriced? It's hard to tell from the junior engineer, you know, it's, I think like, you know, you get a window, which is great, but it's also, it's Elon show and like the risk of over indexing to something like that is like, okay, this guy saw it, he worked there. That's cool. But what is Elon really thinking? And I think that there's this new thing coming from the AI companies that nobody's expecting.

1:13:48And, you know, I wanted to see if he was going to drop sort of that. And I think all of the labs are gearing up to think about this. And it's sort of this new derivative of kind of the outcomes of AGI. And I think almost nobody in the market is talking about it. So it's going to be very interesting to see how the next, you know, leave a teaser there. See how the next... No teasers. Unpack it for us. Break it down. Five to ten years. I can't. You can't? Okay. The next five to ten years comes. But I think that the question is really what can AI really do? Yeah. And I think that we very much over-indexed to chatbots and LLMs.

1:14:29And a good new direction people have talked about endlessly is robotics. But imagine there's 50 new directions. You know, and I think that, you know, I hate to say something like this changes everything, but like there's a couple of clever ones, I think, coming that are just haven't been seen yet. And I think the smaller, like the non-Google, non-Meta, I'm sorry, non-Google, non-Microsoft type of players, you know, including X is kind of what I'm thinking about, are actually thinking about this very, very heavily. And macro heart, I guess macro heart is kind of, you know, an example of like, what is that really?

1:15:04And what's really going to be macro heart as we see, you know, macro heart be revealed more and more. I think there's a lot more to that than, you know, in some ways, I think macro heart is going to be bigger than XAI. It's all one entity, if I'm not mistaken, but macro heart is probably worth more than XAI. And I think that we just don't know what's coming out of that. We also know about Elon's gaming, AI gaming ambition, which I think, and I don't know if you guys have seen Martin Casado's coding. It's very cool stuff. And, you know, it's amazing that guy can invest and do that stuff. And I think one leads to the other.

1:15:40I sort of do the same thing for my end. I want to talk to you about really fast this idea of photonic computing, which I think is time is coming. Okay. Explain. Yeah, so photonic computing is what it sounds like. It's using light instead of electrons to compute. It's been around for 50 years. It's sort of like quantum's ugly stepsister. But I think it's actually the bell of the ball because there are very few companies working on this. You probably know all about it from Interconnects and Silicon Photonics as well. But recently in Nature and a few other preeminent journals, especially Chinese companies, have shown that you can do matmoles with light.

1:16:21And it's actually pretty remarkable. light goes to 100 terahertz, but it's more important than that. It actually is you can do 3D Mammals in their O of 1 complexity. So it's a pretty insane speed up versus GPU. GPUs are great, but they're almost too general, which is why TPUs have sort of been there. But if you throw out the need for, you know, lots of different instructions, you just need to do a Mammal, which, as you guys know, is 95, 90, 95 percent of what a GPU chip is doing all the time. A map mole is done with light very naturally through constructive and destructive interference. It's just nature does it.

1:16:55So I think this could be an interesting new space. And I'm spending a lot of time on it and, you know, very curious about, you know, could there be a chip? There's a couple of private companies. And I'm starting to dig around and, you know, see what's what here. I think most people stop short of actually making the full photonic computer because they need revenue. They need earnings. They need something. and Celestial AI got brought out by Marble. You know, there's some, or Marvell, there's some attempt to sort of push more and more of, like we used to use copper wires to transmit information.

1:17:29Now we use photonics and fiber. And more and more of the chip and more and more of the entire global kind of computer infrastructure is going photonic. And maybe the whole thing goes photonic someday. And that electricity is actually the odd man out. Photons don't have some properties that are perfect for, you know, computing, But others, like in the case of Matt Moles, which is now like most of the compute we need in the world, they're actually kind of perfect. So, you know, it'll be interesting to see if there's one private Chinese company called Lightelligence. And it sort of has a foot in the rest of the world.

1:18:01They're in Singapore, but they're also a foot in China. They started at MIT, so they're privately held. There's Light Matter, which it's not really clear that they're still pursuing kind of all optical photonics. And Microsoft's done a little bit in space, too. So I'm really interested in this as the potential next, like, as we think as Grok, the acquisition of Grok has sort of in the quantum insanity. This is kind of why I care about this in the first place as a biotech guy. You know, this is very weird for me to think about this. But the Grok acquisition,$20 billion, the quantum crap that's trading for$10 or$20 billion, people really want to think about what's next in compute.

1:18:37And, of course, the unconventional and their raise at$5 billion just for a new co. So it's all sort of clear that anything that can be an NVIDIA map model, in my opinion, could be worth trillions. What's the state of the quantum computing market? It seems like the companies have not sold off, but you went viral on a podcast sort of breaking down your overall thesis. Is the market sort of waking up to your way of seeing the timelines around quantum computing or the impact of quantum computing? or is there still more work to be done on your end? Yeah, so I think there's no more thinking that needs to be done.

1:19:17I think this is a war of attrition where both sides are right. And this is why I kind of started to look for other kinds of compute. And as you know, our great friend, Beth Jezos has his own kind of compute. So like the point is you've got a 5 trillion market cap. And if you bought a stock at 5 billion, which is not a small cap, that's a big company. Most of us would be very happy to exit at 5 billion. You can still get a thousand X, right? If it's the next NVIDIA. So if you even have a 0.1 % chance that one of these quantum stupid companies is the next NVIDIA, I mean, take a flyer to get a thousand X.

1:19:53So I feel like there kind of is a rational thinking there, even if it's impossible. What's the difference between 0.1 and 0.0? It's the kind of difference I think about, but most other people don't care. And so any chance to get – it's like the menu problem. What's on the menu for the average Joe versus what's on the menu for you and me? We might get in a grok round. We might get in an unconventional round. The average person can't, but they can buy$2 ,000 of Regetti or D-Wave or Robinhood, and maybe they'll get the same kind of, you know, self-manifesting in some ways, you know, beta. But I'm also curious if you guys want to talk at some point, expand this and talk about biotech more.

1:20:32Yeah, I would love. I mean, give us a lay of the land currently, how you're thinking about the year broadly, what's exciting, what's maybe being oversold about AI's potential impact in biotech versus being underappreciated. Yeah, I think that this that the NVIDIA, Lily, had announced this big partnership. And I've been thinking about drugs and AI more and more and more. And I feel like there is something to do here, but it's not the stuff that everyone thinks about, like discovery itself or small molecules or even proteins. I'm not sure AI is very useful. We have great forcing functions for that, whether they're physical chemists or software.

1:21:16You don't need a 10x or 100x improvement of that. The actual part of drug discovery, and I can say this as a very worn out drug company CEO, the biggest problem of drug companies is the people. And there's fantastic people that work at drug companies. It's just that you have 10, 15, 20 different departments that have to talk to each other, whether it's toxicology, preclinical, PK. There's all these little subsections. They all have to talk to each other and everything gets lost in the translation. Project management is supposed to sit on top of all that. It just often makes it worse. And so you have this like impossible.

1:21:53I mean, you talk to anybody that works in big pharma and they'll tell you how how ugly it is. And so if you had a machine doing it, it probably probably worked better. And I think that more and more companies are going to whether it's a drug company or law firm or anything else, there's going to be these sort of self-organizing, you know, more decision making gets pushed down. You know, sometimes I ask ChatGPT about, you know, what would be a better marketing strategy? Or how do I get more users? What should I do with it? It's like, oh, have you thought about, you know, this? And, you know, sometimes you'll get something surprising.

1:22:26So I think that's sort of the angle I play for drug companies. There's so much inertia. Oftentimes, a CEO will, like, kill a drug for no reason. He'll promote another drug for no reason. Data says no. the chief scientist has a personal vendetta with another company, so he really wants to push that drug. There's all these irrational decisions that get made over what's basically hundreds of millions of dollars to spend. And it's remarkably inefficient. So when Vivek came in to try to really upset the apple cart here, as I did too, we started to do the same thing. It was a little tricky because it's just not an industry that moves fast and breaks things.

1:23:04And for all the the occasional luck you get on a drug, there's a lot of that luck is distributed as well. So Vivek got a lot of bad luck sometimes with Roy Vant. He got a couple of things of good luck. It's done fine. But it's very hard to get the compounding effects that you get. So Lilly has somehow been able to do it because that drug is not going to go generic. But again, the problem is you make a great drug, you have a great hit, you have 10 years to make money and that's it. And so none of these AI companies would be worth anything if in 10 years they all had to give up their assets because there was a generic AI company.

1:23:38So I feel like drugs still are in this perilous situation. The whole industry is sort of cooked because there's fewer and fewer drug targets to go after. So we keep fighting over the smallest number of patients in these random diseases, sometimes like the skinniest slice of cancer where there's like 600 people have it. And there's like 12 drug companies that are developing a drug for it. And so there's overcapacity in the industry. I mean, the same thing happened in steel and some other places. And we arguably have too much going on. And the real fear right now is China. The China drug development capability has gotten so good that there's really no...

1:24:14Who was it that was saying like, yeah, I would hate, I would hate, we all want to cure cancer, but I would hate if China cured it before us. It's a wild quote. It's wild. On the AI impact in pharma, drug discovery. What is the lever that we should be watching between more frontier models, basically raising the IQ, making them smarter, more knowledgeable versus just speeding them up, getting the state of the art models running on Grok or Cerebris so that it's faster, so you can do more? Or is it more of like a adoption diffusion training? You need to just get the CEO to actually turn AI loose in the organization.

1:24:59It's like a change management problem. Well, I think you mentioned HR. And one thing I think XAI should do is take a little bit of macro heart and build an HR function that when employees say, hey, can I go on a podcast? It just says no. I think it'd just be an if statement. And it's like, yeah, if Elon, yes. If podcast mentioned, say no. Anyway, it's keys to success in AI and bio. I think that, I think pharma is a misunderstood business. So there was this sort of drug goat, this guy named Paul Janssen. So he created Janssen and Janssen was acquired by J &J, but Janssen himself just personally invented, you know, I think half a dozen to a dozen critical medicines, not like me too's or anything like that, but like he personally discovered, you know, I think the first antipsychotic fentanyl several like massive, massive.

1:25:57From the creator of fentanyl. We have It wasn't that back then, right? No, it was good. It's very useful. What are you, how are you processing the peptide boom? Yeah. Well, that whole thing is another story, but I want to get back to Paul. So there's statues in the sky et cetera. We need more Paul Janssen's, whether they're digital synthetic sort of intelligence or not. The drug industry is an idea industry. It is not an industrial process. It's some guy sitting down saying, you know what I think would stop cancer if we made an HDAC inhibitor? Well, actually people tried that. It doesn't work.

1:26:34And then somebody saying, I read a paper about this and actually maybe this would do it. It's the same thing as like getting a digital ilia who can just go and try a bunch of AI research ideas. You get the AI researcher in a box, turn that loose, and then that unfurls into progress. Certainly a challenge. Is that constrained by hardware, compute, scale? Are you still scaling PILD? Or is there an idea that will lead to that? It's a great question because in the old days, if you wanted to run computer experiments, you didn't need much cost. In fact, that was the traction. I used to sit there and write code.

1:27:11And now it's like, no, no, I need a cluster. And for drugs, you know, it's sort of has always been an expensive capital, you know, thing. But in the scheme of things, you know, it's cheaper than AI. And, you know, it's kind of like, you know, you could probably get some proof of concept in animals and people for, you know, less than you might imagine. And then from there, you know, you can develop a portfolio of drugs. So I think that, you know, getting the idea part of the drug discovery industry more sort of capable is one thing AI can help a lot with. And that downstream makes less expenses in clinical trial development and also the decision making.

1:27:54Somebody looking at the data and saying, look, this drug is a dog. Let's not do a phase two. Let's not do a phase three. And oftentimes the scientists and the companies have like, I don't know, Wall Street wants to see us in phase three. We can't really fall behind our rival. And they'll often push. I mean, I'd say up to a half of all drugs in clinical trials shouldn't be in clinical trials. And, you know, that's a lot of money wasted. That's a lot of patient time wasted. And a lot of it is pushed by other incentives and other things like that. So I think AI is a lot of, you know, potential for drug discovery.

1:28:26But every time I talk to somebody that wants to talk about AI and drug discovery, they've never been at a drug company. They've never actually done it. And when you do it, you realize it's much more about who's Paul Janssen discovering the next, you know, bunch of hits and who was lucky and got a one hit wonder. And then who just sucks in general, like Pfizer or something. And so you sort of have to like generate those ideas. It's almost like an investor in hedge funds. It's interesting because I feel like I can't remember who exactly, but there was somebody on the show that was saying we don't need it's not that we don't need more ideas.

1:28:56is we need a cost reduction across the rest, everything from that point. But most of the cost is clinical trials. So what you were saying is like, if you have these 15 departments and if you can somehow make it so that, you know, you can have, is it a head count? Like how much is like a head count reduction actually going to impact the overall cost? I think it's all clinical trials. Like, so once you've invented a drug, say that costs 10 to 50 million, the next four or 500 million is going to find a bunch of people to try it and compare it to placebo because you have to monitor those people so carefully and so closely that it ends up being this massive cost across, you know, tons of clinical trials.

1:29:37That's the hard part. If you could just take shots in goal for 50 million, 20 million left and right, you know, you'd have a really productive drug industry. But unfortunately, getting 10 ,000 people to take a new diabetes drug is, it's pulling teeth and you have to like really, incentivize doctors and patients, meaning you have to pay them. And once they do that, you have to pay the company that collects the data, you have to pay the company that puts it together, sends it to FPA. All that stuff is just massive paperwork. And a lot of it's done. How is China's process for running clinical trials different than ours?

1:30:13Very similar, but I think the wages that they pay the people to collect all the stuff is less. and I think that they sort of have it a little more streamlined. I mean, it's amazing what they can do. So one of the things that confused me about China the first time I looked at it in biotech was the average mid-sized American biotech is like 100 people, 200 people, maybe 300. But in China, it's like 6 ,000 people work. Same market. Wow, what are they doing? It's like, no, we have like 800 chemists sitting here. What the hell are they doing? Is that just because of the structure in America? we have more like labs and consultants and we still employ 6 ,000 people, but it's all at secondary and third tier supply chain companies?

1:30:55Maybe. I think it might be more capital markets where it's like they raise the same amount we raise in the Hong Kong markets. They raise 300 million. What is 300 million volume in China? A lot more people. Why is there 800 chemists? 800 chemists. Why is there about 40 chemists in San Diego? Yeah, yeah, yeah. That's crazy. What is your take on the peptides boom? Yeah. Yeah. Peptides are, you know, every time somebody asks me about it, I have to laugh because peptide is defined as a small protein. And so what you're really asking is what is my take on the unapproved self-medication boom? And, you know, when you phrase it that way, it sounds insane.

1:31:35No, no, no, no. And that's, that's like, to me, to me, I, I, uh, I, I tried a peptide cycle probably five years ago and I was like, okay, this is like cool. And I noticed like a moderate effect, but I also acknowledge that like, this is, I'm like administering, administering, like effectively an experiment, like, uh, I don't know, experimental drug is not the right framing, but it's, it's not like the same as like walking into Erwan and getting some vitamin D and, you know, having, having a daily supplement. It's like you're like inject, you're, you're, you're, you're doing an injection. The source matters a ton and it is far closer to, you know, taking a round of anabolic steroids than it is, you know, uh, adding, adding some creatine to your smoothie.

1:32:22I just think it's nuts. I mean, if you think it's a good idea to do your own research on medicine, self-medicate and on uncontrolled substances that are manufactured, God knows where that have not been clinical trials. And I mean, I'm, I'm as libertarian as it gets, but I just don't understand why, what the impetus is to like take foreign substances without like some emergency need. And it just, it's, it boggles my mind because, you know, people, there's some psychological phenomenon where people need to feel in control or they need something in their life that they feel like they're taking a risk on or, you know, it's nuts.

1:32:58I mean, why, Why would you do it? And explain why there probably will never be clinical trials for these drugs. Right. There's this edge case where, you know, you have a cosmeceutical, in essence, or kind of like something that doesn't quite fit what's a disease to the FDA. So a good example is premature ejaculation. So the FDA does not consider that an illness. European drug regulators do. And so you can get approved for impotence in America because it's actually sort of a medical problem where it's like, oh, if I can't copulate and produce, it's a big medical kind of emergency in a sense. Premature ejaculation is sort of like, well, you have it.

1:33:38This is what it is. And so the FDA has never been willing to approve. A company a long time ago tried to get an old J &J drug called Apoxitine approved for premature ejaculation. And the FDA basically said there's no set of circumstances where we will approve this. And, you know, what's weird about that is, OK, well, what if you actually do supplements? And it's a severe psychological detriment. As much as we joke about it, people joked about erectile dysfunction a long time ago. And it's it's a serious illness to me. So you sort of live in this twilight of like, well, I know what drug works. This drug to poxotene actually does work.

1:34:10But I also know it'll never be FDA approved. So what do I do? And in some ways, you know, you're sort of like I'm taking an unapproved, technically illegal substance and I'm buying it off some shady, you know, third party market. So eventually I think we need to find a way, just like, almost like what we just talked about for prediction markets. You have these third markets where it's like, yeah, you can bet on, you know, what Trump's going to say in a speech and it's not registered stock exchange, but it's also more than just you and me making a solid bet. And it's the same thing with, with this, where it's like, dipoxidine is not a peptide, but you know, I, dipoxidine is medically useful for that disease.

1:34:45If you call it a disease, FDA won't say no. So it's not like I'm taking some random goop. It's a real medicine. But it's in this middle of like, what do you do with this thing? And so much of what we do as investors or whatever is in that middle. You could sort of define Uber as in that weird middle as well. So I think eventually we need to address like, it's not as insane as maybe I'm making it out to be. I think some of it is. But eventually you need to address this like, how do we do pet things right? Well, even like BPC 157, there's no incentive to spend the tens of millions of dollars on trials.

1:35:22You would need some type of trade group or something in order to fund that because if one company did it, then every company would get the benefit of it. So why would one company spend the tens of millions of dollars? We haven't asked for that. In drug development, we haven't asked for that. We can take that drug, twist it, change it, figure out, make it better, you know, and then you get a new patent and then you have an incentive. So there is sort of a way to do that. And that's not that's not uncommon. I mean, you know, trolling around PubMed looking for, oh, this drug actually does something.

1:35:54I'll make, you know, take it to my boss at Merck. Boss says no patent. Come back the next day and say, here's the same drug with the nitrogen atom coming out of here. It's going to work the same way. And he says, all right. maybe we should run with that. Run it. Well, thank you so much for coming on the show. So much fun. Come back soon. Yeah, come back on again soon. I have a bunch more questions I want to ask about Godel and everything, but good to see you and happy new year. Congrats on the progress. And congrats on becoming a father. Yeah, fatherhood. Massive. Massive news. Great stuff. We'll talk to you soon.

1:36:26Have a good one. Goodbye. Phantom Cash. Fund your wallet without exchanges or middlemen and spend with the Phantom card. Our next guest is Bradley Tusk, the founder and CEO of Tusk Ventures. You might know him from his work with Uber. Thank you so much for waiting. Welcome to the show. How are you doing? Good. Thanks for having me, guys. Fantastic. I am upset that it took us over 1 ,000 guests to go on. Yeah. I know. I have to say. I know. So you just hired a friend of mine. I got him. Jackson Fortis. Do you guys know him? We love Jackson. So I literally had the like call with him a few weeks ago where he's like, what do I do with my life?

1:37:05And I said, take the job. Amazing. Those assholes have never had me on. And so those two things came together. I'm glad we could do it. Yeah, but you're getting an incredible person in that hire. So we could do a really nice job there. That's great news. Tell us about the bookstore. I heard that you have a bookstore. I do. So first of all, anyone is looking to lose money. I highly recommend owning a bookstore. So, you know, I've written a couple of books. I had started this thing called the Gotham Book Prize in 2020. In fact, we announced the nominees for this year today. And I'd always thought like one day it'd be fun to own a bookstore, but the kind of thing you would just do later in life.

1:37:45And then during COVID, New York City just got whacked by everything. And we lost, you know, half a million jobs. And my thought was, look, if you're going to do something nice for the city from a retail perspective, because I knew I wasn't ever going to make money on it. Like, now's the time, right? Don't wait until you're 70 or whatever it is. And so we started looking for space, and we found one on Orchard Street on the Lower East Side, if anyone knows where that is. And it's a great spot. So it's about 3 ,500 square feet. It's an indie bookstore. It's a podcast studio that is free for anyone to use.

1:38:20You just go on our website, and you can sign up and use it. That's awesome. If you guys are over in New York, you're welcome to the studio. We've got an event space. So we built an amphitheater that seats about 80. So we have authors or different types of events like open mic teenage poetry or whatever, pretty much something every night, and a cafe. And the store has a funny name. And the reason why is so when I leased the space, I texted my dad and said, didn't the store that my grandfather started when they came to this country in 1950, wasn't it right around here? Because that store was closed before I was born, but I knew it was right around there.

1:38:56And he's like, yeah, it was one block over on Allen Street. And I said, remind me of the name. And he said, P &T Knitwear, because if you were an uneducated Jew that survived World War II, the one job you could get in the US, especially New York, was in the garment business. And so my grandfather started a sweater company. And then my dad wrote P &T Knitwear, but you cannot name a bookstore P &T Knitwear. So of course, our store is called P &T Knitwear. So that's our store. And yeah, other than the fact that nobody buys books, it is a really fun business to own and run and a fun place to hang out.

1:39:35So anyone who is watching, if you happen to be in New York, you like bookstores, check us out. That's very cool. What's it going to take to get it profitable? I don't think it's possible. You know, there's actually not one. So because I've looked at like, okay, let's say I ran it aggressively. We charge for the studio. We charge for the event space. we didn't give the employees health care, all this, you know, other stuff, we would still lose money. So I don't quite understand how somebody actually makes a living running a bookstore. I'm sure Manhattan's sort of the worst place to do it, but still, but even like in LA, like my friend Zibby has a great bookstore in Santa Monica, but the same thing, she's, you know, lucky to, you know, have other money and she, part of the way she uses it is to have this indie bookstore.

1:40:18So it does seem like sadly, you know, bookstore ownership these days is somewhat equated with philanthropy. Yeah. Yeah. Oh, well, yeah. Let's get back to, let's get back to business. I want to talk about specifically these businesses that you've been involved with, where you've been so helpful to founders on the local and state regulation level, how companies position themselves. It feels like every company right now is focused on what's happening in DC, the federal level. How do you think about what's different when you're working with a company at the local level? What's the playbook? What works if you have a problem and you need to change local laws?

1:41:06What are you recommending to new founders? Yes. I mean, the first thing is, look, human beings naturally, or Americans at least, when you say the words government or politics, your mind kind of instinctively goes to DC. So you think about the Capitol, the White House, Trump, whatever it But the reality is most regulation, including most tech regulation, does not happen at the federal level. There's some crypto, biotech. But by and large, most things that touch a consumer in some way, directly or indirectly, are regulated either by state government or local government. And that's actually a good thing.

1:41:41So what everyone thinks is, oh, I'd rather solve my problem at the federal level. It's a one-shot stopping, one-shot whatever. I got my thing done. I'm set. But, you know, the phrase and act of Congress is literally synonymous with a miracle, right? You can't base investment strategies on miracles. Whereas in states, and you guys had my friends, you know, Adam and Matt from Doctronic on a few weeks ago. We worked with them to make Utah the first state to allow prescription via AI. In states, you can get things done. And we don't necessarily win in every state every time. but by 50 bites at the apple, we can figure out this makes sense for a red state strategy, blue state, whatever it might be.

1:42:21It might not matter in that way. And because states have sort of six-month legislative sessions in which typically a couple of hundred bills on average pass, you have a reasonable chance of getting your same thing done if you run a good campaign. And then once you establish precedent for it in one state, you have a stronger argument to then go forward in others. And so when we decide what we want to work on and what we want to invest in, we highly prioritize companies that can get their stuff done at the state and local level, because I don't invest in stuff where I don't think that we can move the needle for the company itself.

1:42:56And I think our odds of doing that are so much better outside of Washington. How confident were you that you guys could get this, a deal like the one you got done in Utah with Doctronic. I was a little surprised. Yeah, yeah, because I was... Yeah, we kept having to be like, really? That's what happened? Really? You can do this? I know. Even we, we literally are the people who do this for a living, right? I think I'm the only one that does this. And even we were like, wow, that one was pretty cool. But, you know, there are states that are very thoughtful around AI regulation. And there are a couple of states, Utah, Texas, Arizona, Kentucky, that have created regulatory sandboxes.

1:43:39You guys know what that is? A little bit. So it's effectively like a notion of saying, okay, you know what? There's this thing that we know is really important. We know it's going to take experimentation. And rather than subjecting it to the political process, let's put together a team of experts who can, at least on a provisional basis, say we're going to try out these ideas in our state. And as long as it's working the way we thought, it'll keep going. And those are four states that have chosen to do that. And so those were the obvious places for us to start. Utah is just a really well-run state.

1:44:10Governor Cox is an excellent governor. And so we and they're not a typical state for a lot of reasons. And also they're becoming a very, very tech forward and friendly state because you see migration, especially from California, let alone before the billionaires referendum tax thing happens. Utah is becoming, you know, Silicon Slopes are calling it. So, yeah, we're going to Texas next, which is a giant market, obviously the second biggest in the country. And, you know, once we start showing that it works, you know, even if you think this is the very first tech company that I worked with was Uber, sort of doing that early 2011.

1:44:49And the thing that that Travis Kalanick, who was the founder and CEO of Uber time, understood and he was right about is just get the genie out of the bottle. I mean, Travis didn't pretend to understand politics, but what he did know was that he had a product that was meaningfully better than the status quo. And if you guys think back to the first couple of times you used Uber, it was magic, right? You kind of couldn't believe it. You just pressed a button and this nice car showed up and here they were and you have to like hail it down or deal with some crazy cab company or anything else. And we knew that once people saw that it worked, they would fight for it.

1:45:24And the arguments against it would get weaker and weaker. and we were able to effectively mobilize our customer base. Well, you have the two, it's both participants fighting for it too. Correct. You have the drivers and the users, which is the most powerful dynamic. Right, right. And so when you can do that, but you know, it's funny. There is a whole question as to when do you have a product or a service that people will fight for? And sometimes it's, you might have a really good company that's going to be a good investment, but it doesn't sort of excite people in that way. But what we found is FanDuel worked really well.

1:46:02Bird, Scooters worked really well. Ease, which is a cannabis delivery, worked really well. Crypto, that works really well for us. So, you know, you have to think about if you're a founder, you're obviously passionate about it because otherwise you wouldn't have founded the company in the first place and made all the sacrifices that you've made. But you have to be somewhat objective saying, okay, do regular people find this thing that I am doing truly critical and fascinating. You said FanDuel. So are there people out there that will stand up and protest their state government for banning sports betting?

1:46:36We did that. So in 2016, the New York Times published an article effectively accusing employees at FanDuel and DraftKings of conducting a form of insider trading where, according to the Times, they were giving each other information to then put bets on the others' platform. It turned out to actually not be true. But putting that aside, it unleashed this total shitstorm where a couple of dozen states in the course of a few days issued cease and desist orders to us saying you can no longer operate. And we had to all of a sudden go all over America and either pass bills to legalize daily fantasy sports betting or at least get them to understand that we were complying with the existing laws.

1:47:17And the way that we did that is we went to our customers and said, listen, if you like this product, we need you to fight for it. And every time they were on the app, we knew where they were located. And we'd say, hey, click here to tell your local state senator, state rep, whoever it might be, that you want FanDuel to stick around. And enough people did it that, you know, it shifted the political inputs. If you guys just take a half a step back, in politics, if there's one thing to take away from this interview, it's that every policy output is the result of a political input. Every politician makes every decision solely based on the next election and nothing else.

1:47:56And there are some exceptions. I was Mike Bloomberg's campaign manager when I was mayor of New York. Not for the good of the world? No. Not in the slightest. Right? Mike Bloomberg. That's pretty much it. So flag-pilling. So as a result, that's incredibly depressing. admitted. But it's also kind of a useful place to work from, because then you have to understand, what can I do to make said politician feel like doing what I want will either help them win the next election, or if they don't do what I want... So the tech industry should tell Ro Khanna, we won't trash you in the next election cycle if you don't put our industry in the dirt.

1:48:33Yeah. Or quite frankly, it's funny, just by Larry Page and others acting in their own self-interest, and already saying, I'm moving to Florida, the billionaire's referendum is already on the ropes because of that. So sure, if you think about why, let's say, the place like California, there are unions such as SEIU, the teachers, have so much power in Sacramento, it's because they have so much money that they can threaten any given politician and say, if you don't do what we want, we're going to spend a million dollars, five million dollars, whatever number they want against you in the next election, and everybody backs down because winning the next election is more important to them than any given policy or any given issue.

1:49:16So not every tech startup obviously has the kind of money that a union might have to throw around. So you have to be more creative about it. And that's where your customers can be really useful advocates. Interesting. Very interesting. What's your current outlook on the venture landscape broadly? We've been talking a lot about the K-shaped dynamic. There's a whole bunch of headlines around lower total venture funding, more SPV activity. What are you seeing? Yeah, a few things. So I kind of have it from both a deal-by-deal standpoint and then kind of a macro fund model standpoint. So from a deal-by-deal standpoint, I think we're in this very bifurcated world.

1:49:52So I started investing outside capital on my first fund in 2016. In fact, FanDuel was the first deal I ever did. And at that moment, there was a period of irrational superiors, right? Everything was incredible and everybody threw tons of money at everything and every startup was going to be unicorn or decacorn or whatever. Then 2021 came, everything fell off the cliff and all of a sudden things like unit economics and margins and burn and all that stuff started to matter. And then I came along and then so for AI companies, we're back to the 2016 days where people are just having valuations to me that sometimes make no sense whatsoever.

1:50:30You know, I've seen seed companies, you know, with valuations in the hundreds of millions of dollars with no product or certainly no revenue. That's crazy. And yet some of the discipline that you saw post 2021 crash is still being applied to non-AI companies. And while I do think that overall, if you had an ETF of private AI startups, it would do well over time. That doesn't mean that any individual valuation is right simply because the letters AI are attached to it. some way from a fund model standpoint i really believe that there's effectively only two types of funds that could work if you were to start now one would be a fund with billions of dollars of aum simply because at that point the management fees are sufficient that the gps can make a living from that alone um as long as they can raise the next fund you know it doesn't really matter or a sub hundred million dollar fund or even sub 50 you can get away with it where the hurdle is relatively low, you can get into carry pretty quickly.

1:51:31You're not making much on the management fees. So that's certainly you've either got to have other resources or really be super scrappy. If you wanted to start a fund with the lowest chances of success, what would it be? For me, it would be a$200 million Series A only fund. Sure. Yeah, a generous, generous with nothing differentiating. Yeah. Yeah. Even that sub hundred million still only works if you have something truly differentiating. Right. So for me, because we're the fund that does the regulatory stuff for better or worse, we're the people you call if that's the if that's your issue. And if you don't have.

1:52:09But then how do you what's your framework? Because because there's times where I've invested in a company where I'm like, OK, this this founding team is fantastic and they don't understand. but they don't, they're, they're good at all this stuff, but they don't understand marketing. I'm very good at marketing. I can invest and I can help. The problem is I'm, I'm going to be in the room with them like less than 1 % of the time, right? Like I'll do a handful of calls, I'll make intros, I'll help however I can, but I'm not actually on the team. And so I've ended up getting, I've, yeah, I've ended up getting burned because I'm just like, I, I can't, you know, 1 % of my time is not enough to take them from bad at marketing to even good.

1:52:50Right. Right. So for us, the first thing is we ask two simple questions. Is there a gating regulatory issue or opportunity that if it were solved could really drive growth and valuation? And then if so, can we solve it? And that's what left us investing companies like FanDuel, Lemonade, Bird, Rogue, Coinbase, Circle, Latch, so on. And then the other part, though, is take Doctronic as an example. We are their GRPR department. We effectively become the in-house group for a company until they get to a point where it's like, okay, we are now big enough that we probably need an in-house team, in which case like Roe is a really good example where we started working with Roe back at the seed.

1:53:35You know, it's funny with Dottronic because it's prescription via AI when it was rogue, which was really not that long ago. The question was, is tech, is prescription via text legal? Nobody knew, right? Because they had no till Z. No one had ever tried to do it before. And he saw something that was incredible. And our job was to legalize us everywhere. And we worked on that and did it. And then eventually they grew to a point where we helped them, you know, put together their kind of in-house people. And then we still talk to them all the time. We jump in on stuff, whether it's GLP-1 compounding rules or when they want to be a COVID test provider or whatever it was.

1:54:12But for those early rounds, we really are part of the team. And that's why we take equity for our work because, you know, we're lucky to be in a position where I don't need to take cash usually because I have the resources to sort of fund it myself. But, you know, we are there day to day. Talk to me about mobile voting. We've heard about it for a long time. Do you think it's ever going to happen? What are you doing in this space? Why is it important? Okay, explain. So, all right, let's start. It's important because in a world of gerrymandering, and I'll make the assumption that the viewers know what that is, the only election that matters is the primary.

1:54:49So the New York Times did a study of the 2024 general election. Only 8 % of congressional races were decided by five points or less, And only 7 % of state legislative races were decided by five points or less, which means well over 90 % of elections are really determined in the primary. And the problem is primary turnout is incredibly low. It's usually about 10%. And so who are those 10 %? They're the extremes, the far left or the far right, or they're different special interests, whether it's the NEA or the NRA, either side, that effectively dictate what happens. And if you are a politician and all you want to do is stay in office, then you cater to those groups at the exclusion of everyone else.

1:55:29But because those groups are purists, they don't want compromise. They don't want consensus. They just want to sort of hold true to their principles no matter what the cost is. And so nothing gets done. And the simplest way to solve that is to meaningfully increase primary turnout. But the only way to do that is to meet the voters where they are. Every single one of us has a supercomputer in our pocket. And if you can allow people to vote in at least state and local elections on their phones, you can get primary turnout from 10 to, say, 40. And now all of a sudden, if in order to get reelected, you have to represent 40 percent of your constituents instead of 10, that 40 percent is much more moderate and they want things to get done.

1:56:12So if that means working with the other side to find a reasonable compromise on immigration or guns or health care or climate or whatever it is, you can do so. It's not that government doesn't know what the answers are. It's that embracing them in peril is re-election, and they won't do that. And so I started something called the Mobile Voting Project in 2017. And the first thing we did was test the hypothesis that I just laid out. And so we funded elections out of my foundation that's totally philanthropic in seven different states where either people with disabilities or deployed military were able to vote in real elections on their phones.

1:56:47And the answer was, of course, as you guys know, when you make something a lot easier and you put it on people's phones, they do it. So that was true. Then the next question became, OK, but is the tech really good enough to do this at scale? It was totally fine for 2 ,000 soldiers from West Virginia, 3 ,000 blind people in Oregon. But if we're trying to get every district from 10 to 40 and the tech that was on the market at the time wasn't good enough for that. So we spent five years building our own mobile voting technology that is end-to-end encrypted, end-to-end verified, has biometric screening, multi-factor authentication, it's air-gapped, and all of the code is open source, which means it's on GitHub right now.

1:57:24So it's totally transparent, totally audible. And the next part is then, okay, getting government to say, yes, you can actually do it. So that's about to happen for the first time in April. Anchorage, Alaska is making mobile voting accessible to all of their voters. Hit the gong for mobile voting. Overnight. Thank you. And we are running legislation now in five states, Minnesota, Colorado, Vermont, New Jersey, Maryland. That would allow cities in those states to start offering mobile voting as an option for local elections. citizens i can imagine are super excited about this who's not excited really well yeah yeah because you're just like oh no i love going to the poll i i love i love taking you know an hour out of my day to go do this thing that right uh so there's the people there are a few smaller opponents and there's in reality the big one so the smaller ones are there are some cryptographers who have said it's impossible can never be done something can go wrong and they're not wrong something could go wrong but things go wrong at in-person pal places all the time things go wrong with mail-in voting all the time and the risk of inaction is where we are today and this is not sustainable for 24 years um and there are groups like this one called verified voting that are kind of the anti-vaxxers of voting so they're just like in paper on paper in person nothing else You know, they're the crazies, just like the anti-max people, in my view, are a little crazy.

1:59:00But the real opponents, at the end of the day, are the status quo. Because if you are an ideologue or a special interest and primary turnout is 10 percent, you have a lot of power because your threats go a long way. Well, yeah, and you look at who's showing who's actually turning out for the primaries and you can concentrate budget on influencing that subgroup, which is much better than having to appeal to an entire region or a huge age range. Right, exactly. So that's what we're really going to have to fight through. But, you know, if you think about every major right in this country that's ever been won, Women's Right to Vote, Civil Rights Act, Voting Rights Act, Same-Sex Marriage, America's Disabilities Act, whatever it is, status quo didn't want to do any of those things, right?

1:59:46Of enough people stood up loud enough and long enough and demanded their rights, and eventually they won, and it's not easy, and it takes a while and it's expensive, but that's what we have to do here too. And we're trying. Really quick. I know we're mostly out of time. How do you think, what do you think the dialogue will be around AI in the midterm elections? Negative. I think it's going to be pretty bad. And is this the kind of thing that certain politicians are going to make their entire brand anti-AI? Well, so I think what you're going to see we should see if we can get CalShare, Polymark, or someone to take a set of line on this, is the word that will be cited the most in the midterms, by the way, it's not just Congress, there's 36 governors, there's state, most states legislatures, is affordability, right?

2:00:38So that's the word, right? You heard Trump say it, you heard Mondami here in New York say it, everyone's going to say it. So then the question is, where does AI fit into the affordability puzzle? And it fits in a couple ways. One is job loss. I do believe that some of the jobs data that we saw in 2020 25 reflected layoffs, typically kind of bigger tech companies that are the hurliest adopters of AI and new technology, using it, realizing this works pretty well. I don't necessarily need as many people to perform any given function. So you're going to have layoffs. The second is energy prices. So right now, in the way that data centers were conceived by the hyperscalers was, you know, we build these things and we would consume a lot of energy and that would be from the grid.

2:01:20And the problem is everyone else who's on that same grid then has to pay the price for that. And so there are a lot of states where AI data centers have already come online, where people are paying 30, 40, 50 percent more in their energy bills every month. The problem is no consumer on either side of the aisle is interested in having to spend more money out of their pocket to help make Sam Altman or Johnson Wong a trillionaire. Right. So, yeah. And the thing I push back when people were like, I want Uber in my area around where I live. But nobody's like, if I don't get a data center in my backyard, I can't use chat.

2:01:53ChatGPT is going to work no matter where the data center is. It's going to work. And by the way, the data centers don't even create that many jobs. So there's not even like you can mobilize that for it. And you see now a bunch of different states on both sides of the aisle start proposing legislation that would find ways to curtail the ability of the data centers to impose those costs and externalities on consumers, whether it's around having to use different types of energy or alternate forms of compute or water or whatever it might be. And so AI is going to get blamed, not wrongly in some ways, for both job loss and higher energy prices, along with things like mental health chatbots having weird problems with teenagers and things like that.

2:02:33And so I think it's going to be a huge attack vessel for challengers against incumbents, you know, in pretty much every election this year. And I think that when we in the tech world just think everyone loves AI because we love AI, that's not true. And if that is the mindset you take, you're going to lose. Yeah. No, that makes a ton of sense. Agreed. Yeah. It's going to be a really important story to follow this year. We'd love to have you back on the show to talk about it more. Yeah, but same. Anytime. Love talking about this stuff. So anytime it works for you guys, just let me know. Thank you so much for taking the time.

2:03:08Expect emails in the morning when the story's great. You got to come on. You got it. I'm a real expert. And thanks for pushing Jackson over the line. No, it's great. He's going to love it. Yeah. Have a great rest of your day. All right, guys. Thank you. And Barnes & Noble is opening 60 stores. They're going to IPO. So maybe there's a boom. Somebody sent us. Maybe, yeah, maybe. So I think just don't limit yourself, right? Make it up with scale, right? Open 50 more stores. You'll make it up with volume. Yeah, for sure. There we go. There we go. I just have to make it. I need another Uber to be able to afford that.

2:03:43Thank you so much. Thank you so much for having me. I appreciate it. Cheers. Goodbye. Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in-store, on mobile, on social, on marketplaces, and now with AI agents. And we have a very special segment on TBPN today. We have Lucas Zinger from Divergent in the Ultradome. He's actually outside the Ultradome. So we're going to take a walk and we're going to do a walk and talk. I think this is the first time we've ever done this. So I should be able to click over to the lav mic. Do you have the lav mic?

2:04:20Yeah, I think it's live. Let's go see what's waiting for us outside of the Ultradome. We're getting audio from outside already. I think we have audio outside. Let's go and see. Okay, we are ready. Wow, look at this. Thank you so much for coming. Look at this walk us through what we're looking at it reintroduce yourself for those who might not remember I'm Lucas Zinger I'm the CEO of divergent and Zinger vehicles and you're looking at our first car the 21c. Wow You had to bring it you had to bring it in green too. Thank you so much for bringing this. Match outfit here for you. This is actually remarkable.

2:04:57I've watched the full video I watched Doug DeMuro's review Okay, and seeing it in person the scale is remarkable. Yes, it is a wonder. It's a spacecraft for the road It's a Landjet. It's a hell of a car to take in traffic from Torrance to here. Wait, did you drive this on the street here? Yes. I was assuming you were in a trailer. No, no, we drove it. We drove it. An hour in traffic on the way over. I cannot believe you drove here. Okay, full carbon. Full carbon body. Before we go into the details, tell us the story of your journey with cars. Why did you want to build this car? When did this project start?

2:05:31This doesn't look like it was built in one day, but how long have you been working on this? This is a 10-year build a company from a vision moment. So we started as a family business. My father had the vision for Divergent. I joined him soon thereafter. And we have a business that has changed the way that we engineer and manufacture fundamentally. So when we talked last time, we talked about the defense application, the manufacturing application. That company is scaling right now. But Zinger was the first product within Divergent. So when we said we've got new design software, new additive manufacturing technology, new robotics, how are we going to prove this, validate this?

2:06:11We turned to making our own car. Then that evolved into making our own car company. So Zinger Vehicle is about five years old. Five years old. Within 10 years of what is the divergent journey. And now it's the world's fastest hyper car. It's American made. It's got five plus international track records and it's road legal and most excitingly There's over 16 of these in the wild now. They've been delivered Los Angeles, Miami, London All the usual suspects This is remarkable And this car it represents fully that 10-year journey. Yeah I mean, Joy was just making the joke with me today that you could build the ultimate American-made hypercar garage for remarkably cheap, compared to an Italian hypercar collection or even a British hypercar.

2:07:06You've just raised the stakes significantly with this, I think. This is remarkable. So what has the testing been? Like you said, five records, who's driving these cars, what tracks, what are the records that you're going for? What has stood out in the journey? So Joel Miller has been our test pilot and he's an exceptional driver, factory driver for Ferrari, great coach. And we nailed five track records in five days, which has never been done before. We made a documentary of it and it was all about the street usability of the car and the performance envelope it has. So instead of showing up to a...

2:07:41So five in five days, what does that actually look like? You're getting a record, immediately packing it in a trailer, driving it to the next track? Driving. Driving it to the next track. So that was the epic journey and we made a documentary called a California Gold Rush Okay, and we drove from NorCal through SoCal and we had every track on the way and every day We'd set the track record every night We get in the car and we drive a couple hundred miles to the next track get in late Get some sleep get up early the next morning and do another record So it was as much of a human feat as a vehicle feat But what it was meant to show was these hyper cars can be driven on the road This one can it was built as a road car not only a track car And it's got so much performance that quite literally you can show up at the track on used tires Instead of track record and not need a pro team next to you supporting it.

2:08:28That's amazing. Talk about the decisions in usability. I see three pedals in there Between, you know, the Carrera GT Super Raw, no supports, extremely dangerous to some of the more point and shoot. Where did you want to land? Where did this car land? So ultimately this was about creating a road car that was otherworldly in the way that it drove on the street, but could beat the best on the track. And that's a hell of a design brief because usually those two worlds don't marry well together. We also wanted to create a vehicle that was engaging as part of that otherworldly experience, right? You wanted something that wasn't numb at low speed, that was very, very lively on the road, but ultimately could shave 10 seconds off the thermal track record when you go 10 out of 10.

2:09:17This has been the thermal. Thermal, yeah. 10 seconds off the track record at thermal. Do you know who had the previous record? I won't say. That's too big of a margin to say. But we've had an epic back and forth at Laguna Seca. You know, because I've been in a Valkyrie at thermal with our friend. With our friend, yeah. Who I won't name them. That's a great car as well. But I was in the Valkyrie at thermal. almost every single corner I thought that I was gonna die. So 10 seconds faster than that? 10 seconds faster than that is insane. Sounds brutal. Yeah, you're pulling major G's, the cornering, the acceleration, but back to your question, this car marries those two worlds, track driving with road driving, it's a one design brief.

2:09:59You got the center seating, that's of course intentional, that's based on the aero package, that allows for less frontal surface area. It also gets you centered in the car, which is the best track driving position you can have, right? Your laps feel the same as your rights. I've always wondered why more manufacturers don't go with the center seating. I mean, it's iconic from the McLaren F1. And then maybe it's just a practicality thing. Yeah. Yeah. The back seat, right? It looks a little tight in there. Your feet go outside of the front seat. So you actually have much more room than you think.

2:10:29You don't need to get your knees up behind that front seat. And when you're in there, you're packaged pilot, co-pilot. And it's a really dynamic driving experience. I mean, our customers that drive with their partners, they actually love the layout. And in terms of hybrid, strong hybrid, ground up developed in the U.S. as well. So we've got a V8 that we took all the way through California Carb to road homologate. It's about 750 horsepower, twin turbo, 2.88 liter. We have another spec that takes that up to 850. And then you get another 500 horsepower out of the front wheels. So you're doing 1250 to 1350 total horsepower, all-wheel drive, seven-speed AMT, automatic manual transmission, but you have that EV fill on the front axle.

2:11:13You've got the middle EV motor, the MGU, so you have a really, really seamless gear shifting experience. It feels much more like a DCT than a single clutch, but you get the weight of a single clutch package. How early did the door design come together? It's pretty striking. I mean, people have done scissor doors, butterfly doors, all sorts of things, but there's something about how long these doors are that really stands out. Yeah, it had the record for the world's longest door for some time and that's because of course you're opening the front and the rear at the same time and you've got this 45 degree angle.

2:11:48It looks like wings coming up. It's part of the design brief, part of the functionality brief. We didn't want to split the passenger side into a separate door from a styling perspective and then... Can I get in the cockpit for a second? Yeah, yeah. Yeah, let me show you how to do that real quick. Maybe spin around this side. Let's do it. It's a two-stepper. So you're gonna take a seat right here. Take a seat first. So this is the battery box and then swing your legs in. Yeah, you don't want me to put my feet right on that exposed carbon. Okay, turn it on. Yeah, swing the feet. You got it. Nice.

2:12:21Good. Here we go. Yeah, when you're in there, because you got the center seat, you actually have a lot of room. I mean, compared to the Valkyrie where you're pretty squished and you're also at an angle. This actually has a lot of headroom. When you have a helmet on in the Valkyrie, you hardly have any. Yeah, you're probably touching a tiny bit. And that seats almost all the way forward. You got seat controls on your left. What is the name in there once? So stands for 21st century. Okay. So divergence mission statement is build the 21st century industrial base. Sure. So this car represents the 21st century in the vehicle space.

2:12:57Can I pull this down for a second? I want to see. Wow. How important are the materials? I mean, I imagine... Put your foot on the brake. I see. Do we have a key in here? Yeah. Oh, there we go. Oh, baby. I'll see you guys later. See you later. Whoa. Okay, so you get a feel for it. You've got your different modes. You've got an EV-only mode. You've got a track mode. You've got a sport mode. And you're pulling gears on the steering wheel there. you've got all your controls on the wheel centralized cool features you got some 3d printed ac vents even you've got your 3d printed uh speaker grills you've got the wheel the center node and it's a full carbon body but the full chassis then is printed underneath yeah lock them in there there you go wow wow it's such a different experience actually feeling like you're in cockpit here insane in the green yeah man that urge to just throw it in reverse yeah we should go for a drive this america's hyper car insane yeah how important are the materials i imagine this door if you made it out of steel it'd be incredibly heavy you couldn't lift it what what are you using as building materials and how important is that so the entire chassis architecture is what really differentiates the car as well.

2:14:29So when you look at it from an optimized system, we didn't just optimize rear frame, then front frame. We actually built a full vehicle model with our topology software. And what we're building on the chassis is additively manufactured aluminum, which are actually our own chemistries. So they have the right strength and durability, but also for auto elongation to absorb energy. So if you get in a rear impact, you need to have a material that doesn't shear or break. And that's really, really hard to get in an additive material. We had to actually make that aluminum in-house. The body then is fully carbon fiber.

2:15:04So you're marrying essentially 3D printed aluminums and ink canals with a carbon fiber exterior for aero. You're taking all of your load cases, your main large forces, through that additively manufactured chassis, and then all your aero load through the carbon body. And then what pieces of the cars, I mean, Michelin tires, you don't make the tires. What else is in the supply chain where you're buying instead of building? Yeah, we have great partners in the supply chain, right? We can't build every piece of this. So you named it tires, infotainment system. Oh, sure. Right, we've got Apple CarPlay on there.

2:15:37Of course, yeah, you don't want to write your own software necessarily. We have a partnership on our transmission where we actually did part of the design. We additively manufacture the housing, but then someone else cuts the gears for us and integrates it. Okay. On the engine side, we designed that V8 ground up. Really? actually build that V8 as well. John did you get a look back here? Yeah let's go look at it right now. On the battery side we've got a partnership there so almost every system on this car is bespoke either fully engineered and manufactured by Zinger and divergent or done in a JV where we have a partner where we've taken a design brief to them said this has never been done before will you sign up to do it with us?

2:16:13And then this sort of structure, this sort of structure, it looks very organic. Where did that come from? Whose idea was that? Is that someone in CAD? Is there AI involved? So that's the computer's idea. Okay, interesting. That's not the designer's idea. So when we talk about divergent process, design, print, assemble, the design portion is a fully integrated set of microservices that takes requirements and turns it into optimal CAD. Okay, so you're setting parameters. You want the lightest weight for the most strength with this performance in this volume because it can't be bigger. And then it will go and try all sorts of things.

2:16:48And then you can finally use additive manufacturing to manufacture the part. You got it. So 10 years ago, we started in machine learning, the basis of modern AI, right? And AI-based algorithms and prediction to do a weighted optimization function for designing these parts. Importantly, we simulate the full manufacturing side as well. So once we've designed say this rear cross brace and that wing is going to interface directly on that point So that entire rear Cross brace is gonna take the aero load of 5 ,000 plus pounds onto it Interesting, three of these bolts Yes, so you're going to give it that force vector, that downforce Oh, and it just loads and then that just fuses across the whole So then you also give the software the crash impulse case, the durability case, the load cases coming off the road, the vehicle the tires all into that functional system.

2:17:39The software actually then says add and subtract material against those requirements until we hit a Pareto optimal point. That means it's the lightest weight while meeting those requirements. And then the beauty is it's already simulated for manufacturing. So there's no design for manufacturing loop. We go through our MES homegrown and actually make that part and then assemble it with our robotic cell. So this is chassis 28 of 80. You said there's 16 that are out in the wild already. Do you see a 12 or something that are maybe ready to go? Is that right? Yes. So there's cars in between. This one's also a special number for us.

2:18:12Sometimes we can't request the exact number on a vehicle. But we're typically around 20 cars a year. These are the records that this car has set. Yeah. So this is Laguna Seca and Circa of the Americas, which are two of the main American iconic tracks. Right. This is probably our best track in California. Yeah. The beauty about this one is it's outdated. So we just had an epic back and forth with Koenigsegg, the Swedish hypercar maker. And we set a 122.3. So we beat our previous record by over two seconds, beat their record by almost two seconds, and put a whole new benchmark for Laguna. I love that you have a rivalry with Christian von Koenigsegg, but just settle it on the track.

2:18:53It's just beautiful. I don't know. It's industrialist for hundreds of years in America. I've been building the fastest train, the fastest plane, now the fastest car. Is there any ambition to one day make cars for everyday consumers, or is it racing and hypercars? Zinger will stay at the tip of the spear. It's all about big R &D budgets, making a car that is a next step forward in terms of consumer experience, and therefore it's got to be expensive. It just has to be. It's the most modern, high-performance niche that does develop systems that then do make their way into more mainstream vehicles.

2:19:31but Divergent is the company that's going to do the Ford F-150 one day on the chassis technology side. With partners. Exactly. So that tier one supply, the volume game of doing millions of vehicles, that'll come through Divergent. The tip of the spear, highest performing vehicles in the world, that'll come through Zinger. That's amazing. Very cool. Anything else we should know about the car? Should we go sit down inside, talk about business and how things are going? Let's talk about business a bit. I think that's that for you all is to drive it. This view right here, Michael, you gotta get this view, this shot from this angle is just absolutely insane.

2:20:12What a beast. Stunning. Absolute beast. Yeah, that's real functional era. Yeah. The scale of the car is just, is, is. It's wide. It's just everything. It's wide, long, tall. Yeah, that front splitter. Everything about it is. yeah you can put a lot of weight on that it has such a presence give you a quick shot here there we go it's actually functional it's got to insane let's go sit down we'll chat and dig in more we are back in the ultra dome we'll have you sit right here and And we're back. It's fascinating. I mean, I think we hang out long enough. Jordy's eventually going to browbeat you into doing the McLaren track and eventually making road cars or something.

2:21:10Or, you know, just like. Yeah, just fully road legal. Yeah. How difficult is that process? Oh, it's difficult. Yeah. Because we went through. This is not an EV car. It's a strong hybrid, so it has an EV system, but a combustion engine as well. And we went through full California carb emissions compliance on that V8. Just heard it start up, right? And that is a difficult process because they're looking at while the car is running, when the car first starts, what the evaporation is of fuel or oil in what they call shed test as well. So there's a huge amount of requirements to actually certify that car.

2:21:46I think the amount of fuel you actually burn in the certification process might not actually make it logical for cars that are this low volume. But you go through that and you show that this car is actually designed to be environmentally friendly with that V8. What would you do if you're making a track only version? We would strip out a lot of weight and a lot of the safety monitoring software and systems. So first off, you're talking about all your airbags go, all your passive safety environment goes, a lot of electronics that are there for redundancy. Do you have any that have like a full roll cage in them?

2:22:21or was the intention just like, hey, we want to set records with - Yeah, so road car, right? Maybe in the future, we'd make a track only car. It would be different almost entirely. That roof design, the carbon A pillars that run all the way across, that is rated for a rollover case. So you can roll that vehicle and have similar performance to a rollover cage, but integrated there. You don't want a roll cage in a road vehicle because it's a very hard surface in the car. and if you're not harnessed in, you're going to move more and you're going to hit your head against it if you get in a crash. So the roll cage on the track where you have the harness on and the helmet makes sense, but the roll cage without the harness and no helmet, it actually is a major hazard on the road.

2:23:05Yeah, talk more about the safety approval, getting the car road legal. Is there a hurry up and wait element while you're in the queue? There are other cars and there's a rate limiting factor there or is it a quick back and forth, but you're making dozens of changes constantly? What's the actual process like? So it is an iterative process. It is known to be a real headache for certain OEMs in terms of the timing and attention that they get and how quickly they can iterate. And I'm sure if you're like Ford and you're like, there's a new Corolla coming, or that's Toyota, but there's a new F-150 coming every year with a refresh and we're changing this.

2:23:43They have a whole system set up. It's not as bespoke. So they must have a whole team for it. You sort of have to build that out. To California CARB's credit, I think they saw that this was one of the first new OEMs in America, period. And there was a narrative that they wanted to support there as well. And we managed to get their attention. And our team was able to work with them effectively. How much were you using this as part of the sales process when you're pitching, let's say, other OEMs to make parts with you guys? Are you saying like, hey, if you don't believe me, get in the, let's go for it?

2:24:19Early on, it was like the proof is in the pudding moment, right? You come to our factory, you see our car, you can drive it. If you're Bugatti or McLaren or Aston Martin, you see that. And that's a huge benefit, right? So we use that a lot early on in the company as a sales tool. Now in defense, we're hosting anyone from Secretary Hegseth to a CEO of a prime. When they see a company that can integrate a road legal vehicle and they see that manufacturing process, it just builds a lot of confidence, right? There's a cool narrative. XF loved revving that and car up. But there's also a capability narrative.

2:24:53And that systems engineering is proven out through that car. Yeah. I mean, I was looking at your careers website. You're hiring tons of people. How scaled is the manufacturing plant? How large is the team? How large is the business? How are you growing these years? So we've gone in the last three years from not being a defense-focused business and divergent to being a defense-focused business. We're dual-use, dual-tech. So we do have our commercial arm. The auto business is growing. Our supply to really the European luxury brands is growing. But the focus is on defense right now. Three years ago, we had zero contracts.

2:25:29Today, we're on over 40 active contracts. And that's what's really unique about divergent is we're not competing head-to-head with Andrel and Lockheed and Raytheon. rather we're servicing all of the primes. So you can build this horizontal manufacturing layer as a service to all of them. You can bet not on one to five programs record, but rather service a hundred programs a record over time and be that manufacturing supplier for them. We've gone through that journey of first refactoring the company to work on defense, winning our first contracts, delivering our first prototypes, flying those, proving the performance of a new technology and then actually in the last six months winning our first scale production programs now if you come to the facility you'll see exit rates at around 100 a month of main bodies for large cruise missile systems so 80 a month or 100 a month it's not high volume for auto but for our defense landscape making a thousand missiles per year that's on the high volume side of things that's what divergence factory in torrance is already churning out So now we've almost gotten to full capacity on our Torrance facility with those contracts.

2:26:37We're starting to look at factory two, three, four, five. We've got multiple states that we're kind of shopping and partnering with right now. And then also looking at California because it makes sense for especially the startup community here, which are great partners for us to have another manufacturing platform here. Yeah. What about other applications outside of defense and automotive? maritime comes to mind. I can imagine a lot of that additive manufacturing technology being useful on high-speed boats, yachts, I don't know, aviation, planes. We were talking about Harrison Ford crashing Cessnas all day long.

2:27:15Maybe you need some more reliable manufacturing. But what other areas have you at least thought about expansion into? Yeah, the TAM for this system is so large, and that's a blessing, but you need to be disciplined in what you address first and second and third. Right now, it's a primary focus on defense within that looking at largely munition systems and then looking at torpedoes and similar structures, but on the Navy application side. Outside of defense, we look to space. So we're doing quite a bit of work in space. Optimization, mass saving there really matters. So you think about things like satellite buses as we get more and more launch vehicles and they're carrying different payloads into space.

2:28:01What is the structure that essentially attaches those to the launch vehicle? We've got several programs ongoing there. And then you look out at Divergent over 10 years. Really, any metal structure that takes a complex load path, so has some sort of force that it's interacting with, should be designed by the platform and manufactured by it. So you think commercial air, you think oil and gas, you think mining vehicles. You even think over time as the prices come down enough, construction and where can it get into almost the city planning side of the equation as well. Yeah, yeah, that makes a lot of sense.

2:28:35Wow. What are you seeing on the hypersonic side? I know a lot of it's under wraps, but we've talked to Castilian and some other folks that are thinking about this. And just broadly, the defense industry has been talking about the hypersonic gap over the cruise missile. Are you excited about that? Are you following that story? Oh, absolutely. You know, we're working on multiple hypersonic programs right now, designing primary structure for these vehicles. You know, in terms of narrative or when I zoom out a notch, you need both. You need the affordable cruise missile that's$200 ,000 instead of$3 million.

2:29:11You also need the hypersonics capability across different classes of hypersonics. And for us, what do we need to be good at? At Divergent, we need to be great with our material science. So we need the right nickel-based alloys that can get very hot and still perform. So that's different than the aluminums that you saw mostly on the car. We've designed several of those materials. We have commercially available ones. We've got ones that the government has actually contracted us to do a special materials development for a higher performing version of what's commercially available today. And then we need to be able to print these at scale.

2:29:43A company like Castilian is really not only about hypersonics, but hypersonics at scale, right? They want to make a lot of these. They're down the road from us. We want to help them make a lot of them. That's cool. Talk about what's unique about additive manufacturing in aluminum, nickel. I think most people will be familiar with 3D printed plastics, but is it similar to just soldering lines of solder and just building up slowly? It's the most advanced way to make a metal part, and it's the most controllable way as well. It's not really related to, in terms of technology, 3D printing of plastics.

2:30:22Maybe the narratives are tied a bit together. Yeah, it's not like molten push through a nozzle. We do something called powder bed fusion. We have a powdered aluminum or ink canal in a build chamber. That build chamber has no oxygen in it. It's got nitrogen instead, so it's non-reactive in there. And then you've got high-power lasers overhead. And those high-power lasers essentially melt, laser weld, one layer of that part at a time. You put a new layer of powder on top, and you laser melt that layer to the layer below. So if you think about a part, think about it standing vertical, and think about it as a layer cake, maybe 5 ,000 layers, and you're laser welding with those lasers every single layer, layer by layer.

2:31:02and you're getting full data at every single layer. So maybe that one section of a defense product is 7 ,000 layers. I get 7 ,000 layers of information on that field. So this is the most controllable process in terms of in-process monitoring. It's the most advanced way to make a metal part. And you also end up being able to make exactly the net shape you want. Because when you have a cross section, you can do the exterior and the interior. And you have full control of that geometry. You vacuum away your unused powder, you end up with your net shape. So in terms of waste and efficiency of the process, you can get something that runs at a higher OEE when we talk about manufacturing efficiency.

2:31:44And you can get something that actually performs better. So when we talk, you know, application focus, if you're a missile system, we typically are seeing about 30 percent lighter structures while hitting the requirements. And more importantly, we're able to functionally integrate that structure with the fuel tank, with the avionics mounting, with that whole complex system. We're usually increasing fuel volume by about 30 to 40 percent. So if you're a missile and you get 30, 40 percent more fuel volume, that's a real game changer. And that's what the customer looks at Divergent to do is, one, get the best engineered product.

2:32:17And that's going to change in a positive way the performance of your vehicle. and then two, scale the manufacturing. And that second part is where in the modern age of defense where we are looking for volume is where there's a big gap right now. And that is part of the reason why Divergent has gotten one, so much business, but two, so much attention as well is it's starting to be viewed as that manufacturing layer. Our mission statement 10 years ago, build a 21st century industrial base, it's actually starting to happen now. That's amazing. Last question for me. the pieces that seemed clearly made in your factory with additive manufacturing, they looked white.

2:32:58They didn't look like metal. Have they been coated? What's the post-processing once you make a product? Yeah. So our aspirin and aluminum comes out almost like a silvery white color. So what you're seeing is probably some of that, but we do do coatings, especially on the defense side. We have a fully automated anodized line. So it's essentially a process where you clean the part, dip the part, and then one of those stations, one of those tanks is actually electronically charged. So when you're anodizing, you're essentially plating that part with a certain coating. That makes it more corrosive or more resistant to corrosion for Navy applications.

2:33:35But also, depending on the system, once it's fully integrated, if it's a stealth system, there might be a special paint that goes over top. There might be a gap and flush process over it. And Divergent will either do that or partner with the prime to do it. At least prep the material so that it can go get the special paint that might be classified or something. Yes, exactly. You're handing out. Last question from my side. Want to understand how local kind of politicians and how, like, what's your relationship like? How does Torrance think about kind of neo-primes and overall building up the manufacturing base in Torrance?

2:34:10Yeah, we've had a strong relationship with the city, which is part of the reason we've been able to get fast permits for our factory. You can think about something like automated anodize. That's usually something that can take some years to stand up. We've been able to build relationships that allowed that to accelerate. Are there states that are lower cost and probably even faster for permitting and manufacturing? Absolutely. In terms of engineering headquarters, Los Angeles or the greater Southern California area is very special because we have the environment to create hardware and test hardware.

2:34:45And we've got some of the brightest people in the world all centralized. And now we have a VC community and general, I'd say, growth funding community that is focused in this area as well. So Divergent will stay headquartered here. we will go through some serious scale production here as well but over time you're going to see factories across all the states in the u.s and some of those factories will be at a lower cost basis than the ones here yeah very cool it's so awesome i love it i just everything about this company is so fun uh congratulations on all the progress and seriously thank you for bringing the car what a treat what a special moment we've been wanting to do a tbpn track day yeah you'll be one of the first to know on a date set.

2:35:28You've got to bring it out. Yeah, you can come out and smoke us all. Absolutely, dude. Laps around us. But thank you so much for taking the time to come on the show. This was an absolute pleasure. Thank you. I'm sure we'll be talking much more over this year as you roll out everything. Good to see you both. We will talk to you soon. Have a good rest of your day. Enjoy the drive. And I will tell you about Graphite, code review for the age of AI. Graphite helps teams on GitHub, ship higher quality software faster. And we have our Lambda lightning round starting in just a few minutes. We have Andy McCoon from Cosmos coming on first, then FabFitFun, the co-founder and CEO of that company, has launched a venture fund.

2:36:10He's coming on. And Aaron Katz, the co-founder and CEO of ClickHouse, is coming on. And it sounds like we have Andy in the Restream waiting room. So let's kick off our Lambda lightning round and let him in to the TBPN UltraValue. What's going on? How are you doing? What's up, guys? How are you? Good to see you. Welcome to the show. First time on the show. Thank you for wearing green, too. Yes. Great sign of respect. We love green. It's perfect. My Cosmo Racing jacket in honor of you guys. It's amazing. We just had a green hypercar. Yeah. We were doing a quick tour of. Anyway, first time on the show.

2:36:43Please kick us off with an introduction on yourself and the company. Yeah, for sure. I'm Andy McKeown. We're based here in New York. Cool. 20-person team building the new home for inspiration on the internet. Visual search engine just announced our Series A today. Super exciting. $15 million round. Hit it. And that's what you need to do. I'm slipping. It's live around. Amazing. You got me fired up. Congratulations. Yeah, there's so much to talk about right now. And it's such an interesting time to be building this company because there's just like an explosion of content. You get a pretty picture now.

2:37:22You don't need a camera. Anybody can generate one from their own home. Why don't maybe we start and you can give us kind of the history of the company, what the initial inspiration was. Maybe even go back further about the previous company you worked on. Yeah, for sure. Yeah, in 2017, I founded a company called Unfold. So we are like a mobile content creation tool, kind of canva adjacent um ended up selling that company to squarespace did a few years there leading creator product strategy and i've been working on cosmos for the last four years or so now um and yeah i'm a really creative person always working on a lot of different creative projects and i very much grew up on tumblr back in the day and ever since then i've been an obsessive collector and curator of things on the internet um always just saving things to a million folders on my desktop and folders on my phone, things just in all of my different bookmarking folders on different platforms.

2:38:17And so Cosmos just started out as me wanting to build a place where I could bring all of these references into one home, where everything would be tagged by AI so I could quickly search and resurface anything that I had saved. I wanted to be able to work with collaborators in these different projects. So yeah, I mean, just started out as building a side project for myself as everything I ever built has kind of started out as, um, started getting into the hands of my most creative friends and people were like, this is awesome. Um, and so were those friends, the ones that were uploading the original content, were you scraping the internet to bring on the first and like seed the community?

2:38:54Like where was the actual content coming from? Yeah, we, we weren't scraping anything. Um, it's always come from the users. Um, we have great importing tools and people can upload links from a lot of different platforms and around the internet. So yeah, all of the content is seeded by the community, which has grown in a really beautiful way. We have tens of millions of images on platform now, an amazing global search, all of the recommendation layer that you would expect from a product like this. And what were the early adopters looking for? Is it more like, because when I think about image search, I think about people that are doing home renovations, but also designers that are looking for inspiration for clothing or web design.

2:39:32There's some people that just sit there and scroll just to enjoy their favorite images like you enjoy your favorite books What were the early drivers of growth? Yeah, I think you know, we amassed a waitlist of a few hundred thousand folks before we launched the product and I think if you Look at the cohorts of users that were coming in really early. It was you know designers across every discipline We're talking interior designers architects graphic designers brand designers photographers folks like this which really helped seed such an amazing you know original corpus of content but really where we're seeing a lot of the growth come from now is sort of this you know gen z wave um you know i call them kind of the next generation of tumblr kid sure um you know they they love to mood board for fun you know maybe they're not saving typographical references on a day-to-day basis because you know they need it for their client work but maybe they're just dreaming up a version of themselves that they aspire to be.

2:40:26And they want to surround themselves with beautiful things. Yeah. And then how have you been grappling with the AI generation, image generation boom? You started the company four years ago. It's pretty much pre-AI content, essentially. Now it's indistinguishable. What's the journey been like for you? Yeah. And there's so much pressure on each platform to take a stance. Right now, when you look at most of the platforms, the stance is like AI content is welcome here. We may try to identify. Yeah, we may try to tag it. Yeah. But it's being harder and harder. But for a lot of platforms, you just have to search, okay, I only want to see images before 2022.

2:41:06And it's like almost like an archive that can be really good. But yeah, just how have you been processing it? Yeah, for sure. I think that, you know, we're definitely seeing a lot of backlash within the ecosystem against a lot of other platforms where AI-generated content is kind of proliferating all of the feeds. And, you know, people are engaging with stuff that they don't even know is AI-generated content. And maybe they search the thing and then they click into it and they see that it's labeled by AI and they find that really frustrating. And, you know, we're definitely growing a lot of users now that we kind of consider like refugees from this.

2:41:42You know, we really want to be a platform that champions human creativity and human-created content. Now, that's not to say that we're drawing a hard stance against AI-generated content, but I think that it's really important that people are able to decide what they want to see. And so we've done what some of these other platforms have done as well, is every image that gets uploaded on platform, we're running it through multiple different algorithms to detect if that is AI-generated. And so every user has settings within their profile where they can choose to show or hide AI-generated content.

2:42:14Yeah. Interesting. How confident are you that you're catching all of it? Yeah, I mean, it's an uphill battle, right? I mean, it keeps getting better. You have to keep retraining the models. It's only going to get more difficult. So, you know, I think that it's only a matter of time before it's pretty difficult to be able to do. There's also some AI models. I've noticed where I will upload an image, ask it to do something, and it kind of just spits out the exact same image, or maybe it just, like, adds a piece of text. And it feels like it didn't actually run a new diffusion model. It just sort of used some tool to like layer over text.

2:42:51And it's like, well, it's kind of just acting as an agentic Photoshop at that point. It's not really creating an AI image. Like all of the details and fingers are the same because it's literally the same pixels that it's just using. And so does that count? Yeah, it's going to be all hairy. I'm interested to know how you think about for you pages, algorithmic feeds versus search and being more of a search engine. Is that line blurring? What are the important trade-offs? What do you like and dislike about algorithm design? Yeah, for sure. So we have a For You page on Cosmos that learns your taste and learns your interests as you can say things.

2:43:28I think it's important that you have something like that so people have a new vector that they can discover content through. What's really interesting is the search problem is actually very similar to the For You problem, especially when you get into the world of serving people personalized search. So you train your 4U, we have an idea of what your tastes and your aesthetic leanings are. And then when you search something, you're actually just adding that keyword over sort of this vectorized database of visual aesthetic that you've already identified with. So the problems are actually very related and intertwined.

2:44:00I think one thing that we've done that's really interesting is we've started developing in-house aesthetic prediction models. So we've actually been using all of our own data, user data, user uploaded content. And we've been training models that understand kind of like what good versus bad looks like. And we're using these aesthetic predictions to loosely re-rank our search results in our For You pages to try to keep the slop off of the platform. The slop wars. You deep in the slop wars. Frontline soldier here in the foxhole. I salute your service. I think we believe that your outputs are a reflection of your inputs.

2:44:39And so it's very philosophically important that we're feeding people high-quality inputs because so many creative projects start with the mood boarding phase. And so if the inputs are going to become the output, it's really important that we're giving people a place where they can find the highest quality things. What's been the growth loop for you? What's been the key to growth? I know that you've been featured by Apple as one of the top 25 apps for 2025. five, you're 500 ,000 followers on Instagram, but is there a traditional like referral flow? What are you using from like the web 2.0 boom era versus viral social playbook?

2:45:18What's been the big growth engine? Have you had to just run ads? Like what's worked? Yeah, we haven't been running ads. I mean, it's really been organic up until this point. You know, we had an early like invite friend flow, which was pretty successful for us. I think that that's not as successful as it was a few years ago when we ran it. But I think the nature of Cosmos is really collaborative. You start what we call a cluster, it's like a collection. You start a cluster, you start inviting your collaborators to those. The experience gets so much more valuable. And we're seeing really high density as well within some of the best creative teams in the world.

2:45:55I mean, we have hundreds of users on platform on Nike.com emails. We have teams at Chanel using Cosmos, teams that Apple use in Cosmos. So that density within creative teams has been really, really powerful for us. And then monetization, I feel like you're going to be anti-ad just because as soon as you have an ad, it's hard to influence like, okay, you can't say like, well, you can only serve pretty ads. For$50 ,000, you can post the sloppiest AI slop with six fingers, but we're going to charge you a pretty penny. No, no, I want to know the real monetization strategy. Yeah, for sure. So we're not monetizing yet.

2:46:31I mean, we have a premium subscription right now, but it's pretty demoted within the product. Sure. I think, you know, there will be a lot that we have yet to reveal. But I think that where we're going is, you know, going deeper on sort of focusing on those creative teams that are using Cosmos and how can we better serve those folks. And then we're also starting to build into commerce as well. So we launched a commerce pilot last year. And you know, you can imagine, you know, we're running hundreds of thousands of searches a day right now on the platform. And if someone searches for brown leather couch, they're probably looking to buy a brown leather couch.

2:47:02Um, and so how can we, yeah, how, how can we create a compelling commerce experience that doesn't feel like we're trying to force products down your throat? Um, but it's really commerce through, through inspiration. Yeah. Yeah. It's like aesthetic first taste first, as opposed to like feature set price, like all the filters that you see when you're on Google shopping, it's a complete, like those are probably going to be way lower in the funnel than the aesthetic. Yeah. Shoppable mood board. You can think of people doing inspiration for, you know, interior design, their wardrobe, all this stuff.

2:47:34It's like, that's, that's a high intent moment when somebody is trying to, um, yeah. Can you, can you explain a little bit more about the aesthetic trends that are popular? I I've seen these different, uh, like, like phrases used to describe how certain people describe their whole aesthetic? How would you describe the communities that have formed around these modern sub-niche aesthetics? Yeah, it's a really interesting question. I mean, we launched a new feature today, so we gave the profile a really big refresh. Historically, everything that you saved had to go within a cluster. The release that we put out today gives you a core feed of images that sit at the forefront of the profile, which is really akin to your Tumblr blog back in the day.

2:48:20It's really like a snapshot of your taste and what you're into. We're starting to go really deep on the data and the trends that we're seeing on the platform between what people are searching, what clusters are they creating, all of those sorts of things. And we're going to start putting out trend reports, which we're really excited for. And we've started sort of seeding these into our newsletter, which has been really, really fun. That's great. That's cool. But yeah, it's been really cool to see these different sort of niche aesthetics blossom. I think historically, a lot of the usage has been just around, again, the core use cases that we were talking about, like graphic design, interior design, whatever.

2:48:54But if you think about something like Tumblr, right, there was like fandoms that were popping off and, you know, like emo aesthetics and culture, right? So we're starting to see some of these pop up, but it's a little early to tell. But we're really excited to start publishing these more publicly. Okay, last question for your personal aesthetic. The Pantone color of the year for 2026 is Cloud Dancer. It's a soft, lofty white that symbolizes calm, peace, and fresh starts. Underrated or overrated? You know, I'm a neutral palette guy, but I know it was really contentious within the creative community.

2:49:30I think we could have gone a little bit more creative with it, but I'm never going to hate on a neutral palette. Okay. This is pretty out of pocket for me. Okay, so it sounds like Cloud Dancer might be working its way into your cosmos. That's a good name. eSpooky in the X chat says, how do you avoid creating an automatic house aesthetic if you're algorithmically filtering slop? Any thoughts there? Like, is part of this, like, it is so... Yeah, like, a lot of... You've generated that somebody might have, like, you're not trying to, like, have an overarching aesthetic other than, like, somebody could come on and create a bunch of clusters for an aesthetic that is not aligned with your personal taste or the taste of Cosmos.

2:50:12The art station look became the mid-journey look, and then now that just looks like the AI art look. But, you know, what if you're actually just into, like, you know, matte painting? Like, you're going to get filtered, maybe. I don't know. Yeah, I think, like, the point of the aesthetic prediction model is less to peddle what we think is good and is more to set a bottom bound on what we think is bad. And so that bar is set pretty low. But it's a great question. You know, we're continuing to retrain the aesthetic models that we have. And we're kind of going through this right now, but we're trying to widen the aperture of that to service more folks and to reach a broader audience.

2:50:52So I don't think there's a clean answer to it other than just like, you know, continuing to iterate on your models and pass them into production and measure and learn from there. Very cool. Well, thank you so much for taking the time to come hop on the show. Great to meet you and good luck. Great to see you. We'll talk to you soon. Goodbye. The New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. And with that, we will... Oh, you can hear. I hope you can hear it. I don't know if it's coming through the mics. The 21C is fired up. That V8 is roaring. And we have Michael from FabFitFun and Deep33 in the Restream Waiting Room.

2:51:30We'll bring him to the TBP and UltraDump. How are you doing? Welcome to the show. Welcome to the show. Hey, guys. Doing great. Good to be here. Thanks so much for hopping on. Give us the news. Look at this jacket. I want to ring the gong. First of all, this jacket is fantastic. You have news for us. Watch the news. I mean, you guys always dressed to impress, and I didn't want to let you down. Incredible. Thank you. You raised some money. How much did you raise? We've got a$150 million fund. We've closed. There we go. Oh, my God. Let's go. $150 million fund. What sub-markets are you most interested in?

2:52:05What strategy? What stages? How are you thinking about focusing your investing work over the fund lifecycle? Yeah, so we are a deep tech fund focused on the U.S.-Israel corridor. So we'll be investing across U.S. and Israel. And so, you know, what's unique, especially about the kind of the Israel ecosystem focus is there isn't a deep tech fund, let's say in the mold of what's, you know, an eclipse, a DCVC, a lux capital focused on that ecosystem. And we just think there's enormous opportunities. What are the great recruiting hubs for entrepreneurs in defense in Israel? We've heard about Talpio and a few other groups.

2:52:51Where are you finding talent? Are they, you know, ex-founders themselves, repeat founders? Where are you sourcing deals from? So interestingly, actually, the former commander of Talpiot is actually one of my partners in the fund. There you go. For those of you who aren't familiar with Talpiot, you know, there's a lot of different elite units in the IDF. But Talpiot is the unit of the physicists. It was about the IDF made to really train physicists and combine that with the defense industry. So we do, we work closely with the different groups. Another of my partners is from 8200, which many of you might be familiar with.

2:53:36It's kind of the cyber unit within the IDF. We have tentacles into those alumni groups and networks, as well as a lot of the institutions of higher learning. So Weitzman Institute, Technion, et cetera. We're kind of, I'm actually the only non-Israeli on the team of five partners. So we're deep in that ecosystem. Five partners. That's cool. Out the gates. It's sort of the counterposition pretty squarely against a solo. Solo GP. Yeah, there were a lot of people that were doing that. How did you all meet? And what was the concept between the maybe larger partnership at this phase? I don't know. Yeah.

2:54:13So we have what we call an expert first model. My main co-GP is Lior Prosor, who has been a venture capitalist for 15 years, decided to start a new platform because of the opportunities we thought were emanating specifically in deep tech and out of the Israel ecosystem. But to do deep tech right, you actually have to have significant capabilities to go deep on the technical side, to have people who are domain experts. And so three of our other partners are actually domain experts in the categories that we're most bullish on, one being quantum computing, the second being AI infrastructure, a third being energy.

2:54:56And then we also have a venture partner and advisor in the robotics side. How are you thinking about point solutions versus defense platform companies? Are you looking for a multi-product defense firm that can build sensor towers and drones and submarines? Or do you want to invest in a bunch of different teams that are doing smaller product bets, see what works, and then maybe scale out of that? What's the current meta on the ground in Israel? Yes. So interestingly, although defense is an important part of the ecosystem and shoots off a lot of the innovation that we are investing into, we don't think of ourselves as a defense fund.

2:55:44In fact, we're not focused on kind of kinetic battlefield companies, although a byproduct of the types of companies and technologies that we're investing in oftentimes do have defense ramifications and capabilities on the battlefield. We just think that the commercial opportunities of those technologies, when they extend beyond defense kind of specific companies, are much bigger. They can have much bigger outcomes. So there's a lot of dual use in the types of things that we're looking at. And what we're looking for is just insanely capable founding teams, really groundbreaking kind of technology and IP and enormous end markets.

2:56:28And when you look at the confluence of things that Israel has in terms of the defense industry, the institutions of higher learning, the concentration of elite scientists, RF industry, silicon industry, optics industry, et cetera. There's really, you know, the only other geography in the world that has that sort of concentration of those types of mixes of assets is really California. And you see what California produces. So we think Israel is kind of this looming deep tech giant. In a lot of ways, the ecosystem there in deep tech is kind of where cybersecurity security was maybe a decade ago in Israel.

2:57:05And now it's this powerhouse that's producing these, you know, just insane outcomes. We think of the next decade, the deep tech outcomes from Israel are going to be right up there. Where did the name come from? Yeah. So I think deep speaks for itself. You know, one, we go deep and we're focused on deep tech. And 33 is actually, it has a mystical connotation in Judaism. It's the day of Lagba Omer, which is a Jewish holiday where the kind of the depth of the hidden secrets of the universe were revealed for the first time. And that's what we really think our founders are doing. Our founders, our companies, and kind of the mission of the firm is to go into the technological and scientific frontiers.

2:57:47And we think of that as a way of revealing hidden truths. The type of work that these companies are doing, what we're trying to build is really frontier tech and new innovation that moves the world forward and moves our understanding of the world forward. That's cool. Last question for me. Energy. Do you put that in the deep tech realm? Are there particular areas within energy that you're interested in? Solar, nuclear, wind, anything? Yeah, it's a great question. Absolutely. We We actually have two energy companies already under our belt. We, you know, in particular kind of our thesis on energy is where the point of production and the point of consumption are brought together.

2:58:32Sure. So where can you create unique types of energy outputs on location, on site? So for instance, if one company we haven't announced is able to produce carbon-based fuels out of a laboratory, and there's a scale -up model around that that we think is extremely compelling. you know, the name of the game for a lot of, you know, U.S. security interests, Israel security interests, and the shared interests between the two of them is this kind of supply chain resilience. I think we saw the president speaking at Davos and even, you know, describing maybe, you know, the kind of the counter swing of globalization is here.

2:59:20And that means people having these capabilities on site being less reliant on any specific partner and trade for their national security and so that ability to bring things on site becomes incredibly important It's amazing, well thank you so much for taking the time to come chat with us, congratulations Next time you join, you can join from Tel Aviv, I'm assuming are you calling in from LA right now? I'm actually calling in from New York here, it's Deep 33 we're across LA, Tel Aviv and New York. So the team all came together, you know, announced the fund and do some critical work together. So amazing.

2:59:58Well, I'm sure you'll be back soon. Yeah. Congrats. Thanks for having fun together. Have a good one. Goodbye. CrowdStrike, your business is AI, their business is securing it. CrowdStrike secures AI and stops breaches. That's right. Up next, we have Aaron Katz. He's the co-founder and CEO of ClickHouse. He'll be joining in just a minute. In the meantime, time, let's head over to the timeline. Elon Musk has confirmed that he's unhappy with the Twitter algorithm. Data Hazard said, the Twitter algo has gone to hell. I want to again see obscure technical posts from outside my follows, not this cornucopia of low IQ rage slop.

3:00:38That's interesting coinage. Elon chimed in and said, I agree, sigh. So that's the thing. Anytime that there's a bad day on the timeline, bad week on the timeline, I feel like they're pretty aggressive about just mixing it up. I don't know. You might not like the new thing, but it's going to be different. Command C. You're going to see some doing stuff. And there's been some fantastic deep fakes of Elon Musk as the farmer from Babe, Pig in the City, looking at the slop and being sad. So fun day on the timeline. Good that there will be some changes coming. Anyway, without further ado, we have Aaron Katz from Tech House.

3:01:10Welcome to the stream. Thank you so much for hopping on the show. Good to meet you. How's your year going so far? Well, the calendar year is behind us, but our fiscal year is about to finish the end of this month, and it's going really well. We got, you know, whatever, eight or nine days left, and, you know, we're off to a good start to finish the year in the fiscal quarter and get the new year up and running beginning of February. What's the theme of 2026 for the company? Is it just more AI? It feels like AI is so obvious, but also what else are you going to focus on? Well, you know, ClickHouse has been a back-end database to a lot of agentic workflows over the past couple of years even before the chat GPT moment.

3:01:52We're being used by Lovable, Cursor, Anthropic, OpenAI. Pretty much every leading AI company right now is using Cluckhouse in some sort of capacity. But it's not AI companies. I mean, we're the back end to technologies like Ramp, for example. You're a sponsor if you talk to Eric. We power most of the analytical experiences if you're a Ramp customer. And so it's just kind of doubling down on what's gotten us here. It's a very large surface area that we're going after and just kind of broadening the competitive landscape. Have there been any challenges on actually scaling the business from an infrastructure size?

3:02:30You're at a scale where many AI companies are struggling with data centers, GPU poor. How has the actual build versus buy decision changed over the last few years? How has the infrastructure side of the business changed? Yeah, well, scalability is one of ClickHouse's strengths and always has been. I mean, we've got, you just talked about Elon's. Many of Elon's companies are using ClickHouse. Tesla is a great example of that, where they're ingesting a billion events per second into ClickHouse. And there just isn't another database in the world. I think you need to start a list about who's not using ClickHouse.

3:03:07Yeah, just tell us who's not a customer and then it'll be quicker. Yeah, fair enough. I mean, you just kind of get your head around the volume of data that these LLMs are generating and the ability to ingest that in an efficient way to be able to store and analyze that. You know, ClickHouse really shines. So scalability is something that we really promote and our customers talk about as one of the unique characteristics of ClickHouse. Yeah. How are you thinking about just walking through the ladder of different analytic suites and analytic functions and just case studies that can actually, that are best practice today?

3:03:44I mean, people are familiar with, you know, like big data, MapReduce, you know, count a bunch of things across a bunch of databases. We're in a much more modern era. What are you seeing companies do that's unique, surprising, interesting, or uniquely enabled by ClickHouse? Yeah, you mentioned MapReduce or any sort of kind of batch-oriented processing. It's all moving towards real-time, and Ramp is a great example of that. If you're a user or a customer of Ramp and you want to run analytics on spend patterns inside of your company, you're not going to wait two to five seconds for that dashboard to render or that graph to update or the report to produce results.

3:04:20You expect to see those in milliseconds. And so there's just this broad shift from an analytical perspective to real-time experiences. Batch processing, even waiting three to five seconds for any sort of dashboard to refresh is just an eternity. And so that's a big pull to why people are moving towards ClickHouse. I'm sorry. Yeah, I wanted to ask about the engineering culture. last week we were at a dinner and somebody was telling me about i don't i'm i'll kind of mix up some of the details but they were telling me about like the an investor was going through the diligence process with click house and click you guys had shared like that uh that they couldn't believe how like how few engineers had created some of the early technology and they just honestly didn't believe it so i wanted to ask about kind of the early engineering culture and and how that's evolved and what you guys are doing on the talent side that has allowed you to work with so many of these massive, important companies?

3:05:24I mean, I love that question. I don't get it often enough. The engineering team at ClickHouse, the broader community behind ClickHouse, but the creators and the committers that are part of our company are the best engineers that I've ever worked with. And I've been in the industry for quite some time. I know you've had Mark Benioff on your show several times. I spent 12 years at Salesforce. It had an incredible engineering culture. The team that we've assembled here is specifically in infrastructure software is the best in the world. And you're right. It was a very small team of primarily Russian engineers inside of a company called Yandex, which was commonly referred to as the Google of Russia that actually created ClickHouse originally.

3:06:05My co-founder, Alexei Milovitov, who lives in Amsterdam, is the one that actually named it. It's short for Clickstream Data Warehouse. And so he was thinking about the data warehouse use case when he created the database itself and then made the brave decision to open source it in 2016. And that's essentially you just expose your source code to the world, and then anybody can adopt and deploy the technology with no attribution to the creators of the software without any sort of commercial relationship. And it was a small team. It was a dozen engineers who are the primary committers that created this incredibly powerful feature-rich database.

3:06:43And it's a total joy to work with this group. They're all in Holland, in the Netherlands now, all part of our company, and continue to advance the project. And then we've built a managed service that runs in the cloud called ClickHouse Cloud that they're also contributing to. And that's the primary business model. And this is a managed service that runs on Amazon, Google and Microsoft infrastructure. And then in Asia and primarily in China, we have a partnership with Alibaba. How are you thinking about the way that developers and teams are selecting database backends, data storage products in the age of agentic coding?

3:07:25I feel like we talked to So a lot of folks who they go to an agentic coding solution, they say, I want to build this type of application. I want to build some custom piece of software. And the database choice is sort of made for them by the coding agent. And this feels like it's hyper relevant at the vibe coding, the DIY hack project level. But is this something on your radar where you think that in a few years, Maybe it's already happening. How you show up in LLM search results and how the different AI agent coding models feel about ClickHouse will be important to your business. Well, you know, Anthropic spoke at our event last year in San Francisco.

3:08:14So did OpenAI and Tesla. And Anthropic discovered ClickHouse by asking Claude what they should be using for observability. and Claude said you should be using ClickHouse. And I had a call with the CEO of a very large fintech company in Europe last week. And he said, every LLM we ask what we should be using for this new real-time analytics product we're building suggests ClickHouse. And so, fortunately, a lot of people are turning to these LLMs for guidance on what technologies they should be adopting. And I think as everybody knows, IT decision-making has been completely turned on its head from kind of a tops-down C-level mandate.

3:08:50You're going to use Oracle. You're going to use IBM now to a developer-led evaluation. And open source creates such a fast path for a developer who's building an agentic application to deploy, experience, experiment with technology without any sort of vendor relationship. They can actually push it into production for some of these agentic applications without any sort of vendor relationship. Open source is completely disrupting how these LLM providers and enterprises are deploying these applications. And we made an acquisition last week that we announced on Friday alongside our financing that we're super excited about.

3:09:28It's a company out of Berlin called Langfuse. And they have built an industry-leading LLM observability application so that now enterprises can observe the types of experiences their users are developing on these LLM applications. Very cool. How are you processing all the chaos in SaaS right now, the sell-off broadly, the industry's transition, you know, went through, you know, a couple decades? A big question about the shift from seats to consumption or performance results. Yeah, there's so many, I feel like, open questions, how much of a threat are new. you probably laugh when people say, oh, I made this database with a prompt.

3:10:15And you're like, okay, well, you're missing the 100 other features that an enterprise is really going to care about. But after spending more than a decade at Salesforce and then running ClickHouse, I'm curious what your point of view is. Yeah, John mentioned this move from a seat-based model to consumption or usage-based, which infrastructure isn't that unique. I mean, it's been around for quite some time. Like our customers simply pay for what they use, which is a combination typically of storage and compute. And these are very kind of AI intensive applications. And what's unique about our service is that it automatically scales up and down.

3:10:52So you don't need to provision all of the compute that you otherwise would, because a lot of these analytical workloads are quite kind of spiky in nature. And you want the application to be able to idle during periods of inactivity, for example. And so I do think that, you know, these traditional seat based, although I think we should separate like traditional business applications like Salesforce and Workday from kind of infrastructure software, which is much more of kind of the plumbing of, you know, the Internet and the stuff that you don't see necessarily when you look at a mobile app or you look at a business app or consumer app.

3:11:26It's a lot of the processing that occurs, you know, behind the scenes is primarily where we exist. I'll give you an example is, you know, we where I would historically use some sort of BI reporting to build a graph or a chart to show me how the business is performing. You know, build me a bar chart that shows revenue growth or build me a stack bar chart that shows regional revenue distribution, things like that. And I'd go to some data analyst and they'd go and find the source data and they'd spend a week pulling this together and we'd have like multiple reviews. I now simply ask Anthropics. We've integrated Anthropics, you know, Claude, I think it's 4.5.

3:12:01their latest model with our own MCP server and then a ClickHouse cloud service that has all of our customers, not our customers' data, but like our customers' use of our service. And I can ask that prompt, like build me a bar chart that shows these dimensions and it builds it in like five seconds. And it's just, and it cost me 20 cents, that single prompt. And that simple semantic search is generating like dozens of SQL queries. All of those SQL queries are all being stored in ClickHouse. And if you think about the characteristics of these agentic experiences, like the number one requirement is latency.

3:12:38They have to be low latency in terms of the interactivity between the text prompt, whether it's ChatGPT or anything else, where you put in that query, and then the database itself that's actually storing the data that's providing the response. Yeah. It's fascinating how much of a global company ClickHouse is, the Alibaba partnership, the Russia history with Yandex. We were reading a post about selling software in Japan and how the Japan software buyer often will want to read a lot of documentation and details before jumping on a call with a sales associate. Does that match with your reality in Japan?

3:13:14What can you tell us about the difference of buyer across different countries that you've experienced throughout your career? What's the key to global growth broadly? Yeah, it's a great question. I mean, so half of our employees are outside of the United States. We employ people in 22 different countries. Half of our customers are outside of North America. Half of our revenue comes from outside of the U.S. So the company is extremely diverse internationally. I had the good fortune in 2005, so 21 years ago, Mark Benioff asked me to move to Singapore. And I spent four years helping Salesforce expand across the Asia-Pacific region.

3:13:57And at the time, Japan, I believe, was the second largest country behind the U.S., ahead of the U.K., Germany, France in terms of revenue contribution. I don't think I could be quoted on that because I'm not necessarily true. It's number two or three, but it was like top five. And I would argue it was probably top two if my memory serves me correctly. And you could add South Korea to that same description of just how unique these markets operate and how localized the selling effort is. We recently did, and I've seen both models, and one being you do a joint venture and where you partner with a local company and you share the risk and you share the economics and you share the upside.

3:14:35And they leverage local relationships in the market and local market knowledge, et cetera. or you try to do it yourselves and establish your own KK in Japan. And I've gone down both paths and learned a lot of lessons along the way. And so here at ClickHouse, we did a joint venture in Japan. We recently announced that with a firm called Japan Cloud. And they were previously Sunbridge, which is the firm that did Salesforce.com's joint venture. And so I've worked with these folks for, you know, two decades now. And so, yeah, it's a highly localized market. The buying behavior is very different in Japan.

3:15:10and in South Korea than other parts of Southeast Asia, putting aside greater China or Australia, New Zealand or the Indian subcontinent. So it's a huge market, but not many companies get it right, honestly. They try to do it themselves versus leveraging local expertise. We also brought on some investors, a firm called Geodesic Capital, and that's run by an individual named John Roos. And John, I believe he's still on the board of Salesforce, but he was the previous ambassador for the United States to Japan. And just having those types of relationships are very important for a company like ours that we frankly don't have the same level of awareness in Japan that we do here in Silicon Valley or in New York.

3:15:52Yeah, makes a lot of sense. Fascinating. Jordan, anything else? Yeah, very cool. Thank you for joining. Thank you so much for having me on the show. Yeah, at this rate, I'm sure there'll be many reasons to have you back on this year. We have to ring the gong. There's a recent fundraise. Can you tell us the details? Yeah, happy to. We weren't running a process or raising capital. There's a lot of investor interest. And so we ended up pulling around together kind of over the holidays. That was led by Dragoneer here in San Francisco. And we ended up raising$400 million. We had participation from a number of our existing investors and a few new investors.

3:16:32Pulled together over the holiday. We've talked to a lot of VCs who said that it was easier to do deals over the holidays. There were less distractions at the office. And so plenty of time to hop on the phone with like a tight partnership, get something through. So congratulations. Love it. All right. Well, I appreciate it. Thanks for having me on. Yeah. We'll talk to you soon. Have a good one. Goodbye. TurboPupper, serverless vector and full text search built from first principles on object storage. It's fast, 10x cheaper, and extremely scalable. We are puffing. We are puffing. Back to the timeline.

3:17:03Did we miss anything else today? Yes, we missed the fact that Keith Raboy has hit 5 ,000 Barry's classes. Some people would say that's too many. It's an incredible amount. Only possible, I think, mathematically because he does multiple per day. But what an absolute generational run, sometimes a literal run, on the Barry's treadmill from Keith Raboy. Demis Hassabis from Google told Alex Heath at Sources that he has no plans to put ads in Gemini. It's interesting they've gone for that so early. He said of OpenAI putting ads in ChatGPT, maybe they feel they need to make more revenue. What's interesting is I feel like, yes, it's like At what level do the ads bake in?

3:17:55Because you can access some level of Gemini model through the AI search overviews, the enhanced results that come from Google searches, and they've already monetized those. And it doesn't ruffle any feathers. Of course, on this show, we're very pro ads. And my answer, my response to Demis should be, let's do it, buddy. Let's put some ads in this. I'm ready for ads in Gemini 3 Pro. But of course, they're in a very different position. The business is much larger, much more established. And so they have much more flexibility with timing these things. I do wonder, you know, the whole ads and chat GPT news cycle, it was released on a Friday.

3:18:40It was very much crowded out by the Elon OpenAI lawsuit. it had been rumored and discussed multiple times. Sam Altman had mentioned it in many different podcast appearances. He talked to Ben Thompson about what he liked about Instagram's ads. He had been messaging. Fiji Simo came on to find a couple things and talked to us on our show about it. And so people, it was a very slow process of building up this idea that ads will be coming to ChatGPT. And the benefit of that is that, you know, it's not just all of a sudden, hey, the thing that I like now, stuff full with ads. It's been a very slow rollout.

3:19:19I think it's been deliberate. I think it's smart. I think it works well. We've been discussing whether or not the first chat app to add digital advertising might get like branded with a scarlet letter of like, that's the one that has ads. And you always remember that one is the one that has ads, even if they all add them eventually. I don't know that that will happen, but I have yet to see any real backlash across the internet around, oh, ads are coming to ChatGPT. It's bad. So in general, I feel like the rollout has been handled well. I'm sure we'll see more interesting things when the ads actually go live.

3:19:56People will be taking screenshots of them, sharing them. As long as they partner with good companies like the Ridge Wallet, I think they're going to be, I think people would be thrilled to see Ridge Wallet ads in ChatGPT. Imagine getting one of those like really sketchy like supplement ads, you know, they run these on like display. Yeah, it's like some x-ray and it's like this one thing is destroying your metabolism. Doctors hate. Doctors hate this, yeah. The Bank of America CEO says he hired 2 ,000 recent graduates from 200 ,000 applications. I wonder how much of that is driven by AI applications.

3:20:34Some people are wondering how many of those applications are duplicates. You apply to multiple jobs. You can only be hired once. But Bank of America, clearly growing. Humans and the new NeoLab, founded by Exanthropic, XAI, and Google staff. They're trying to build interactive AI. They raised a 400... This is a Neo-Neolab. It's a post-Neolab. They raised a$480 million seed round from NVIDIA, Jeff Bezos, and others at a$4.48 billion valuation. And it's in the New York Times. You can go check it out. Cade Metz has the story. So that's exciting. We have to do a proper deep dive on the Neolab landscape.

3:21:19What do you have for us, Tyler? I mean, okay, you said it was post-Neolab. I think it's actually closer to, like, Trad Neolab. Okay. I mean, it's kind of like, look, we're going to release models. They're going to be for consumers. We're going to do it with this human-centric view. But it's not like they're going, you know, we're combining this with BioWetLab or something. This is like down the pathway. Yeah. Like Fei-Fei Li's World Labs is many ways more of a post-Neo Lab than this. So if I'm getting this right, Tyler, this is less about, okay, I'm going to send off this agent to go work for a week.

3:21:56and I'll check in with it when it's done and more of this kind of like co-working approach. I believe. I mean, they haven't released much, but it seems like it's very interactive. I mean, like I'm the expert on what they've put out, which is very little. So the name of the company is humans, plural, and then an ampersand, humans and. And Dylan Patel says, we are all calling this startup human sand. Sorry, I make the rules. It's a fire name. Nir says, what are you guys actually doing? Or what are you guys actually making? I literally cannot tell at all. Well, they haven't been on TBPN yet. They haven't broken it down for us like we're a bunch of adventure companies.

3:22:36In other news, our very own Dylan Abriscato got a major scoop. Scoop master. Scooping like a Ben and Jerry's employee. He figured out Instagram replaced the following tab or the following sort of word with friends, which are followers you follow back. Yes, and this has implications. Explain. Yes, the implications here are that some people aren't following accounts because they want to keep, they have some idea of what their follower to following ratio should be. Yeah. If I follow 800 people and I have 100 ,000 followers, it looks really elite. But if I have 100 ,000 followers and 100 ,000 followers following, it just looks like I reply, I'm like doing some scheme.

3:23:22I have some bot going or something. I mean, hilariously, Barack Obama had a follow back bot on Twitter, I think throughout his entire presidency. So this comes up all the time because someone will find some crazy account and they'll be like, Barack Obama follows this. What's going on? And it's like because he follows like three million people or something. Anyways, this is a continuation of the kind of product decision where they allow people to not show the number of likes a post has, right? They want people to post more. So if they're not showing the number of people that follow. Sometimes I just see a part that's blank because they don't want to show me how many people like it.

3:23:55Yeah, it's the user gets to decide. Okay, I didn't realize that. I don't want to be measured by how many people tapped, right? So the idea was to encourage people to post more because they're not feeling this pressure to put up crazy numbers. Oh, I always get 100. If I get 80, I won't feel good that day. So I will just turn that feature off entirely. Yes, correct. So I think they want people to follow more people. And yeah, we'll see. I don't think this is rolled out to everyone yet. Yes. And the net effect that you believe will happen is that there will be more following activity on Instagram.

3:24:27So lots of accounts will be followed. Yes. Yeah. So I will say when I go on Instagram, instead of seeing on this screenshot. This is a very limited test. Yes. So I see friends. You do? But I don't see the post number. I see the friends, the followers, and following. Okay. Where in this screenshot, I think it was, I don't see it anymore. But it was posts. Oh, so maybe you're in another test. I'm in a different test. Adam Masseri is treating us all like guinea pigs over here, lab rats in his maze. It's great. Anyway, is there anything else you want to talk about? We got to end on this. This is extremely important.

3:25:00This kid was sent to the principal's office for lying that his uncle was Superman. The next day, his uncle, Henry Cavill, dropped him off at school. And Andrew Feldman said, dip SHIT school. Andrew Feldman, the CEO of Cerebrus, a$22 billion company, does not like that the school sent the kid to the principal's office. But justice was done. Henry Cavill showed up and stole the day. And it's the wholesome side of X. If your algorithm's all sloppy, maybe you need to adopt Andrew Feldman's algorithm because he's got some great posts and he's not afraid to reply to them. And last but certainly not least, John Palmer says the TVPN guys are just two Fortnite archetypes.

3:25:45John has maxed out stats, 6 '8", 230 pounds, zero V-bucks. Still on default skin after a year. No emotes. Pure fundamentals. Jordy's mid-cracked, only 6 '1", and absolutely loaded with V-Bucks. Rotates legendary skins. Spams rare emotes on the soundboard every five minutes. That I do. This is extremely accurate. I certainly... I feel seen. I feel seen. John knows us well. He knows us well. I'm going to hit the repost on that one. That's a good one. Anyways, folks, thank you for tuning in today. It's only Tuesday. We've got three more shows for you. We've got lots more in-person guests this week.

3:26:25Leave us five stars on Apple Podcasts and Spotify. Subscribe to the TBPN newsletter at tbpn.com. We will be live tomorrow at 11 a.m. Pacific. And have a wonderful afternoon, evening, wherever you are. Yes. We love you. Goodbye.

3:26:42Nice work, brothers. I'll see you on the next one.

3:26:53You were about to slam on the gas. You were like, I'm taking it for a spin. That actually would have been so funny. You were so close to doing it.