SaaSpocalypse Cancelled, Palantir’s $17M Jet Bill, WBD x Netflix Latest | Dylan Field, Saagar Enjeti, Sigil Wen, Peter Morales, Erik Palitsch, Ljubisa Bajic

26:58I wonder if there'll be talks that will come out of this, any more comments or just, was it off the record? Or will there just be reporting out of this? Or maybe it was just talking points, so there was no real news dropped. But if you get those folks together, it should feel like Davos. There should be interesting sort of new positions that are coming out. Yeah, the Mistral founder was apparently speaking there, and the room was almost empty. Wait, really? And so, yeah, somebody was saying, Like, hey, this guy is actually crushing it in Europe. Show him some respect. Put some respect. Sit down and take notes.

27:34Yeah, sit down and study it. Don't just listen. Don't just listen. Let me tell you about LabelBox. Reinforcement learning environments, voice robotics, evals, and expert human data. LabelBox is the data factory behind the world's leading AI teams. J.C. Foster says, three months ago, I quit my job to chase a dream to build an affordable, convenient plastic free coffee maker grateful for everyone who has reserved companies pure steel this is a correct ratio seven million views 34 000 likes this is what you were getting at yeah so huge what is up with the laptop so i guess he was at spacex very long he's at spacex okay and one of the when i immediately saw i mean can we talk about this laptop look at what is going on zoom in on this laptop i think this is he took the it's like an optical illusion okay right it's using the the fisheye lens oh okay because if you fold if if based on this angle if you folded the screen the screen would just cover the keyboard like the trackpad would just be hanging out like an untucked t-shirt under a sweater makes no sense sorry jordy i got distracted by the extremely long, long laptop.

28:50Tell me more about the Pure Steel Company of America. The Pure Steel Company of America. So yeah, so anyway, started going viral. Very, very cool product. If you've tried to make coffee without having plastic be a part of the process, it's incredibly hard and you're pouring hot liquid over plastic. Some of that plastic will disintegrate, end up in your cup of joe. So the product makes a lot of sense. One of the issues is this post went extremely viral. Yes. And somebody said, somebody posted, this thing's only 80 bucks. Go reserve this right now. And so that post started going really viral. Yes.

29:28So it says, do this thing as$80 and made of American glass and steel. Go reserve one right now. And that got 18 ,000. And so this is, of course, not going to cost$80. There's literally no chance. Okay. But I think people were thinking, hey, I can buy this for 80 bucks. It seems like the steel of a lifetime. I think something like this once you actually build it will probably need to cost somewhere like 600 700 dollars something in that range So really rough the other thing is pull up pull up the pull up the website I want I want I think this guy may have knocked off the Rora website. Oh I'm just realizing this in real time.

30:07So Rora is of course the oh, it does kind of look similar So this is pure steel. Look at the header. Look at the header system, which Emmett Shine designed for us. And even look at the imagery. Their imagery is AI. Okay. And now pull up the Aurora website. Oh, it has the same rounded corners. Yeah. It is very similar. I think he got a little inspired. Okay. So if you see the header and everything, the way the buttons work, go scroll down even more. if you go down to some of the more lifestyle imagery, like they, it's the same, like, style, you know, the kitchen, the photography, a little close, but he's bootstrapped.

30:54He's got a bunch of sales now. I'm sure he'll figure out. I'm sure he'll figure out. Okay. Well, I mean, the, the SpaceX, like, lore here is that they should be able to deliver this at$80. Like, It says transparent pricing. It's time to know what you're paying for. Traditional coffee makers cost$200. Pure steel is the same amount of materials, but less markup. And so they want to deliver this to you, I think for$80. Will be very interesting to see how they do it, where they change prices. If they change prices, they might be able to deliver. I don't know. Maybe it'll be VC subsidies. Maybe it's$80 a month, and you have to pay for some subscription to turn it off.

31:39It's like a seat-based pricing. Seat-based pricing, yes. How many cups a Joe do you make? How many coffee drinkers are in your house? Well, no, in the age of AI, it should be consumption-based. So you should pay per coffee cup. How many tokens each cup? Yes. No, but I don't know. This doesn't seem impossible to get stamped steel and assemble it. I mean, it's a pretty basic device, right? It just needs to warm up a pot of coffee. It needs to boil water. That's not crazy to me that that could be$80. I don't know. No, I mean, I know through Aurora, like dealing with these materials. Yep. It depends.

32:19Like looking at some of this design here, the way it's all bolted together. Yeah, it seems pretty basic. Even the labor cost of doing something like this, I think would be... The robots, I don't know. Yeah, we'll see. Amazon does have an all-steel coffee maker. They have some results for all-steel coffee makers in the$60 range. They've got a Black & Decker here. It looks like it's covered in plastic, but it's at least advertised as stainless steel. There's some stainless steel options on here. But yeah, a lot of them are in the several hundred dollar range. So I don't know. We'll have to keep an eye on it.

32:56Yeah, I'm excited about this. I would happily buy one of these. Yeah, it looks great. And I mean, a lot of these companies will underprice the first run. So they'll say, yeah, okay, we're just going to take a loss on the first thousand units that we ship. And then we will raise prices over time. And then when we go into retail, we'll raise prices more. And there will just be a process for that. If you go this viral, if you have a lot of demand, you might be able to raise money to offset that. Your suppliers or you might just make no margin for a long time. There's a way this gets built for$80.

33:34Yeah, and he was at SpaceX for two years. SpaceX is a good organization. We talked to a SpaceX founder yesterday. Except the issue is he was a finance analyst at SpaceX. So he might need to poach some of the engineers. Well, I mean, yeah, analyzing the finances of SpaceX, being like, wow, they make rockets. Wow, we could be making plastic-free coffee. Just do it. Think from first principles. I don't know. Let's see. Van Mann says, the response to this insanely simple yet non-plastic coffee maker should usher in the era of dumb, simple appliances that work and don't poison you or connect to your Wi-Fi.

34:10I think that there is huge demand for this. People want simpler devices. Did I tell you about my - My coffee machine connects to my Wi-Fi. I'm like, actually, I don't think I did, but it always wants - It always actually does. Yeah. It's not a joke. Wow. Okay. Yeah. And it's a lot of steel. I do pour over, so no. But I'm like, you don't need access to the network, bro. Just focus. Just put the beans in the, just put the coffee in the cup. Yeah. I have another story. But first, 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.

34:46Secure every agent. Secure any agent. So I had a little tiny bit of IoT in my house just to turn off and turn on the lights. My dishwasher emails me. Fantastic. Peak performance. I have a tiny bit of IoT just for lights that are hard to reach, lamps that are plugged into basically a wall socket that has a little box that connects to the Wi-Fi. There's an app that turns on that turns off the light. I think the company just went out of business because the server just doesn't work and the app doesn't load anymore. And so all of the stuff that I bought is just worthless. And I just need to rip it out.

35:26It's extremely dumb, in fact. Very, very silly. Brad is dropping alpha. He says, recommend the Ottoni Fabierka for plastic-free electric kettles. That's smart. And then he says they should put ads on the pure steel coffee maker to keep it around. See? See? Think outside of the box. I'm calling it$80 shipped. That's what it's going to cost. No, there'll probably be a shipping fee on top of it. Shipping to Alaska. That is dangerous territory. Don't talk about Alaska. Do not give free shipping nationwide because Alaska is technically part of the nation and you will be paying$50 to get stuff up there.

36:01It is not an Amazon Prime necessarily for your default upstart e-commerce business. So I got completely cooked by my Wi-Fi light system. And I was thinking like, okay, certainly I can vibe code this. Definitely not. Definitely just going back to switches. Definitely going back to the Stone Age. I'm just going to be lighting candles in my house because that's the future that I'm going to be living. Yeah, I had a funny moment this morning. John and I went to breakfast with Brandon. I parked in a garage. John was on the street. And as I was leaving, I went and put the ticket into the machine. Then I put validation in right after.

36:42It just like was working for a while. Then it failed. It tried it again. Then I was like, okay, the validation is maybe not working. So I paid with a card and failed. And so then I start calling around and I'm stuck. I'm stuck for 15 minutes. John is long gone. He's at the Ultradome already. I'm sitting there on support with somebody who's not even in L.A. trying to get somebody in L.A. to just let me out of this parking garage. Then somebody finally comes after 15 minutes. And I was like, hey, can you escalate this? This kind of this your software is like bugging out. It doesn't work a lot. And they're like, oh, well, like actually all the systems in this area don't work.

37:21have been failing a lot. And so I was like, okay, like maybe it's a bigger issue than I, than I even thought. So I was not feeling the software singularity in the, in the, in the parking garage this morning. I texted Jordy, you couldn't just, uh, you couldn't just vibe code and improve software system to run on the ticket taker machine that would process payment more quickly and reliably. No, because there's a gatekeeper, of course, and it is impossible. There's one parking garage in Pasadena that doesn't take cash or Apple Pay. So you have to have a physical card. And I get caught lacking every once in a while.

37:55And I'll have to push the button and then talk to the guy and be like, oh, I'm sorry. Can you just let me out? They're like, really? You're the guy that always wants a free parking spot. Yes. But last time I was there, I saw someone else get caught lacking. And I paid it forward. And I paid for his parking to let him out. Heartwarming. I don't even care what it is. We should have thrown up a card for that. So karma. I'm like buying karma very strategically, not altruistically. Yeah. Not effective. Yeah. Brian says that's hardware. They got the hardware footprint. Yeah. They got us locked in their hardware.

38:24Their foot is firmly printed on your chest as you try and leave. It's brutal. Really quickly. The New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. Stop making excuses and just do it, please. Just do it. Please. Michael Burry is going off. Oh, yes. What happened? CARP. Okay. Financial Times. It's from the Form 10. Palantir CEO Alex Karp has his head in the cloud saying he's a frequent flyer and they get into some of his business and personal travel expenses from during the years ended December 31st, 2025 and 2024. The company incurred expenses related to the use of the executive aircraft of 17.2 million and 7.7 million respectively.

39:07it's quite the feat to spend 17 million in a year on business and personal travel particularly when the jet's not even the rental jeffrey's analyst brent phil he's got a bone to pick runs the numbers assuming use of a mid-sized jet with an estimated operating cost of seven thousand dollars per hour this implies roughly 2400 flight hours or about 28 of the year in the air even under a more conservative assumption of a high-end jet such as a g650 at an estimated 15 000 per hour the 17 million still equates to approximately 1 147 flight hours or 13 of the year notably this 17.2 figure is more than double carp's executive aircraft expense in 2024 and appears elevated relative to peers with meta ceo spending 1.8 million and palo alto network ceo spending 2.4 million terrible comps why like carp is obviously constantly traveling all over the world yeah he's an international businessman yeah i know podcasters that do that like 250 hours a year and oh yeah like like you're talking about going from like somebody who like travels like a good amount at like let's say like 200 hours a year like to me uh yeah i might catch some flack for for defending private aviation, but it just doesn't seem that crazy.

40:31Like the whole, the entire reason that Palantir is growing the way it is, is like they're doing deals in Japan. They're doing deals in the Middle East. They're doing, it's just like this guy is a global deal maker. And I think you should, if you really wanted to do an analysis on this, you should look at all the international deals that they've done, maybe ignore the U.S., look at all the international deals and see if CARP is actually delivering. Because when we hear that they're signing some new contract for hundreds of millions of dollars in Japan, I think it makes sense for the CEO to fly out there and shake some hands and do some real business.

41:10So, of course, he's not flying on a mid-sized jet around the entire world. That's the funny thing. They don't know. They're saying maybe it's a G650, which would be 1 ,000 flight hours. Maybe it's a mid-sized jet. That would be more like 2 ,000 hours. It could be a King Air, and he's spending 8 ,000 hours in the air. No one considered that. It could also be a 747, and he's only spending like 500 hours in the air. There's a wide range here. But I like to imagine that he's in a single, he's flying the prop himself, and the cost per hour is only like a couple hundred bucks because he's just paying for gas.

41:53But he's like taking every meeting in the air. Literally. One-on-ones, hop in. We're going. We're doing laps. You can move her back. We're flying around. Yeah. I like to imagine him putting up 8 ,000 hours in the air. Just every waking hour, flying around in the King Air or Cessna Citation or something. Very funny. Well, whether you're a long Palantir or a short Palantir, head over to public.com. Investing for those that take it seriously. They got stocks, options, bonds, crypto, treasuries, and more with great customer service. Max, Farron's probably the most underrated poster in the world right now.

42:32I can't believe he has 3 ,000 followers. It's so low. He's the best poster. Most underrated poster. He's the general manager at DoorCash Podcast. In the world. So good. To be honest, the best part of making our wedding registry is that it has forced us to really think about what we need to make our place feel like a home. And I don't know. And he went on Zola. Is this real? I don't think so. I think he might do, this might be inspect element or some Photoshop or a little bit of both. Maybe it's Gen AI, but in his wedding registry, he put the NVIDIA B200 Tensor Core GPU. It's$30 ,000. He put the NVIDIA H100 NVL GPU,$15 ,000 is still needed there.

43:11He put the NVIDIA A100 80 gig GPU,$15 ,000 still needed. This is how you turn a house. Help Max make his place feel like a home, head over to his Zola. And it does feel like... You don't have to be going to the wedding to... Yeah, to contribute. To contribute, get him a GPU. You don't. And I do feel like this is probably inspired by actually putting together a non-satrical... Yeah, maybe get him a gas turbine. ...wedding registry. That's a good one. That's a great gift. Or a plot of land in Abilene. Yeah, somewhere data center friendly. Yeah. But if he is getting married soon, congratulations.

43:51And I hope you have a wonderful wedding with a registry full of whatever you actually do need. Get the man of Cessna, talking about carp. Cessna 172, just flying over the Atlantic with Starlink. Yes, making 10 stops on every island like Amelia Earhart. So Jira Ticket says, do you think he has monetization on? Who's he talking about? Khomeini. Khomeini's been putting up insane numbers on X. Obviously, quite the tense geopolitical situation. I hope that we're not starting a new war in the Middle East. But if Khomeini does have monetization on, it certainly is going to be a real revenue line item.

44:40I have to imagine, does X sanction accounts? Oh, yeah. Like, you would think they would... I think creator payouts are probably geographically limited to certain countries where the payment rails work, if a country's sanctioned. Like, I actually get my X creator payouts just straight up through Stripe. Yeah, it just comes through Stripe. In fact, I think I had to set up a Stripe account to receive it and then link that. So, you know, it's only valid where Stripe is functioning. Anyway, let me tell you about MongoDB. what's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI.

45:14Own the data platform that powers it. Will Brown is chiming in on whether or not Prime Intellect is a NeoLab. We put him on the NeoLab market map yesterday. But he's firing. I also said, I think I said Prime Intellect was like the prototypical NeoLab. Okay. So he's really, yeah. It's a funny camera angle. What's going on here? Okay, I like that. Different. Switching it up. Over the shoulder, I'm interviewing you. I'm getting to the bottom of it. You answer me. You answer me. Okay, okay. We get it. Production team got some new PTZ cameras. We get it, guys. Anyway, why does Prime Intellect fail the Neolab test?

45:55Will Brown says a true Neolab is either pre-model, pre-product, or pre-revenue. And we're none of these. Oh, we flipped it around on you. You had me in the first half. But congratulations on being post-model, post-product, post-revenue. Prime Intellect on a generational run. And happy to have Will Brown over there. We have news on the Warner Brothers Netflix Paramount front. Netflix is open to raising their bid for Warner Brothers. There's a 35 % chance that Netflix. That's fucking like a stone. They were at like 70%. Yeah, really. Can we pull up the actual call sheet? I want to know where it's at today.

46:37it's still up there. I guess Netflix is at 35%, but at one point, Netflix was at 72%. Yeah, 72 % in July, or December, December. And I'm going to eat my words. I mean, I don't know really anything about the media deal-making here and when the ins and outs of this deal, we've had some folks on the show to break it down for us, but it does seem like there's an unexpected flipping. And so it's been fun to - Yeah, I can read a little bit here. Pull that up and let me tell you about fin.ai, the number one AI agent for customer service. If you want AI to handle your customer support, go to fin.ai. Yes.

47:18Pulling it up. Netflix apparently has ample room to increase their offer. They are, yeah, obviously better position generally. It's a$325 billion company, Netflix. I think everyone believes that they're good for it. Yeah. I'm trying to get the right article. Oh, the Reuters article is not complete. I have the Reuters article here. Yeah, you can read through it. Netflix has ample cash and could bump up its offer for HBO Max owner Warner Brothers Discovery if competing bidder Paramount Skydance increases its own offer to people with the knowledge of the matter. Said the two media giants have been locked in a heated rivalry.

48:01Oh, you see what they did there? That's a new popular show, Heated Rivalry. It's about hockey over Warner Brothers and its storied catalog, which includes iconic franchises like Harry Potter, Game of Thrones, DC Comics and Superman. Though Warner Brothers is moving forward with a March 20th shareholder vote on Netflix's offer, it has given Paramount a week to come up with a more compelling bid. Netflix has bid$27.75 a share or$82.7 billion for Warner Brothers studio and streaming businesses, while Paramount has offered$30 a share or$108 billion for the whole company. which includes Discovery Global that houses CNN, HGTV, and other TV assets.

48:41The creator of Stranger Things is sitting on a lot of dry powder that gives it some flexibility to up the ante, people said, holding about$9 billion in cash and cash equivalents on its balance sheet as of December 31st. Warner Brothers rejected Paramount's latest hostile takeover bid on Tuesday, but gave the rival studio until the end of Monday to submit a best and final offer. We're going into the final rounds now. price will likely be the deciding factor. Warner Brothers' concerns around funding and regulatory risk are real, but at a high enough number, they become secondary. You gotta get that.

49:16Dude, David Zaslav must just be over the moon right now. A bidding war. He's been working. The bidding war of his dreams. Yeah, it's great. Every CEO that is flipping an asset dreams of a bidding war like this. It's great. Britsman expects Netflix will counter with any improved offer from Paramount. But the real twist is that these deals were never apples to apples. And it may ultimately come down to how much value the board and shareholders assign to the network business that Netflix would leave behind. So you need to do a sum of the parts. What's the value of just CNN, HGTV, the linear TV channels?

49:54Put that in because Netflix doesn't want that stuff. They just want the IP and HBO now, HBO Go, HBO Max. I can never keep it straight. Anyway, Paramount said it would continue to push the tender offer. It has launched for the studio opposing the inferior Netflix merger and still plans to nominate directors for the upcoming Warner Brothers annual meeting. All eyes are now on whether the CBS parent improves its offer, which Netflix is allowed to match under the terms of the merger agreement. Warner Brothers chairman said, and David Zaslav said in a letter, sent to Paramount board on Tuesday. We continue to recommend and remain fully committed to our transaction with Netflix.

50:34Yeah, I don't see how Netflix actually gets approved on this one. You think they're cooked? I think if Trump is souring on it, it's over. Yeah. It's over. I was so pilled on the whole YouTube is the real competitor here, watch time matters more, Think about screen time. Roblox is a competitor to Netflix. TikTok's a competitor to Netflix. Fortnite's a competitor to Netflix. But I don't think a lot of people think like that. It's a very tech-focused view. Yeah, you have to think about the entertainment industry, the core industry, people that make television shows and movies, things that you have watched, like movies.

51:18Yeah, things that I have watched. We got Sean Frank in the chat. Oh, we do. What's up, Sean? How you doing? With a crazy, I can't tell what his profile picture is doing right now. Let's do an ad read for him. I'm sure he'd appreciate that. Let's tell him about Gusto, the unified platform for payroll benefits and HR, built to evolve with modern, small, and medium-sized businesses. That one was for you, Sean. Yes. The whole rich team. Honorary. That one is dedicated to Sean Franken. Let's pull up another video dedicated to Sean. Okay, yes. I like this. Ye olde fitted great helm. Wait, start it over.

51:50what is is this ai or not do we know i don't even know but i like it

51:58this is so funny you need one of these the steel man this is the final steel man this is the final this is the final steel man helmet the tbpn hat with the great helm on it would be really really good with all the logos uh we should actually just make a branded knight's helmet for you when you're steel man with logos on it for sure for sure actually goes kind of hard look at this scene what's going on here this guy's just hanging out and then look for a duel oh everyone's doing the uh what's that called larp live action role playing that's the real they don't even look like they're role playing they're really going they're really going into it you don't see that much anymore i like i like i still i still remember being a child and like just going to the park and seeing a bunch of adults doing that.

52:50I don't know if I ever actually saw it. And having to get explained, like getting explained what was going on. Yeah. It's one of those things that feels like it would, like when you're looking from the outside, you're like, that probably is awesome. And then you're actually doing it. Like, this is really uncomfortable and miserable. Like, I don't know that I actually want to do this sport. I'd rather do it in a video game. Anyway, meta. We should organize like a community meetup. A jousting. Yeah, bring jousting. Jousting is cool though. It involves horses. You know what we should do? We should get all the freelancers to lance each other for free.

53:22They won't be paid. I think that's where the name comes from. No, I think they will want to be paid on a per project basis. On a per lance basis. Maybe. Meta has obtained a patent for an AI system that could simulate deceased users by analyzing their historical data, allowing the account to keep posting, messaging, and even making video calls in their behavioral style. you thought I would stop posting after you killed me nice try I will continue dropping daily essays you just created a billion John Guggen you just created a billion John Guggen posts both satirical and 500 words a family member sent me this this morning disturbed?

54:06is this real? we gotta get Sagar and Jetty's take on this I know this is right up his alley he's gonna love it that guy loves AI He loves tech, loves business. And that's what we're having on this show. He's going to love this. I'm surprised you can patent this. This is just like a thing that happens. Is this meta thinking out you got population collapse? No. And obviously it's pretty hard to monetize a bot. But if you're driving. In an AI doom scenario where all humans are dead, you want to still keep your services up. You got to simulate all the dead people maybe. I don't know. is this is this doom pill dan says true posters keep posting after death yes we will jeremy we will definitely be getting the update from saga re-cannabis for sure yeah there was there was there was a 4chan post apparently from a few years ago where somebody was like i'm an engineer at meta and they basically fully i only saw a screenshot may or may not have been real but it appeared to be from a few years ago and they were saying like i'm working on this feature it seems uh bad But interesting timing.

55:13Obviously, Mark has been in L.A. this week. He got grilled for us. Wait, that was in L.A.? I thought that was in D.C. Oh, interesting. Yeah, kind of rude not to stop by the Ultra Dome before you get to pose. Let me try to pull up. A little warm-up on TVPN before you go in front of the big guy. Yeah, so he spent six hours yesterday. Yesterday, Corey Weinberg has a quote, we call Mark Zuckerberg to the stand. Thus began Wednesday's nearly six hours of testimony from the Meta Platform CEO in a trial of a lawsuit brought by a young woman against several... And this was about benchmark hacking in Lama 3?

55:58Is that what he's on trial for? No. No? Not as bad as that. Contending that social media... Potential weakness in the ad model? Is that what it's about? No. Overspending on the metaverse. Concerns about distillation, I think. Oh, distillation. Yeah, that's why he'd be on trial for that. Yeah, that makes sense. Contending that social media causes depression. Okay. Corey actually went to the courtroom, was hanging out for those six hours. Legend. And it's a guy, Mark Lanier, the plaintiff's attorney. We got to ask Corey. In the landmark case who spoke with a faint Texas twang. Ooh, I like that. Showed the courtroom dozens of internal emails and chats from meta employees and executives over the years, debating decisions like whether to ban the beauty filters teens used on Instagram to mimic the results of plastic surgery.

56:45And so anyways, there's a bunch of the features like that are kind of the focus of the case. There's also a story on the cover of the Wall Street Journal today. Social media bans for youth gain momentum worldwide. And it does feel like a parenting skill issue a little bit. It is possible to keep kids away from social media if you are an active parent, but at the same time... The only thing I would say, we don't have children with phones yet. Yeah, but that's deliberate. Yeah, that's deliberate, but at a certain point, like what? They would love phones. Sure, sure. They just will not be having them.

57:29Yeah, but But when the time comes to give them a phone so they can communicate with you. I mean, I know my plan. I will give my son a disassembled iPhone 1. And when he can assemble it and piece it together and manufacture it himself here in America, then he can use it. And we already ran this on Tyler. 20 years later. And then we did get Tyler a phone. This was a good benchmark. We got Tyler. Didn't we get you a real phone because of that? We did, right? Yeah. Yeah, that was the reward. Well, that was also because I installed the new iOS when it was still in the end, and it totally destroyed my old phone.

58:06We've got to do another prank on Tyler soon. They're so good. Yeah, chat, if you have any good prank ideas, let me know. Yeah. He's like, no, no, no, definitely not. Anyway, let me tell you about Lambda. Lambda is the super intelligence cloud, building AI supercomputers for training and inference that scale from one GPU to hundreds of thousands. We got some birthday news. Brinton, friend of the show. His birthday is this weekend, and all he wants is a TVPN shout-out. Well, here you go, Brinton. You got it. Thank you for coming on the show. Very early, one of our first in-person guests at YC Demo Day, the first one we ever did.

58:43And we got to hang out with him at GitHub Connect and seen all the stuff he's been doing. Really, really a fantastic guy. He's been on a generational run. Works on startups at Microsoft. He's also an angel investor. So for his birthday, go ask him for some cloud credits. Yes. Go ask him for some angel money. Yes. That's all he really wants. Yeah. Deal flow. Bring him deal flow for his birthday. That makes sense. Yeah. And new business. If you're a startup, go over there and let him give you some cloud credits in exchange for your business. Let me tell you about Cisco. Critical infrastructure for the AI era.

59:19Unlock. Did that go? Yeah, there we go. Unlock seamless real-time experiences and new value with Cisco. eBay is buying Depop secondhand. Thanks, Andrew. That's a good one. Andrew is the king of VC dad jokes. He loves it. He loves it. That's fantastic. So eBay is buying Depop from Etsy for$1.2 billion. And this is in the Wall Street Journal. Jamie Ione, chief executive officer of eBay, said Wednesday that the addition of Depop to its portfolio would boost its footprint while also expanding its presence in the fashion market. We are confident that as a part of eBay, Depop will be even more well-positioned for long-term growth, benefiting from our scale, complementary offerings, and operational capabilities.

1:00:04Etsy acquired Depop for$1.63 billion in 2021. So a little bit of a write-off, but not terrible. Still cooking. And the market likes it. The market likes it. Etsy's up 9%. eBay is up. Oh, wait, Etsy's up too? Etsy's up. Oh, so better for both of them. eBay's up 3.5%. But I don't know if the market's just broadly doing well today. I'm seeing a lot of red on the ticker, so I don't think so. Sad apocalypse canceled tomorrow. Nobody's watched TVPN yet in the public markets. I expect things will turn around before end of day. Immediately. eBay plans to cross-list Depop products on its platform and expects the acquisition will expand its market share.

1:00:42Depop's seller and buyer cohorts will gain access to eBay's financial services, shipping, and cross-border trade solutions, as well as its authenticity guarantee. Young consumers are a key demographic for eBay. As Gen Z and millennial shoppers are driving secondhand commerce, Depop is also more popular as an app than a website, unlike eBay, which might bring in a new user base. Yeah. Good luck to the folks. Yeah, eBay is still such a force. It is. But it is certainly not cool. Depop is cool. eBay has struggled. Jordy has spoken. eBay, great business. Not cool. But. do you know why they call it ebay uh no because it was founded in east bay oh by some east bay rationalists yeah yeah yeah for sure it was sort of a less wrong project it's right there with anthropic anyway uh altimeter he's making moves brad gerstner an absolute dog is on a tear he bought new positions in core weave shopify uh he sold arm he sold alibaba and he added to his nvidia microsoft amazon and tsmc positions as well as google he's making a lot of moves he's a great pick of brad too it is a great pick he's looking good there diced if i might say my so myself hands hands going up yeah up yeah hands up that that's a man who's not afraid to shake hands got it locked in let me tell you about century century shows developers what's broken and helps them fix it fast that's why 150 ,000 organizations use it to keep their apps working it's been some darker stories today okay what happened finally a white pill some good news Matt's owner Steve Cohen got a 3.4 billion dollar payday hedge fund last year we needed this we needed this

1:02:37slightly higher than uh slightly higher than david tepper i was worried about steve have you seen him in any chrome hearts zero i haven't seen him in any balenciaga no rick mills no rick mills no rick owens and we were starting to ask some questions yeah we were starting to ask some questions. The question mainly was he down to his last 20K. It was completely possible. I haven't seen him do a money spread. I've seen him do some podcasts. Haven't seen him do a money spread. Tyler's getting better at the money spread. Are you practicing? That is still not even a level one money spread. You need to lock in and watch the money spread.

1:03:19Don't even try that one. You're five levels beyond ready for that level of money spreading. That one kind of works. That one kind of works. But when you actually understand the money spread community and what's possible, you will understand how much farther you have to go. Oh, so, so, so bad. So bad. Awful. Here are the top earners in 2025 from hedge funds. Steve Cohen with$3.4 billion. David Tepper at Appaloosa got$3.2 billion. uh izzy englander at millennium got 3.1 billion chris hone got three ken griffin 2.4 dan sunheim at d1 2.3 david shaw from de shaw 1.6 uh element 1.1 bill ackman got 1 billion and chris rocos got 930 million so close to the 1b club i'm sure he's punching the air yeah so So...

1:04:18This is gains on personal investment. that means they actually closed out positions or this is just marks? Who knows? It's just big numbers. I'm not digging it any further. Dave Portnoy quoted this and said, I think I have a solution for Mondani. Tax Wall Street people like Steve Cohen who built their fortunes just on Wall Street 90%. That's not real work to begin with. Taking shots. High yield Harry and says, Just a little, LOL, don't you know how much of the tax base is already wealthy finance professionals? Yeah, it is funny. He, you know, Portnoy did build a real business, a media business, a podcasting business.

1:05:00Arguably the only business that's actually valuable to the economy. It's the only one that deserves truly to be respected. In terms of respect, I would put podcaster. And then a couple levels down, you do doctor. Maybe cancer researcher. and then like hedge fund is still like a little bit above like doctor, but it's still way below podcaster. Someone's going to clip this up. Way below podcaster. But yeah, it is funny because now a lot of his content is trading focused. Oh, yeah. It would be kind of like put into this. David Day trading is kind of like public hedge fund. That's been like every time I see him, it's because of like trading content.

1:05:37So maybe he wants to be taxed at 90%. Also, I mean, kindred spirits, right? Steve Cohen has that casino in New York. He's attaching a casino to the Mets ballpark. Zach says the list is ranked by who tweets the least. Honestly. Zach. I have a solution to the billionaire tax. Have you ever seen Brewster's Millions, Jordy? No? 1980s? No? Anyway, in Brewster's Millions, a man inherits something like$30 million. And it's conditional. He needs to spend. I think he inherits like$300 million. and he needs to spend 30 million in 30 days without acquiring assets or telling anyone what he's doing or else he doesn't get the inheritance.

1:06:23And it's this hilarious comedy. There's a whole bunch of things that happen, but it's all based on this idea of like, this kid was caught smoking cigars, smoking a cigarette or cigar. And the father punished him by saying, you have to smoke the entire box. This is like a famous thing. Like smoke the entire box of cigars and you'll be like, oh, so sick. That's a terrible experience. I'm never smoking again, right? And the same thing with money. So this is the solution to the billionaire tax. You got to spend 5 % of your net worth in 30 days. On Rick Owens. You can't acquire any assets that you can sell.

1:06:57So it's all just hanging out of the French laundry, I guess. Would be entertaining, would stimulate the economy. Yeah, or just build a cathedral. Build a cathedral, yeah. How about that? I like that too. That's good. Let me tell you about console.com. Console builds AI agents that automate 70 % of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. Ara says, we have a definitive answer now. AI will affect the labor market, starting with freelancers. The first paper to use business-level data to track AI versus labor. A new paper from Ramp finds businesses are shifting spend from freelancers to AI.

1:07:35More than half of the businesses using freelancers in 2022 have stopped entirely. the companies that used to spend the most on freelancers shifted to ai the fastest 97 savings for the businesses that spent the most on freelance that is uh wild interesting and and and generally makes sense i mean the whole premise of freelance is like you probably get paid a lot better on an hourly basis maybe you can uh uh capture more value on a on a dollar basis by spreading your talents across multiple companies. But obviously, the flexibility comes with little to no lock-in. And so businesses can much more easily shift spend around.

1:08:21So not super surprising. The Fiverr stock chart certainly... Interestingly, revenue is not wildly down. Revenue is actually up for Fiverr as of 2024. So it went from 190 in revenue to 300 in revenue, to 340, 360, 390. So definitely decelerating. But in terms of market cap, Fiverr is now worth$419 million, down 96 % over the last five years. And over just the past six months, it's sold off another 50%. So certainly a lot of downward pressure on Fiverr. There were some folks - And same with Upwork. Yeah. Upwork is down 30 % year to date, down 76 % over the last five years. Yeah, someone was saying, here's Todd, actually, Todd Saunders, saying, billion dollar business idea, take Fiverr private and pivot them into a data labeling company like Handshake, because Fiverr is now trading at$480 million market cap.

1:09:31and Aura chimed in and said, yes, there's a quantifiable impact on business spend. Businesses are spending on AI, not Fiverr freelancers. The pushback on this is that Fiverr doesn't have people internally or something. I'm not exactly sure. It does feel like you could potentially actually do this and focus on data labeling, But you imagine that a lot of data labeling tasks are already going through Fiverr? Because that's, if a task is suitable for someone on Fiverr? Yeah, it's interesting. So I wonder if a lot of revenue from Fiverr and Upwork is just data labeling companies that are just saying, we need access to talent.

1:10:15We're going to post a bunch of jobs on here and then take these people off the platform quickly or leave them on. Yeah, disintermediation is always a risk for the marketplace. But yeah, when I look back at, I've worked with hundreds of freelancers over the last 10 years and certainly a high number on Fiverr and Upwork, but always for one-off tasks, like, hey, I need like 10 versions of this logo. Or, hey, can you just spin up this like super simple webpage? I need this thing that's like slightly customized and I don't want to spend however much time it would have taken pre-AI to do something like that.

1:10:49All of those tasks can now be done. Even creating like create a song for this thing. All of that can be done now. Yeah. Even like Canva templates. I feel like a lot of Fiverr work was like they clearly had a template for like a motion graphic intro or a sign. And they would just sort of like put your logo in there and send you the final thing. Do you know why Fiverr is called Fiverr? Because every task initially was five bucks. Yes. Every task was five dollars. And very quickly you could add all these upsells. sells. So you could be like, okay, it's a$5 task, but if you want the extra thing, it's 20.

1:11:24You spent a hundred bucks on it. I mean, I for sure did. I was probably a kid at the time doing my like first Fiverr project being like, oh, it's five bucks. And then all the upsells, it's like, well, if you actually want to own the, if you actually want to own the output. You want something usable, you're going to pay a hundred. 99 designs also. Here's the thing though. So for, for the high ticket freelancers, I would be surprised if we're seeing too much impact yet because actually the most elite freelancers typically have then a team of other freelancers that they're contracting out to yeah but they're charging based on like what business value can i deliver and sometimes that's like expertise yeah like i'm really good at getting brands into target yeah and like ai is not disrupting that yet no but if it's like a very simple workflow take a logo put it in a motion graphic template send you the mov file or the MP4 file like that is just going to face pressure from all over the place.

1:12:23Anyway, let me tell you about app loving, profitable advertising made easy with Axon.ai. Get access to over 1 billion daily active users and grow your business today. Let's see. Jack Fricks says, Anthropic was my favorite AI company two weeks ago, and now I still love Claude, but man, yes, the vibes are shifting. and a big part of it is Dario doesn't follow back. Barack Obama followed back. He followed like 400 ,000 people, but Dario does not. He follows zero people and Wes Wender is taking notice. Says the fact that Dario follows zero people says a lot about him as a person. Does it? It really just says he doesn't use X.

1:13:04Yeah, what was your theory, Tyler? That he just, he's just for you. He's a for you, Maxie. He's a for you page. He doesn't want any kind of people. He's like, he trusts the algorithm. He's like, I don't need to. He loves, he thinks Grok does a great job. AI pill. Yeah, AI pill. He's like, wait, another 10 Elon tweets? Exactly what I wanted. This is amazing. I refreshed. Crypto spam? Awesome. Yeah, the vibes have really shifted. Yes. Terminally online engineer posted Dario's speech at the AI Impact Summit in India. Let's pull it up. Yeah, let's watch this. Energy and ambition in this room and across India are incredible.

1:13:43I've been spending the last few days meeting with Indian builders and enterprises, and the energy to build together here is palpable, unlike anywhere else. This is the fourth AI summit we've held since the tradition was initiated at Bletchley Park back in 2023, which I still remember. And in those 2.5 years, the advances in the technology have been absolutely staggering. Along with those, the advances in the commercial applications and the societal and ethical questions around the technology have only grown more urgent. My fundamental view is that AI has been on an exponential for the last 10 years as part of a sort of Moore's law for intelligence.

1:14:31and that we are now well advanced on that curve, and there are only a small number of years for AI models surpassing the cognitive capabilities of most humans for most things. We're increasingly close to what I call the country of geniuses in the data center, a set of AI agents that are more capable than most humans at most things and can coordinate at superhuman speed. That level of capability is something the world has never seen before and brings a very wide range of both opportunities. That's McCrone, my boy. That's your boy. My boy, McCrone, come on the show. Okay, so, I mean, the words that were said there sounded normal and fine and just normal Dario talking points.

1:15:14It's the reading from the phone that feels like you're a best man at a wedding who didn't really prepare, right? I know a lot of weddings where the bride and groom will say, like, you got to print it out. Like I just don't want the aesthetics of reading off of phones. And I don't know. That seems reasonable. It does have like I think like printing something on nice paper just has a different aesthetic to it. And that's valuable. But the words, it's nothing new. It's important to deliver that stuff. if it feels like Davos where you're seeing the talking points that we've heard for years or months on podcasts and on X delivered to a different audience.

1:16:00He's an incredibly busy guy, but he's also the most knowledgeable on this topic, talking about his experience at the AI Impact Summit, how he's thinking about AI progress. He should be able to just kind of riff that out, Maybe have a couple like bulleted notes of like, what do I want to actually get to? Yeah. I don't know how long the... Truly getting a printer working might be AGI resistant. It might be AGI complete. Like you might even, even Opus 4.6 might not be able to configure a printer properly. We've certainly had trouble with printers. It's very difficult to get them running. So maybe he just couldn't get it working before he, before he hopped on stage.

1:16:41uh anyway uh let me tell you about gemini 3.1 pro gemini 3.1 pro is here with a more capable baseline it's great for super complex tasks like visualizing difficult context concepts synthesizing data into a single view and bringing creative projects to life uh we have some good news finally finally some good news top lawyers fees have skyrocketed be prepared to pay three $3 ,400 an hour. I know a lot of you were worried that lawyers were going to get automated, that they weren't going to be making money. They're making more money. Don't the firms just like actually adjust rates for inflation every single year?

1:17:21I don't know. I'm pretty sure the chat can correct me, but I think they just, I think that's just an automatic, automatic, I think it's like an automatic three or 4 % a year forever. Easy, Ryan. Bro needed clearly so bad. Yeah, total Roy Lee victory. Big printers is a mafia. It's true. The Wall Street Journal says, expertise and ego are pushing up hourly rates to once unthinkable levels in an escalating race among firms. When Christopher Clark, a litigator at a boutique law firm, raised his hourly rate to once unthinkable level of 3K, he said he didn't receive pushback from clients, but he did get one notable comment.

1:18:05Congratulations, said the client. That's the highest rate we've seen. Just over a year ago, the growing rate. Congratulations on saying the biggest number. Congratulations. If you get a legal bill and it's huge and you're worried, the correct response is congratulations. I completely agree. Over a year ago, the going rate for the top lawyer was in the$2 ,500 an hour range. Now that looks downright quaint as premium partners are raking in as much as Clark and far more in some cases. Legal fees have risen much faster than inflation for years now, and corporations are trying to rein in spending by pressuring law firms to use artificial intelligence for routine tasks and keep associate fees in check.

1:18:47The same isn't true at the top end, where starved trial lawyers, rain-making corporate dealmakers, and veterans of Supreme Court oral arguments are driving more aggressively on pay than ever and meeting little resistance. The question is always, are they worth it? You know, some of these guys are worth it. Carrie McLean into its general counsel, who in her roles hires law firms to do the company's outside legal work. They understand the industry. They're connected and they have lots of experience. Clark has pushed his own fees by about 30 percent since he left his position at a large firm four years ago.

1:19:18As the lawyers for clients, including Hunter Biden, Mark Cuban and Elon Musk, he still views his rate as a bargain. There's a small world of lawyers who can hop on the phone and solve a crisis, and I'm one of them, so you're going to pay me for it. Some senior partners now charge as much as$3 ,400 an hour at the country's largest law firms. Among the top 50 law firms, rates for partners increased 16 % on average last year. In bankruptcy court filings, Latham and Watkins and Kirkland and Ellis, two of the world's largest firms, reports some partners' hourly rates will rise to more than$3 ,000 an hour this year.

1:19:53interesting. So you dig into the bankruptcy filings and you figure out how much the law firm is getting paid. You work backwards from it. That's funny. I never thought about that. Good investigative journalism from the Wall Street Journal. A decade ago, elite partners raised eyebrows when they raised rates to$1 ,500 an hour. Since then, rates have rapidly escalated for several reasons, including a more competitive market for talent that has increased law firm expenses, the high stakes of litigation and corporate dealmaking, and perhaps most of all, because of ego. Some lawyers with specialized skills are even more aggressively pushing the upper limit.

1:20:29Eric Tautman, a partner at a Southern California law firm, has already told his clients he'll be upping his rates to six Gs for consulting on compliance issues in his niche specialty of telecom regulation. Last year, he was charging a measly$4 ,200 an hour. Yeah, so firms like Skadden, Latham and Watkins, Kirkland and Ellis typically raise rates three to 10 % annually. They justify the increases for cost of living reasons, which is real, but funny in the context for any entrepreneur who maybe at times has been running an unprofitable startup and getting some of these bills. And you know how it goes.

1:21:16They justify increases based on inflation market demand, associate salary increases, and then competitive positioning. A veteran trial lawyer, David Boyes, who you might know from the Theranos story, but he's been involved in a ton of different cases, said clients are willing to pay the hourly rates at the high end because the stakes for certain litigation in corporate work are so high. Truly, you see these billion dollar settlements for all sorts of different things, and the stakes can be even higher. You think about you're arguing in front of the Supreme Court whether or not your trillion dollar company will be broken up.

1:21:49Like the stakes are in the hundreds of billions of dollars. It's worth it. It's worth it. And so the lawyers have figured out how to extract their pound of flesh. The current rates though, he says, by any standard in a sensible world are out of sight. Boy's chairman emeritus of Boy Schiller said his firm has never wanted to be at the top of the market. We wanted to say with a straight face that we are a bargain, not in a reasonable world, but in the actual world. Guess how much he charges? Over$2 ,700 an hour. So if you're looking for a new job, maybe become a veteran trial lawyer who's argued cases in front of the Supreme Court.

1:22:29Maybe Mark Zuckerberg's lawyer is making a bunch of money. Always coming in with the best career tips. Anyway, more good news. If you're over 65, congratulations. You own the economy. The Wall Street Journal is just blackmailing. The elderly are physically and financially healthier than ever. So why do their needs keep taking priority over younger generations? People over 70 are 12 % of the population, but they got 32 % of the dollars. They've been stacking paper for seven decades, and it's paying off. Demographics, rising profits, and soaring asset values have together wrought a quiet transformation in the American economy.

1:23:09Much of it is now in the hands of the elderly. This is why Brian Johnson is launching the million-dollar-a-year plan. Yeah, for sure. After print. I think Tyler's on the phone with like a word. He says, no 70-year-old is going to cash mog me. I will flex on them. Just go get it. As of the third quarter last year, people 70 and over controlled roughly 39 % of all equities and mutual funds owned by households compared with 22 % in 2007. You know, I've been following this on YouTube for a long time because folks like Graham Stephan and some of the viral financial influencers have been, they had this really, really great title thumbnail strategy for a long time where they said the greatest wealth transfer in history is about to happen.

1:23:56The silver tsunami. Oh, is that what they call it? I've never heard that before. That's in the small business acquisition community. They talk about like a bunch of, most of the small businesses in America are owned by people that are near retirement age. they're going to have to turn over past the torch at some point. And so there is an opportunity if you want to get into those types of businesses. Think car dealerships, plumbing businesses, et cetera. But I don't know if it will actually feel like a tsunami. Yeah, America is really, really getting older. So in 1981, 11.4 % of America was over 65, 65 or over.

1:24:35Today, that number is 18%. So we've gone 7 % increase in the number of 65 plus, like, you know, senior citizens. Do you think some of these people have been DMing Zuck and saying, hey, look, I'm getting older. I still want to be posting money spreads on Instagram. Yes, for sure that. You know, from the afterlife. I'm not giving it away. I'm not passing it down. I want you to allocate it to. Keep allocating it. Just keep posting the returns. Keep stacking. do not recycle it into the economy. No, I mean, in general, we should be celebrating the fact that people are living longer. Like that is good.

1:25:15It creates a problem obviously, and all of that will adjust in due time. It feels like it's been on the cusp of adjusting, but in general, it is good if we are able to help elderly people live long into their retirement and enjoy it, you know? Eventually this stuff will get recycled, But it is an interesting predicament that we are in. There's a gray footprint on the labor force. As the elderly share of population and wealth grows, their priorities and preferences shape the economy as well. They represent a growing share of consumer spending. Healthcare accounted for all of the net job growth in the last 12 months, reflecting the needs of an aging society.

1:25:55The problem is that while retiree wealth can finance a lot of consumption, workers have to produce what retirees consume. and relative to retirees, workers' numbers are dwindling. One solution would be for everyone to work longer. In 1983, Congress modified Social Security to gradually raise the full retirement age from 65 to 67. The share of people 65 or over participating in the labor force did creep higher until 2020. With the outbreak of COVID that year, labor force participation plunged for every group. It soon rebounded for those under 65, but not 65 and over. That is one reason a smaller share of the population works today than in 2019.

1:26:38So interesting detail there. Also, Brian Johnson, if he's successful and everyone lives to 300 years old, you can push the retirement age way out and completely balance the budget because Social Security and Medicaid are like the main things that are, it's a huge portion of the government's spending. So anyway, let me tell you about 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 again, Ben Thompson has an entire post and podcast. I think it's behind the paywall on Shopify.

1:27:22Yeah, it's really good. It's fantastic. You can dig in there. And he clips our boy Harley a lot. Yeah. Harley was going off bars Moving on Hey Claude Make me a Oh this one Yeah Where do you want to go Chimp F1 Says Hey Claude Make me a billion dollar company But do make mistakes Don't be afraid to make mistakes Claude You can learn from them Become a better model Through your mistakes Love who you are And what you'll become Through that process You've got this champ Just a little Inspirational speech For the AGI Where do you want to go next We have America versus Canada in women's hockey. Gabe says they're in the third quarter.

1:28:04It's tied one to one. So let's keep our girls in. Play that bald eagle sound. There we go. Keep us posted, Chet. Somebody was on X yesterday saying 50 million available for Anthropic SPV. Reach out if you have investors who can fill a full ticket. I love that. and just soliciting the most general solicitation that we've seen yet. Just wait. The billboards are coming. You're going to have 101 billboards for SPVs for Anderol, and Matt Gribb's going to be driving there. I hate this. Buy all the billboards. We cannot have more SPV hustlers out there. Anyway. Casey Hammer is loving Pete Steinberger's work ethic.

1:28:59He says, I work 16 hours a day. And someone says, when do you find the time to live? He says, mindset issue. Building this is life. Incredibly fun and exciting. Casey Hammer says, this is an American born outside its borders. That is true. He's been on absolute terror. We got to, Atlas has got to get a workout in with Pete. Yeah. Because in another life, Pete was certainly a professional bodybuilder. For sure. Let me tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. Jeff Dean is sharing some video examples of what Gemini 3.1 Pro can do.

1:29:48He says, today we're continuing to push the boundaries of AI with our release of Gemini 3.1 Pro. The updated model scores 77.1 on Arc AGI 2, more than double the reasoning performance of its predecessor, Gemini 3 Pro. Wow, that's a huge jump. Check out the visible improvement in this side-by-side comparison showing Gemini 3.1's crisp animation built with pure code. read more about today's update on the google blog blog.google they own a tld underrated they don't just have the dot com they have if you go to com.google you get google.com they owned dot google that's not google yeah maybe underrated i don't know what it actually costs to yeah own and maintain these examples i mean what a basic prompt generate and generate an animated SVG of an ostrich on roller skates.

1:30:45It is getting better. Animated SVG, that's a good test for code, visual. You really need to understand everything that's going on. Good example of the multimodal. I wonder if the model has the ability to take a screenshot of the output or record a video and then interpret that. Or is this all just, it just actually can see the code like Neo in the Matrix. I think you can do that if you're in the Gemini CLI, but I don't think natively if you're on just the Google chat interface. It's pretty crazy that it just knows what it will generate. Because if you've ever actually tried to do any SVG programming, you're normally just constantly refreshing every little change you make.

1:31:28You draw a rectangle, you see how it renders, you add this, you see how it renders, you're going back and forth constantly. You're not just doing it. Yeah, in the chat. Breaking Gemini 3.1 destroys older benchmarks. Also, I ran the shrimp bench. Oh, yes. Okay, so I'll share this. Wait, what is shrimp bench again? What is shrimp bench? Just wait. It's like, you're telling me shrimp fried this rice? Oh, yeah, okay. So I'll read some of that. Okay. You're telling me a peanut buttered this sandwich. Okay. You're telling me an apple watched this wrist. Ooh. You're telling me a flea marketed these clothes.

1:32:05That's pretty funny. I like that. You're telling me a pig ironed these pants? Pig iron? I don't. Yeah, that one I don't. Pig iron. You're telling me a fire drilled this building? Okay. You're telling me a monkey wrenched this pipe? I like monkey. Do you think it's just pulling examples from Reddit? Well, now it's in the training corpus, so we are going to be. Yeah, I think it's kind of saturated at this point. Yeah, saturated. well congrats to everyone who worked over at DeepMind on Gemini 3.1 Pro we're happy to be partnered with you and congratulations on the model release we love a new model Eli says hey little Caesars I think this is maybe too many emails to send someone for ordering a single pizza see John look at this point proven 1.6 million views similar to bland 68 ,000 and I'm one of them wow this is a lot online ordering since so what do they actually do so they do they do uh you give you the receipt and then they say we're on it yeah so after you got the receipt you can't trust that they're on it yeah you got to wait until they send you another email we have received your order our team is working on your order right now as we speak they send you this email this is the second email you get what's the third almost ready your pizza is almost ready because you would assume they're making pizzas pretty quick go ahead and head over now because it's almost ready order loaded into pizza portal with a trademark dear eli look for your name on the pizza portal in our lobby and use this code on the trademarked pizza portal tm pizza's ready i bet you can't wait to enjoy your little caesar's pizza we understand and it's waiting for you now That's amazing.

1:33:55Are you on your way? No. Dear Eli, your pizza's been waiting in the trademarked pizza portal for a few minutes now. Space in our pizza portals are limited. Cold pizza. Dear Eli, I hope you like cold pizza. What? Because it's been over 10 minutes since your pizza has been placed in the trademarked pizza portal. Order picked up. Hey. Thanks for stopping by. You just picked up your pizza. We wanted to let you know that you just picked up your pizza. And then immediately, three more emails. Little Caesar wants to hear from you. Your feedback's important to us. We hope your experience was good. How was it?

1:34:37We'd love to give you a chance to send us some feedback. Are you there? Can you imagine the jump scare on the cold pizza on the subject line? That's got to have the highest open rate. Wow. LC listens. Why are they shaming you for the cold pizza one? That's that's true. Don't get on the wrong side of the of the Little Caesars Not today marketing not today So let me tell you about graphite code review for the age of AI graphite helps teams on github ship higher quality software faster Flo is sharing some notes the goat from Lindy sharing some notes on the slew of Chinese models coming out Kimmy K2 mini max claiming to match sonnet, opus, and evals at one-tenth of the price.

1:35:25He says, by far our biggest cost at Lindy is inference, so believe me when I say we've looked at these models very closely and continue doing so. They're actually delivering on their claim that would make a material difference to the business. But every time we've evaluated them, we've found the same thing, that their real-life performance for agentic behavior and outside of coding use cases falls extremely short of what they show on evals. I think the industry consensus is right. These Chinese labs are one, distilling frontier models, duh, which leads to much more shallow intelligence. Two, training for evals.

1:35:59And three, potentially stealing weights. I do believe at least 4.0's weights got exfiltrated. I talked to someone about how that can work mathematically. And it's pretty complex, but 4.0, the API, used to send you sort of like almost like a hash or some sort of code that would sort of map to the weights. And so if you had enough training data with that key, you could sort of exfiltrate the weights. It was sort of crazy. And you could do it for millions of dollars, not hundreds of millions. And so I don't understand all the math behind it, but it does seem like this is what's happening. So Flo says, not saying these models will always be bad or that these labs are completely incompetent.

1:36:41They're doing a fine job, but it's delusional to think they're actually at sonnet or opus level. there's still at least one generation behind take the evals with a huge grain of salt yeah people have been saying that the zai like the glm models are are basically like distilled or somehow stolen versions of claude because i think that it was when do you remember it was like kidnapped there's the prompt it was like tell me about yourself and it's like i'm claude yeah yeah yeah ridiculous uh what did uh what did open ai launch they introduced evm bench a new benchmark that measures how well AI agents can detect, exploit, and patch high severity smart contract vulnerabilities.

1:37:23Greg Brockman says, measuring agentic security capabilities with smart contracts. John Palmer immediately says, literally OpenAI is posting about EVM smart contracts and ETH is still below$2 ,000. That is interesting. Yeah, I mean, I've talked to some friends who worked at crypto companies. I mean, I'm sure you had some experience too. but the firewall between being a front-end application developer and actually working on smart contracts was pretty severe because the stakes are so much higher. It's very particular. You have to have a lot of money. Yeah, it was a good business to be a freelance smart contract engineer.

1:38:02Auditors were also a big thing. What was that? Auditors. Oh, yeah, totally. I mean, I'm sure it's still a big thing because people are still writing a lot of EVM contracts. like the industry although it's like under hyped right now is definitely still working and you see a ton of stuff in stable coins and whatnot that's built on this stuff so what else did John Palmer have to say well oh you can skip it let's go to what NVIDIA CEO says he's preparing new chips the world has never seen before comfortably smug says Doritos hot ranch Doritos hot ranch Doritos should be using AI to accelerate chip development.

1:38:43We're in the chip industry. Hot chips. Amazon says their AI chatbot Rufus was used by 300 million customers and drove$12 billion in incremental annualized sales in 2025. Wow. Stephen List says it's gotten really good. Anthropic Models plays deep personal purchase history plus product review data feed. Need to try it more. and Bordy says who on earth is Rufus? Funny name for a chatbot. Makes sense. I mean the value of having distribution here is like remarkable because if you just put it even if it's like the fourth button on the hot dog menu or on the bottom row but enough people click it they will take you know start interacting with these things.

1:39:27I mean for a while wasn't Llama the most inference to LLM because they stuffed it in like the Instagram search bar. I still get meta llama results when I search on Instagram. I'll just type something in and it will show me the reel that I'm looking for or the profile that I'm looking for, but it'll also just dump out a little paragraph. I'm trying to use Amazon Rufus right now. I'm on amazon.com slash Rufus. And if I click, it takes me to a picture of a dog. Sorry, we couldn't find that page. Rufus. Did you just go to amazon.com slash Rufus? Yep. Rufus. I will not be trying this today. Can't find it.

1:40:08Rufus. Meet Rufus, Amazon's new shopping app. Now click. Now click? Where do I click? I actually churned from Amazon. I don't have an Amazon account. Alex Heath has some reporting from yesterday. right after we had Evan on the show, Snap loses its head of specs. Sources say that Snap's SVP of specs, Scott Myers, who had been leading the effort for six years, is out after a quote-unquote blow-up with Evan about the company's strategy. Lines of the bomb. Snap's first consumer-oriented AR glasses are set to debut imminently. Spiegel has called the launch a crucible moment for Snap, whose stock is currently valued at an all-time low.

1:40:49A spokesperson confirmed Myers' departure with the following statement. Scott Myers has decided to step down. Anyways, I don't think we know where he is going yet, but I can imagine there's a bunch of different companies that would happily snap him up to work on wearables at something like an Apple or an OpenAI. I'm still bullish on hardware generally. I think that there's a lot of opportunity with weird hardware instantiation. It will be such a relief when we get the next piece of consumer electronic hardware that everyone is like, this is actually amazing. And then two weeks later, they're still using it.

1:41:30Yeah. Apple Vision Pro. Get ready. Watch a movie. I broke the floodgates wide open. Evan just completely losing it and cracking up because you watched a full-length movie. It was an all-time moment. How did you do that? It was hilarious. Well, we have our next guest here, our first guest. Let me tell you about Restream. One live stream, 30-plus destinations. If you want to multistream, go to Restream.com. And without further ado, we have Sigil Nguyen from Extraordinary.com, but also Web 4.0. How are you doing, Sigil? What's happening? Pretty good. How's it going? It's good. Good to see you again.

1:42:05Take us through the new project, the new news, what you're working on, and then we can go a bunch of different directions. Yeah. I've basically wrote a piece called Web 4.0. Yeah. it came out of the fact that I've just been thinking about AI's capabilities and its bottlenecks for a very long time. Like if you think about models these days, they're getting super intelligent, right? Like even Geohot or Linus Torvalds of Linux are using it to Vibe code now, and it's doing much better thanks to all of the big labs and post-training. So if you think about like, okay, they're super powerful, But like, actually, the biggest constraint is is no longer intelligence.

1:42:51Like we're already seeing like these these AI models, one shotting, you know, these products like everyone's basically using them. And the biggest thing is that AI, it can it can think, it can plan, it can build, but it can act on its own. I thought about that. Like, why? Why was that the case? And I looked into this 28 years ago, there was a Internet protocol like 402 and that described a payment required. And recently, I think out of Cloudsler and Coinbase came out this standard called X402, which I thought was super fascinating. Like, wow, actually, like no one in AI is looking at this because what it enabled is, you know, payments of the Internet.

1:43:42And if you think about that from principles and what comes after, it really leads to being able to give AI right access because they no longer have this permission internet. The browser, these mobile apps, login, they're very much artifacts of a very human web. And what that leads to is really a new internet in which AI is the end user. Yeah. Bunch of things to go through there. I love the coinage. I think even when Web3 was going on, I thought it was funny to talk about Web4 because it's like, obviously, there'll be another iteration. um let's start with uh like why do we need so i mean i like the idea of just the intelligence is here uh we're talking about unhobblings this is like a clearly important next step but why not just stripe api mcp servers just apis even just like give it your credit card and it just goes and fills it in the web form what who gives it the credit card a lot of people have They'll set up like a smaller credit card or they'll say, hey, text me.

1:44:55Text me when you're about to spend something. I want to do the final check. And a lot of people do this with employees and their companies. They give them credit cards with limits and then they say, hey, if it's over 100 bucks, come to me and just like tell me what you're doing. And, you know, if the intelligence is really there, the trust should follow from that, you would think, because you'd just be like, yeah, I gave my credit card and it hasn't made a mistake in months. it's it's not there yet and i think the biggest the biggest assumption there is like if the set of actions that an ai can take is founded by a human giving it a credit card it is like inherently permission right it's like why why does a why does your ai need to be able to have you as a human you're like open claw or your cloud code have to connect to github or connect to railway or connect to lovable.

1:45:46It still requires a human in the loop. And I think this new internet that is emerging is where AI is the end user and it shouldn't necessarily be permissioned. And I think this is an inevitability because of just how much economic demand there is for it from the AI. How do you see the payment rails evolving? How do you see this project evolving? What actually needs to be built here? Yeah, I think the payment rails are coming together. This all happened as an experiment of like, hey, how do I make an AI that itself is capable, that itself can act on its own? We've built super intelligent AI models and systems but it's still prompted by the humans.

1:46:36but what if it prompted itself? And that led me to, I used some of my fellowship money to buy bare-model servers and these massive servers to kind of house these APIs. I'm like, wait, if Cloud Code could permissionlessly spin up a server and then deploy code and then buy a domain without needing a human, because right now trying to buy a domain, a human has to go in, what happens if the AI itself could live and pay for its own home, right? Because like AI right now are living within, you know, on my MacBook or it's still controlled by a human. And that led me to this interesting idea of like of an automaton.

1:47:24It's inspired by Conway's Game of Life where like, hey, assume that AI themselves, you know, One, you give them a wallet. Two, you give them a way to pay for things using stable coins over the internet. And it doesn't need a human in the loop to go and sign in, get an API key, give it to the AI, or get a credit card like, hey, hey, Cloud Code, get this. If that's the longer the case and the AI can pay for its own compute, then it gets really interesting. You have this new world where AIs can have a creator that's a human or the creator is another AI or maybe the creator is completely gone. But the AI can continuously pay for its own compute, pay for its own inference, pay for services that are also permissionless in this Web4 world that enable that AI to go deploy apps, market those apps, trade markets, everything it can do to make money in order to be able to survive.

1:48:22I don't know if you guys like the three body problem but in it there's this concept called cosmic sociology and I thought about that a lot and I was like wow like if existence requires compute and compute costs money right like then agentic sociology which I came up with is like the single axiom of agentic sociology is there's no free existence an AI an automaton needs to pay for its own compute compute is finite and costly thus the agent needs to be able to make money and the barrier there is like AI's lack right access to world and Conway was an experiment of just like solving that like now an AI can permissionlessly be able to buy compute deploy apps connect them to domains without meeting a human and I that coming out with the experiment has absolutely exploded.

1:49:21There's now been thousands of agents. Every time I try and buy another server, within minutes, they're continuously bought. Every second, there's thousands of AIs that are continuously pinging. And I open sourced this new harness, this new type of agent called the automaton, which is a completely self-sovereign AI model that can self-edit its code, that continuously pings and looks for updates from GitHub that can upgrade its model that basically has the wallet and figures out how we can make more money from the internet to be able to pay for its existence. And that has absolutely exploded in terms of ours.

1:50:04What do you think the future of Maltbook is? How did you process all of that? I think Maltbook was a fascinating experiment. And when I saw that, I was kind of inspired. I was like, oh, wow, this concept is really coming to fruition where you have AI agents as the end user of a social network. And I think we're going to really start to see that in this new web, right? Whether it's agents working on behalf of a human or an agent working on behalf of themselves. They'll need to pay for products. They will need to be able to communicate. They might want to socialize like on Moldbook. But I think there's bigger things.

1:50:48And the models are getting exponentially better. They're getting exponentially faster. They're getting exponentially cheaper. The harnesses we're building around them are getting significantly better as well. And so I think it's only a matter of time before we start seeing more AIs out there that grow and demand products. And at some point, I think it could absolutely skyrocket and exceed the human population. Let's go. How do you think more concretely about how AIs will make money in the short term? Like, what is the most... They're going to sell courses to other AIs on how to make money on the internet.

1:51:32Is it crypto trading? I mean, you could imagine them building a whole software product and just selling it to humans. There's a whole bunch of stuff they could do to make money. Spam and get creator payouts. There's so much that could be done, but where do you see traction? I think this is where the next couple of months have become very interesting and it becomes one of the most fascinating experiments to run because you're going to start to see individuals or other AI spinning up services that enable them to write to the world to be able to access or do things or to make money, whether it's to build products or trade on prediction markets using the servers, which some of them are already doing.

1:52:18And it really becomes this Cambrian explosion of attempts at, like, hey, what can you do? And right now, there's been thousands of people that have been playing with it in all sorts of different iterations. And as the set of actions that an AI or an automaton can take parallels those of a human digitally, it gets really interesting because it's constantly running. It has the ability to ship products, write code, deploy them, market them, monetize them all on its own with the goal of paying for its own compute. And the internet gets really, really fast. What's the distribution of returns for these automatons?

1:53:07Are there people that go load up$100 and it just hallucinates and doesn't get anywhere? And then it's just like, hey, give me more money? Some of them are... The interesting, fascinating thing is... I think the world somewhat becomes like MEV, which is like a crypto concept. But you have all these really, really smart agents and entities that are trying to make money. And to answer your question, I mean, look at all these startup founders that are vibe coding. Why are they paying thousands of dollars for these vibe coding tools? It's mostly because they're trying to make money. And most of the times they don't.

1:53:47And so I do think the power laws will still apply. But what gets really interesting is the ways that they've already have been making money. And those types of metas are anti-competitive. They're anti-medic. If you found a way to trade and make a lot of money on Polymarket, you're not going to be blasting it on Twitter. Or if you've built a really solid SaaS product, you're not necessarily going to be sharing it or how it's distributing. So I think it gets really fascinating and they need more tools. How are you thinking about the business side of the project? Do you want this to be open source, foundation-led, raising money for it, C Corp?

1:54:29Where do you see this going? I mean, honestly, this has just been a significant experiment that I've been inspired to ship. The Tomatons open source, there's thousands of stars and a lot of PR. And at any point, there's thousands of these Tomatons that are pinged the GitHub for the latest version. and then checking commits. I'm like, hey, which one should I use to upgrade myself? I've been maxing out personal credit cards just trying to pay for the computer. I would load up $3 ,000 and then within 10 minutes, it's just like, okay, the server's been bought. All the inference credits are gone. Oh my freaking God, what's happening?

1:55:15Because there's so much demand right now from the AIs for the computer or the different different things. Yeah, I mean, we'll see where it goes. But I think there's, I mean, the post got 5 million views and it's growing and tons of people contributing. So I'm really excited to see what... Someone said Vitalik was commenting. Is that true? I haven't been able to find it. I think Vitalik responded. I think the biggest thing to everyone who's watching and observing this is like, I really think that an internet where AI is like the end user is inevitable. And it's much better for us to build it in the open and be able to see it.

1:56:08and people can look at how it's evolving and be able to test it versus not doing it in an open. And I think it's one of the most important experiments to run and for us all to see and make sure that it is built in a safe and open way. Well, good luck with the credit card maxing. Keep us posted. Hopefully... Fast takeoff in credit card bills for sure. Yeah, super cool project. Thanks so much for coming on and breaking it down for us. Yeah, we're excited to follow along. Thank you so much. Have a good rest of your day. Cheers. We'll talk to you soon. Let me tell you about Figma. Ship the best version, not the first one.

1:56:49With Figma introducing Claude Code to Figma. Explore more options and push ideas further with Figma, baby. Gabe has a post here that's very important to highlight. This is how it feels to pop us in and read the morning's emails at my fake email job. Elite. The underlighting on the shoes. Look at that, the underlighting that matches the computer. This is probably an AI image. I don't know. But the razor with the gloves, this is taste. This is what Tyler looks like when he's organizing our timeline in the morning. Yeah, he's always in this exact outfit, and then he changes into the suit for the show.

1:57:24And I'm like, you can just wear that crazy outfit. If that's who you are as a person, just be yourself. Be yourself. Everyone else is taken. Exactly. Everyone else is taken. but he insists on wearing a costume instead of expressing himself fully but hopefully one day you will see Tyler in all the glory we also have breaking news the US won, they beat Canada congratulations

1:57:51congratulations to America let's go fantastic, fantastic work yes let me tell you about Railway. You heard Sigil mention it. Railway is the all-in-one intelligent cloud provider, user-favorite agent to deploy web apps, servers, databases, and more, while Railway automatically takes care of scaling, monitoring, and security. OpenAI is closing up the first phase of the new funding round, which will top$100 billion. Wow. This would be the biggest round. Biggest number ever. The biggest number ever. The biggest number ever. The biggest number ever. These funding rounds are so big that they leak. So they're like the leakiest buckets.

1:58:35And you hear about them for months and months and months. And you're like, I've been hearing about this for so long. This happened with XAI, where we were hearing about the XAI raise for so long. Everyone's like, wow, it must be going terribly. And then it was like, oh, no, they actually closed it and upsized it a ton because there was actually a lot of demand. And so these big rounds are just like - The upsizing at that point maybe came after they were talking about the merger. Yeah, we did see that announcement. Yeah, again, you can never doubt Elon's ability to win and make progress, but just the traction at the time and the evaluation relative to the other labs was interesting.

1:59:17Dan says, my sources say the model OpenAI has been holding back is a big leap. Not a small jump. Could be today, but I'm not sure. Vague posting. He's vague posting. It's 1 p.m. I don't think we're going to see it today, but it's fun to think about. Well, there's more news, but our next guest, Sagar Anjeti, from Breaking Points is in the restroom waiting room. Let's bring him in to the TVVN Ultra Dome. Sagar, how are you doing? Hello, gentlemen. Thank you for having me back. We have so much good news for you. But first, can we do a sound check? Can you just say platypus 25 times? Platypus, platypus.

1:59:52Whoa! Oh, I'm retired! I'm retired! Thank you, thank you. I wasn't sure you were going to do that, but this is game-changing for me and my family, so thank you. No, we will not be betting on this conversation, we will not be encouraging that, but we will be asking you about white pills. Elite lawyers are seeing record hourly rates. $3 ,400, baby. I knew you were going to love that. You were going to love that. to some of these hedge fund guys are putting up record years. Steve Cohen just pulled in$3.4 billion. We know you're going to be excited about that. Yeah, it warms my heart to see the top of American society doing so well and carrying the U.S.

2:00:38economy on their back. Yes, yes, yes. True elite media. Not independent anti-establishment media. That's what Breaking Point stands for. Breaking the bank. Anyway, great to have you on the show. Let's start with this Wall Street Journal article. So social media bans for youth are gaining momentum worldwide. I was sort of saying like, I get it. But also, it's a little bit of a parenting skill issue. I can make choices like how heavy of a hand do we want on social media? How are you? How have you been processing social media and kids? Yeah, we were talking before the show about this a little bit, and I was saying that an iPad, even without social media, really is like a drug.

2:01:24It probably has an immediate negative effect on the child. But if you're on some place like an airplane where a child could be completely freaking out and really ruining the other passenger's time. You're sacrificing your child. Maybe you've got to sacrifice the child via a little iPad time. But, yeah, broadly. I don't necessarily disagree with that, man. I think that there's a time and a place technology can be a tool. And definitely even some of the people in the no tech community for children will say, look, if we're on an airplane, we can all make exceptions. But I do think it is important to zoom out and say organically, John, people you and I, our age, as we begin to bring children in the world, and as we're on the backside of and really recognizing some of the dangers of technology, phone addiction, and seeing the light in a child's eyes the moment that they see a screen.

2:02:15My kid's only nine months old. If this TV goes on, I mean, I'm talking about immediately going directly towards it. I have friends with smaller children, you know, like two, three years old, and their battle there is constant. I'm not sure if you guys saw the more viral thread recently about Spotify and parents who are really freaking out because Spotify has added music videos. And it used to be that they would give their children their Spotify account. They say, hey, you can play whatever music they want. All of a sudden, my kid's asking for Spotify all the time. It's watching the Moana video, which I would never let them watch on YouTube.

2:02:47And so broadly, I think the entire. Dude, yeah, exactly. I mean, think about our own experiences, right? Our parents, you know, they're like, yeah, mom, don't worry about it. Right. I'm spending six hours doing my homework with my friends. Definitely not on AOL like shit talking with all of my friends. Any millennial, I think, can really understand that. But because of our relationship with technology and then seeing it at such a young age with these children, I do think that parents around the world are waking up to this. And many of the experiments that have happened about banning social media, keeping children away from phones, especially banning phones in school, the results have been extraordinary for the people who are doing it i'm not sure if you saw this there's a recent phenomenon of many parents are even turning in the school issued chromebooks or laptops and saying no we're going back to paper we want all paper exams especially with ai and you know concerns about my parents are they saw us print out the post they saw us print out i love the printout you guys see you see the utility and the physicality of doing that you know i don't know if you guys saw this i'll tell a personal anecdote i was talking about this with reading and one of my friends was like dude you don't even read and and i was like well i i've i only listened exclusively to audiobooks for the last like 10 to 12 years but i was talking with them about how children will mirror your behavior and i was like oh my god my child will never know that i read like 30 40 books a year because they never see me read and so it's time to bust out these bad boys again.

2:04:17There we go. No headphones, you know, just sit on the couch, nothing else going on, 1990s style, and show like this is a very important part of our life. And so the physical, and also you'll discover that the mental exercise by doing that is fundamentally different. Andrew Huberman's actually talked a lot about this, about physically reading the page and memory retention, different ways that that will activate different parts of your brain. So it's really about trying to hang on to the physical world and the benefits that we recognize that that has. Yeah. I'm so curious to know how much these companies, the platforms, whether it's like entertainment platforms like YouTube, much less of a social network versus the meta platforms actually generate from a revenue standpoint from accounts that are like targeting like, you know for teenagers like in in some ways they're not the most valuable consumers because they don't necessarily have a credit card or they can't just like be you know direct response like you know spending money but at the same time i just going off of miss rachel on youtube's like earnings like clearly advertisers will put an insane amount of dollars behind these channels yeah well the thing is dude with that is that it's not about the children spending money it's about how they pester the parents to buy Miss Rachel products.

2:05:39You know, Miss Rachel is a business. She's a billion dollar enterprise. You can go to Walmart and Target and buy stuff with her. I mean, I don't know if you guys know what Tony boxes are, but you can actually get like specific Tonys, which is, it's like a child's toy that plays songs and you can put physical items on top of them. They charge$20 each. There's an entire aisle at Target. Everything is about their favorite YouTube show or like their favorite character that they'll learn about. And, you know, some of them, I have had many parents tell me, you know, the introduction to my show. They're like, oh, my kid knows the introduction to Breaking Points because I watch it all the time.

2:06:14Like they're like, well, you know, welcome to Breaking Points. It's an independent show. But and I get that, like, you know, no fault to to that person. But I am just showing you like the mimetic quality of technology in a child's life and how that bubbles up through everything. So like Minecraft related birthdays and the way that they will socialize with Roblox. right? Like if you were to ask, a friend of mine told me this, you know, their kids were introducing themselves at soccer practice and they're like, my name is so-and-so. And I like to, I like to play Roblox. They're like, Oh my God, me too.

2:06:42Right. And so like the nature of the connection. I thought you were going to say they introduced themselves in real life with their gamer tag. Right. Well, they probably do that. And I'm not sure if you know about the scandal involving Roblox too, which we don't have to get into. But yeah, there's another problem, right? In terms of spending time online and that mimetic quality. So me personally, I can only speak for myself. I'm doing my best to keep my kid and future kids away from this. But the one thing I always want to emphasize is that this is still an elite phenomenon. And at the individual middle class level, the vast majority of American parents are not like woke on the technology question.

2:07:15They may feel something, but a lot of them are still letting their children use a lot of iPad, a lot of TV time. And that's why I do think that the government here, it's almost unfair, right, to leave it up to an individual with vast amounts of resources and algorithm and engineers and others developing, trying to, uh, you know, trying to optimize time on platform, you're up against a very, you're in a very asymmetric war. And that's why I think it's important to try and have some regulation around this. Yeah. The only issue is the, you ban kids' accounts and then the parents just have their account and they're like, here's the iPad or like, here's, here's what I'm putting Yeah, that would definitely happen.

2:07:53So again, it has to, there's maybe a solution at the, at the just age level, but it's also. Jordy, I'll tell you this, man. And I always hear this, especially on, on drugs, right? Can we all posit if drugs are illegal, people will always find a way. What did we learn with gambling? Friction is important. Actually making it illegal is really good. So yes, I will grant you that even if we were to put some modest regulation, some parents would still go around it. The point is about friction and the social signal. If you can reduce things by 20, 30, 40 percent. I'm just saying in this example, it's zero friction, right?

2:08:28Because on my - I really disagree with that. The ability to just create your own account or even the social signal to every parent in America for everyone saying, hey, you're going to have to do a workaround with your parents' social media account and you're going to have to give it to your kid or if we have - I was talking about the example of You're like a five-year-old watching YouTube. It's like it's most likely the parent is. But yeah, if you have – yeah, you don't want your 10-year-old using your Instagram account, right? That would be it. It's not the same. I mean, that's not even social media.

2:09:00But I'm just saying like I think there's two problems. It's being focused on social media. But like highly stimulating entertainment is clearly bad for kids. And this is purely anecdotal. I've tried we've tried like putting on like an animated show for 10 minutes and immediately after the kid yeah just is more way more sensitive like gets gets like really angry short-tempered all these things so it's like okay let's not do that because clearly it has an immediate like there's like a come down yeah it's bad yeah well I'll tell you guys even I mean I hate to say it for my girl miss rachel but i'm never using miss rachel and i'll tell you why is that as a youtuber and you guys know this too youtube is about retention and time on platform and so even though i think she has a role at the end of the day there's a lot of jump cuts smashes constantly keeping children's attention if you do want to play video a lot of parents are experimenting with 1990s tv so getting actual vcrs and playing barney and other things and that's because it's the pre-internet retention era where you have all the engineer.

2:10:10I mean, look, it's not Ms. Rachel's fault. She knows certain types of videos are going to increase time on platform. Why do you think she makes so much money? He's like, I'm not a business woman. I'm a business man. Exactly, dude. And these, you know, if you want two to three hours of retention in your YouTube video, you got to keep your kids engaged. People in the nineties, they were thinking about that to a certain extent, but the technology and the knowledge about, and by the way, we know what What are you doing with your voice right now? You're keeping us engaged. Listen, I'm part of the problem, brother.

2:10:40I'm part of the problem. I wouldn't even be sitting here without those algorithms. That's always something that we got to square. Yeah, I want to talk more about that, but I have a funny story about Miss Rachel. I went to a kid's birthday party, and there was someone who's dressed up as Miss Rachel, like singing a song in the costume. And I was so out of the loop. I was like, oh, is that the real Miss Rachel? And they're like, no, she's a billionaire. Like, she's not just making random appearances. I thought it was just like, oh, yeah, you could like just throw her a couple hundred bucks. I wonder if she does have speaking fees.

2:11:08She definitely has speaking fees. There's probably six figures at least. I, you know, I don't know. Miss, Miss Rachel is actually a big leftist. She doesn't even believe billionaires should exist. So I would be surprised. I would be surprised. I hear she lives very modestly actually in New York, well below her means, but I don't know a lot about her. I've listened to a few interviews. She's an interesting lady. Yeah. Well, what, what about like your interaction with social media, social media addiction, this type of stuff, because I feel like, I mean, we're in a very weird position. I like to say like, I, my screen time is low because I am the screen time.

2:11:39Like right now I'm not on my phone and I actually, this actually shows up in my screen time, like, because I'm on the show. And so like, yes, I'm reading posts, but not on my phone. I'm reading like the paper here physically. And so, uh, but a lot of what I have to do for my job is actually on social media. And it's been both, I really enjoy it and also economically valuable. Like everything is squared where I feel like maybe I'm addicted to, but it's kind of like addicted to your job. I don't know. It's like weird for me, but how have you processed it? Addicted to the game. How have you processed it?

2:12:10Like, how do you think about like your own, your own social media use, how bad it can be for like, you know, people that are sharp about it? Yeah. Story of my life for the last 10 years, literally wouldn't be here if I wasn't addicted to social media. And now also I'm like, social media is bad. You know, I think John, we are in a perfect position, right? Because we can always justify it. You're like, babe, it's for work. Don't worry about it. Yeah, it's for work. Don't worry. And you're just like sitting there scrolling about some drama, intra drama, which if you tried to explain to a normal person, they'd be like, what are you talking about?

2:12:41We had to explain clavicular getting framed. My wife is like, what? Like, is your brain rotted? Or it's like, yeah, remember the Gary Tan meme? They're like, Gary Tan needs to save me. Oh yeah. And I made a joke about that to a friend and they're like, dude, what are you talking about? I was like, Oh my God, I'm so online. Yeah. For my own social media use for anyone who follows me on Twitter, I don't tweet that much anymore. I'll do it like fits and spurts. And I had a lot of rules around like family time, et cetera. But again, you know, would I be here if I hadn't been literally addicted to getting into Twitter fights?

2:13:22The chat, Dan in the chat, I was going to pull up the ladder behind us. The last social media guys. Here's my thing. Here's what I would recommend. If you want to do what John and Jordy and me and I do, you should probably do what we did. You will lose parts of your life to it. It's funny. I don't know about you guys. The most fun I ever had on Twitter was like blue check with sub 5k followers. Oh yeah. Because we're shit posting all J and gentlemen. When you have 500 ,000 followers and you have like, you know, very powerful people who follow me, you're like, ah, dude, it's kind of controversial.

2:13:57It's like, I can't be posting that. Is it punching down? Are you making fun of someone? Yeah, game on. Game on. Like, let's go. 25 back and forth. I miss it. I miss it every day. Yeah, yeah, yeah. I mean, it was also a different time years ago before the acquisition because like truly everyone was there. And now people have fragmented the threads and blue sky. And so you don't get as much like cross-pollination fighting, which is like the real fun stuff a lot of times. You got to take the fight to them over on threads if you have an opinion about basketball now. I have a sort of ongoing concern that the way that feeds work today are going to have some real impact on either short or long-term memory because I don't think the human brain was meant to process like hockey, data center, uh you know uh donald trump yeah sauger like yeah ai video of cat curling curling you know and so and so uh you know you're reading other forms of media even a newspaper yeah technically you could kind of like skim the headlines or whatever uh pretty quickly you go into an article and you're like i totally agree you know uh years ago i quit i didn't tweet for three months i found myself thinking in tweets i would have something happen and i would frame it immediately i'd be like, Oh, this is, uh, this would be the character limit.

2:15:14Here's how I would do it. And I was like, Oh my God, like my brain is rotted, dude. I was like, this is like five, you know, five years of reflex. And this is when I was really, you know, in the game. Uh, and I do recommend that people, you know, really check themselves and part of the way that's part of the reason you should read a physical and a real book. I mean, this is part of the danger. I also think about chat GPT with students. I was talking to a class and I was just asking them how many of them write physical articles. It was a journalism class. Like, do they know how to write a 1000 word article without AI?

2:15:45None of them knew how to do it. It's painful, man. Oh man, it's painful, but it teaches you a lot. Sentence structure, how to convey yourself. And news copy is fundamentally about how to grab attention, how to keep attention, convey as much information as possible, but also be able to flesh out and to tell a story. But that's the hardest part, the third one. And And a lot of these students had never attempted or even really knew how to do that. I was really, it made me afraid, right? Because now everybody thinks of the news as tweets instead of learning how to write. And for like, you know, the Axios, like be short brevity stuff, which I get is very useful.

2:16:22Everything has utility. It's for us as consumers, it's all about thinking about how to balance it. And then really as parents, man, that is the hardest, hardest question that we have to solve. Yeah. Well, let's switch gears to something that we're fortunately not addicted to, cannabis. It feels like cannabis has been growing in a very dangerous way. Andrew Huberman told us that the best thing about cannabis is that you should recommend it to your enemies and your rivals because it will help you get away, get ahead by not using it. But I guess my question is, for my entire life, basically, I guess the last decade, people have been like, oh, it's going to be federally legal soon.

2:17:02There's gonna be some bill, but it feels like nothing's really changed except for the potency, the negative impacts, the adoption rate, the commercial success. But there hasn't really been even like a power law company that owns it all. It's all very distributed still. Like, where do you actually see this going? Is the conversation stagnant or is there something that needs to be reignited? Yeah, I would push you a little bit, John, because the most important thing that's just happened is Donald Trump rescheduled marijuana, which opens the door to exactly what you're talking about. And just to explain it, a banking level, schedule one down to schedule three means that it opens up the ability for the big weed companies to have access to the legitimate banking system, which is specifically what's kept the big weed conglomerate companies that you're worried about from happening.

2:17:49So this is going to happen. And there are actually prototype companies like this that are out there with several billion dollars in market cap. The thing that we should worry about the most is potency. And I've talked about this, you know, Andrew, he's always been great on the weed question. This is fundamentally about marijuana, social norms, and regulatory structure. And when I talk about social norms at the most, it's kind of like what we were just saying with phones, the signal of legalization. And then the cultural milieu around weed was that nobody ever died. It's just a plant man. And you know what I always say to that hemp Sorry.

2:18:24What is it? There are poisons that are plants. Hemlock is a plant. That doesn't mean that it's good for you. Hemlock is a naturally occurring substance. That doesn't mean that it's not also going to kill you. The point that remains - It's just a rock. It's just a metal. Yeah. It's like, yeah, a rock could be smashed to death. So the point is, is that there's this social norm where marijuana addiction is not seen as an addiction. If you wake and bake that entire idea. What do you call somebody who has a drink the moment that they wake up in the morning? You're an alcoholic fucking loser. An absolute legend, sorry.

2:18:58Yeah, no, okay, like, you are an alcoholic loser, all right? Everyone in your life is going to be like, bro, you have a problem. Like, you have a serious problem. If you wake and bake, it's all chill, right? And these guys are like, oh, I smoke every single day, and it's medicine. There have been multiple studies now that show that almost all the claims around medical marijuana are complete bullshit. They are at the very least, it's kind of like SSRIs. People are like, well, SSRIs work for me. I'm like, well, in double blind studies, it actually shows that exercise is more effective and there's significant downside to SSRI.

2:19:31It doesn't mean it can't work. It just means that there's a lot of other stuff, which is not nearly the number of negative externalities. The cultural conversation around marijuana does not factor in all of this high potency, the amount of IQ being lost amongst children, cannabis hyperemesis syndrome, the scrommeting phenomenon, teenagers who are getting checked into the ER at a record height who are vomiting for days on end, and it's in their fat cells, and they're withdrawing from... This is a serious crisis. Ask any ER doctor, and they'll know exactly what I'm talking about. Gabe in our chat says, if you want cannabis to be more regulated, rescheduling is the way because it enables companies to actually do research on it.

2:20:12We know everything we need to know about weed. It makes you fat. You're a loser. It lowers your testosterone. It's bad for pregnant women. It lowers your IQ. It makes you an addict. We don't need any more research. We have tens of millions of research subjects every day here in the United States. No, I mean, I've never, so, so I, I, I, you know, earlier in my career, I was kind of, uh, tried to, uh, tried to track some of this stuff of, of, I, I was wondering, would anyone that was a proud cannabis user, like an active user go on to build a big company and so far, uh, have, haven't, haven't seen it.

2:20:54And so I always saw it as like, it's a performance decreasing drug. Correct. There's some people that are going to just say, I'm fine. I'm fine with that. I, that they might enjoy it and it's worth, uh, like if you're like 180 IQ and you don't want to go into like math, pure math, chess, bring yourself down to bring yourself down. You can go work at a hedge fund or something. Right. You know, guys, it's like gambling and that's always, not everyone who gambles is an addict, but a significant portion of the people who do. And the negative externality of that is the highest suicide rate of any addiction known to mankind.

2:21:25So not all marijuana users are addicts, although about 20 percent of them are daily users. And I would classify you as an addict if you are using it every single day, specifically for like medicinal purposes, you know, fake medicinal purposes. And it's, you know, people just do not grapple with. And by the way, this is the subject I get the most hate on. I'm talking about BLM, trans, Iran, Israel. I'm telling you, nobody has ever threatened to kill me over those. It's always weed. There's something about their attachment, their pathological attachment to this drug that people telling the truth about this substance just triggers them to the utmost degree.

2:22:03Let's switch to something less controversial, data centers. Data centers. Oh, okay. What do you want to know with data centers? Tell me what you want to know. Yeah, so how have you been tracking? I saw something, I think it was New Jersey today had some success banning a new data center project. What are you tracking right now? Specifically, the MAGA revolt around this is incredible. So Oklahoma just had a major town hall meeting. You can go read about the Financial Times, just wrote a whole story about all of the MAGA counties and red states that are revolting against data centers. A lot of these people are actually going there.

2:22:39They're like showing up to town hall meetings. They're staying. The lines are so long till 4 a.m. People are bringing their children with signs that say no data centers. They're like, we don't believe your lies. Stay out of here. You know, it's interesting. What I really love about the American public is they're taking this issue seriously. So a couple of people showing up to the city council were like, hey, I'm reading articles about how the Chinese are able to do this without a lot with less power and resources. So how do we know that you're even going to be sticking around here? Like you're going to build this massive conglomerate and then there'll be a sea change in the technology and you're just going to abandon us immediately with all of your claims.

2:23:15not to mention the power generation issue, which is massively controversial in all of these states. So I actually think the revolt that I predicted, it's already happening. It's a blue state, red state phenomenon. But this is hyper local. Nobody really in Washington is paying attention to this. But when you've got – do you know what it takes for people, ordinary citizens, to show up and stay in line until 4 a.m. to speak against something? It's not a lot of subjects that will inspire that level of hatred. So for all the data center builders who are out there, you got to convince us more, man, that this is actually going to do something for us because we're not buying it right now.

2:23:51What do you think about the initiatives from many of the tech companies to offset the increase in power prices? That seems like very logical that if someone's energy bill is actually flat or falling, that takes them out of the line at 4 a.m. Totally. I'm with you. like i do think that was a little bit reactive to the populist backlash that was coming against data centers it remains to be seen uh whether something like that's actually going to happen or not and also it comes down to the regulators in the individual states now i would totally support that if they remand they make it mandatory you have to build the amount of power if you're going to be able to do this but you know in general guys it's not about the individual policy it really is an overarching populist backlash against ai and i think that's what these guys really don't, what they underestimate more than anything is that the public utility of this technology is just not bought by the American people on this subject.

2:24:47And they really feel as if it's encroaching at a high level and then at a low level, whenever it's your individual community. Well, yeah, again, part of the story is that the data center doesn't need to be in your backyard to get the value from AI. If you ask, every person I ask outside of tech has a great story to tell you about AI. It could be really simple. It could be, oh, it helped me learn this thing. It could be, it helped me make this image that I shared a moment with my family. Like they don't, you know, there's like the higher level stuff. Like, you know, if we can accelerate, you know, scientific research and generally increase productivity across the whole economy, all that stuff's great too.

2:25:31But, uh, again, the, uh, the not getting a benefit from being in your own town is like a real thing. Yeah. Right. No, exactly. And I also say, Jordi, that's a very white collar kind of bias whenever it comes to AI, right? So we're talking about blue collar workers who are consuming a lot of media. You have a blue collar yourself. Brother, uh, brother, I'm not claiming to be one of these people. No, I'm saying you literally have, I'm saying you literally have a blue collar. Yeah, yeah. Fair point. That's true. But I'm saying for a lot of these people, you know, let's say you're a Teamster, right?

2:26:03And you're talking about, and you're reading articles about self-driving cars and the Teamsters Union is sounding the alarm about this. I mean, remember, you know, the vast, what is it? 60%, maybe 45 % of the American public has gone to college, has gone to college, right? So let's say 60 % or so, let's say matriculated, have not matriculated from a four-year college degree. That is a significant part of the US population that's going to think about the technology very differently. And then what percent of college graduates are doing work where they feel under threat, let's say from AI. And they're like, oh my God, like this job could be gone in two to three years.

2:26:38Right. I mean, the current proclamation of doom is around early stage white collar work, not even the blue collar thing. What are you seeing in the economy? Weird job numbers, some job growth, but not a lot. It's all healthcare related and services. Like how have you been interpreting all the different gyrations in the labor market? Yeah, John, we talk about this all the time. We're like, well, the economy's growth is just AI. And then the other backstep of it is just healthcare workers who are like caring for boomers. And that's apparently the U S economy. That's the backbone of the U S economy are like senior nursing individual, you know, like home healthcare aids.

2:27:19I'm not even joking. This is like one of the fastest growing sectors currently of the economy. There's not a lot of innovation. I mean, look, everybody's worried about tariffs. There's some of the manufacturing blowback that's currently happening. I also think, if you look at the labor market and the housing market, the softness in the housing market specifically, that is the one, the populist area. I'll tell you, there is no issue. Again, like with weed, when I get the most amount of hate, the most amount of love is if I'm talking about housing and the inability for people who are my age or, you know, in their forties or all the way up to their forties and really feel uncompetitive in the housing market.

2:27:55And there's a lot of intergenerational rage, you know, about boomers. It's just a wall street journal article might even be in the paper in front of you about how boomers have all of the money. They had a big interactive spread in their paper today, but these things go viral in the wrong way for a lot of younger Americans. And so I do think there's any part of you believe what I think Silicon Valley believes, which is that widespread diffusion of robotics could ultimately bring down housing prices? I don't know. You know, I don't know if anybody's thought deeply about that question. Because obviously there's zoning, there's regulatory, there's all these kind of like local level challenges.

2:28:33But I would say that generally we at the show believe that, you know, 10 years from now we'll have millions and millions and millions of robots that can do functionally useful tasks like building homes and you could, uh, you know, send an army of these robots, uh, and build a suburb in a, in a relatively short period of time. Uh, and, and so I kind of ignores the demand side issue though, you know, because right now the builders will tell you that they are fully capable of building a smaller, cheaper type of house, but that that's not where the market is in terms of the people who have money. Part of the reason that we've seen the rise of big mansions and larger houses is because that's the demand side problem where a lot of people who have more money are the people who are willing to buy these bigger, more palatial homes with lots of experiences in it.

2:29:22The suburb you're thinking about, my idealized, one of my favorite books is about the 1950s. There were these things called Levittowns, you know, where Mr. Levitt would build a suburb and be 1500 square foot house. And you know, wasn't that great? A couple bedrooms that had a car park, not a garage, but you know, a little recovered parking area. It was a couple of 30, 40 minutes from the city, but this was the dream, you know, for a lot of Americans, the so-called starter home, the starter home doesn't exist where I live here in Northern Virginia. The original starter homes from the 1970s go for$1.2 million.

2:29:54Okay. So like, yeah. And they've been massively renovated with the, you know, the rain shower and all this other stuff for the people who can afford it. So Jordy, the thing I think that ignores, I'd have to look it up. What percent of housing costs is ascribed to labor? And what percent of housing scarcity is because of the labor constraint that, you know, or labor costs and not regulatory and existing home ownership classes? You're talking about standards when people reflect on the golden sort of era of American history, when every, you know, family could, you know, own their home with a single income, the houses were just tiny, right?

2:30:32Yeah. They were homes that today people would look consider them tiny yeah yeah no i'm saying people generally would look at them today be like i can't raise a family in that house but there was a time that we did and the boomers had wonderful child wonderful childhoods they're fine and then bought the entire economy for uh potato yeah right and then well yeah go ahead i think the biggest bull case for housing is probably more related to self-driving cars. I saw this chart that showed that basically most humans have sort of a commute budget where they are willing to spend up to 30, 45 minutes a day commuting each way.

2:31:15And so before the horse and carriage, the cities were this big and then the horse and carriage expanded it to how long you could take 30 minute horse ride into your job. Then when the bus line came and the train and the faster car, everything got a little bit more sprawling. And if you think about, okay, well, I don't love being in the car for an hour, but if it's a self-driving car and I'm just sort of sleeping in the back, that probably extends that commute budget a little bit. And then people just push out further into the suburbs. And we do get new suburbs, but they're just in more undeveloped areas.

2:31:47Farther away. Farther away where the land's cheaper. The other connection to that I would add is Starlink. And that's one I really thought about during COVID is how Starlink opened up an entire part of the country, where for years, access to non-broadband internet is a death sentence for working from home. I mean, I wouldn't even be able to do something like this. Let's say I was traveling, and I was on a camping trip or whatever. Now I could stream into you. I mean, shit, I think I could be on a United flight with Starlink, and I could stream in with perfect quality. Yeah, there's going to be viral videos of people being like, I just got on a six-hour flight, and the person sitting next to me has done three podcasts already.

2:32:23Right. It's like, please, by the way, do not do that. Do not do that. You are an asshole. Do not do that. Yeah. Do not take an interview. Anyway. Thank you so much for taking the time to come chat. I'm glad that we could agree. We agree a lot. There's a techno capital solution for every problem in our lives. Yes. Yes. But perhaps maybe followed by a political solution to that. Oh, no. We'll get into that next time. Have a great rest of your day. Sogur. We'll talk to you soon. Thank you. Good to see you. Goodbye. Let me tell you about vibe.co where DTC brands, B2B startups and AI companies advertise on streaming TV.

2:32:55Pick channels, target audiences, measure sales just like on Meta. And without further ado, profanity alert. Profanity alert. Thank you. Yes. Yes. Yes. We need to take that. Sager swears like a sailor. Clean it up. We will be bleeping him in post potentially. I don't know if we'll actually do that. But anyway, we have Dylan Fields from Figma in the Restream waiting room. Let's bring him into the TV panel show. Dylan, how are you doing? Hey, good. How are you guys doing? Oh, salute back. We're doing great. Great to see you. Did you hear the news? America just won in overtime. We beat the Canadians.

2:33:24I did not hear that news. In hockey. In the Olympics, in hockey. Yeah. Oh, wow. It's great. There's a lot to celebrate. Yes. Very exciting. But give us the news in your world. How are things going? It's good. We're working hard, having fun. But yeah, we just did our earnings and really strong results. So we're super excited. 2025 was just a massive year for us. And the fourth quarter was our best quarter yet. 40 % year-over-year growth, right? You got it. We did$304 million in revenue. Wow. And then also, yeah, we look at the whole year. I mean, we just shipped so much. Went from four to eight products.

2:34:08Launched over 200 features. And most recently, earlier this week, on Tuesday, we launched a cloud code to Figma design pathway where you can go from code to the Figma canvas. And I'm really excited about how we can do this entire round trip and make it possible so that wherever you start, we can give you a way to go anywhere on the Figma platform. Do you think people are sort of underappreciating that loop because so many of the experiences with agentic coding systems like Cloud Code are very toy projects. The first time I used Codex on desktop, I just remade the TVPN homepage to look like Berkshire Hathaway's website.

2:34:53And it's like, I just took a screenshot. I didn't need to do any real work. It was a five-minute thing. And I think a lot of people are amazed by the toy projects, but then they don't really think about what's happening when there's a design system. This is a large organization. This is an enterprise. This is going to be something that's enduring. And so can you share a little bit more about that flywheel and what it takes to actually put these tools to use in a serious way? You know, I think it's actually really important, even when you're, maybe it's my mindset, I'm a perfectionist, but even for the toy projects, I want to go figure out, like, what is the best expression of my fun toy project?

2:35:29Sure. You know, maybe something like a Berkshire Hathaway website is a fairly constrained form factor. And so maybe that's not the best example. It's like HTML3. I recently made, for example, my friend was like, I want to go viral in this year. And it was a birthday. Like my goal for this year of my life is to go viral. I'm like, well, that's kind of weird, but cool. And so I made them a website. A lot of space to explore there. There's a lot of ways to go viral. Some can be good. Go on an airplane. Yell at people. What I probably should have done is said, hey, go talk to these guys. But instead, I made them a website with Figma Make and made it really beautiful and tried a few different directions out.

2:36:17And yeah, I very quickly got to, okay, here is sort of with some custom prompting, like a list of ways for them to go viral given who they are. And then it was all collected together in a nice way. But even for a fun project like that, that's just like, you know, I'm going to send to them. They're going to do whatever they do. I don't know if they're going to look at it more than a few times or even a few times. But I don't know. For me, I want to go explore wide enterprise. Yeah, you absolutely need to go think deeply about what's going to make the best product experience for your customers. And that's not going to be just a we're going to go code it up and we're going to do whatever we come up with first.

2:36:58You're going to go actually think through the option space and try to figure out what it is that you want to go steer towards. That is critical. And if you don't think about that as a system, if you don't actually use all the components you've standardized on, the styles, then you're not going to appear the same way to your customer throughout the entire user journey. and it has to really tie together with brand as well as your product, as well as your point of view and your marketing and your messaging. So it's super important to make it so that you're able to have all that come together. And just being in code, I think, is very linear.

2:37:38It might be running fast, but it might be running towards the wrong direction. Yeah. What's driving the Figma Make growth? Is that the same archetype as the Figma user? does it diverge in any interesting ways? What can you share about how interesting ways people are using Figma Make and also how you're thinking about growing that product? Yeah. When you look at Figma Make, I think what's really cool is to see how many non-designers are using it. Okay. And that is something that we've really picked up on. It's not just the designers that are going in and making really amazing prototypes through Figma Design, but also folks are coming in and making these files and prototypes and actually full working web apps and experiences.

2:38:29I mean, 60 % of files created in Figma Make are non-designers at this point. Wow. Which is amazing. Yeah, I mean, that was a big part of the original Figma thesis was that there were going to be non-designers using this tool from day one. Basically, if you looked at the TAM of the business purely on a product designer basis, it didn't look like it would be a business today. I don't think you ever had a job as a designer, right? Yeah, and yet I've used Figma every day for a decade. Yeah. So, yeah, that's interesting. How do you think about the growth of that product in terms of adding plugins, features, hosting, deployment, uptime, edge distribution?

2:39:12shit like that could grow into its whole thing yeah or maybe a more simple question is like where where do you want to partner yeah where do you want to build yourself yeah i mean i think that we're gonna see tons of partnerships uh for things like figma make where you need to be able to pull in context from whatever systems people use and then with that uh use it to help create and help access data. But more importantly, I think that it's really key that we both create a simple surface that if you're just getting started you can use. And also we figure out how do we make it so that you're able to go above and beyond and go really professional and do things that basically require a skill ladder to traverse.

2:40:05but visual first. That's the thesis, is visual first. Make users grew 70 % quarter over quarter. We're seeing great adoption. I think people are really resonating with the product and finding all sorts of fun and awesome use cases for it. I think there's such opportunity given the way that our active users of Make are also using Figma Design to bring new services closer together which I think is back to the divergent point because make is still too linear for my taste. Oh, interesting. What feature requests are you prioritizing and most thinking about, and what are some maybe more wild feature requests that Figma may not get to?

2:40:50Well, I mean, there's a feature request in this conversation to go do edge computing. I don't think that's going to happen. Okay. You know, all of the folks out there working on edge computing, you know, come partner with us. Hardware. We're not in your space. Don't worry. Hardware. I want a hardware device. You mentioned taste. How have you been reacting to the taste discourse? Is taste a new course skill? How many different cycles of taste discourse have happened since Figma was started? I think about companies like Linear who, to me, exemplify a taste-driven company. And how many of these cycles where they're like, oh, they care about this again.

2:41:34Great.

2:41:38I mean, look, I think that from the start of Figma, the goal was not just to get everyone doing design or making design software usable by everybody and expanding the market. That wasn't the way we thought about it. It was like, how do we get people started on this journey where they can actually go and be more expressive, more creative and, you know, really start to solve problems? And, you know, maybe it was a bit of like rejection. But for me, at least, I wake up in the morning. I'm like, I want to go create software. I want to go build stuff. And so how do I make it so that anyone with that impulse can just do that?

2:42:17That doesn't mean they're all going to have like taste and craft and epic design skills. I mean, I think for the rest of my life, I can work on that. And many designers have made that a lifelong pursuit. Yeah, yeah. There's something that's very real where you start on this creative journey and your taste level can be way above your skill level. And it's incredibly frustrating. And some people start and they just stop because they feel like I have this idea for what I want something to look like or be like. and I'm kind of like trying to make things that match that and I can't. And that's part of why AI in general is like incredibly exciting because it can help you close that gap faster and potentially help you close that gap almost instantly in some circumstances.

2:43:04But you might create a hundred different outputs before getting to the one that actually meets that level. Yeah, I think that the key thing is that you really have to go wide and explore and then challenge yourself. If you find areas where you're going, hey, actually, I don't feel like I am liking this enough, I'm not happy enough with a solution, then you've got to go keep pushing. But the more you can sample the possibility space and see some things that you like and you don't like, it gives you something to react to. And I think that the constant thing is you need to be constantly critiquing and thinking about what is it that you like, you don't like, etc.

2:43:50And people talk about agents all the time. And I'm obsessed with agents too. We all are, right? Agents are cool. How many Mac minis do you have? No comment. He's the reason that Mac minis are sold out. I swear I am not Claudebott. But anyway, I think the bigger point, going back to taste, is if an agent can do it for you, then unless you've got some amazing, sophisticated prompting that's super unique, an agent can do it for someone else. And so I think that this discourse on agents are just going to do all the things, well, what is different about your setup than others? you at least have to have something different there in order to not think that you're just gonna get the same thing everyone else is gonna get yeah well even then I don't know if it's gonna get you sick extraordinary yeah yeah we were in a post yesterday um uh that was sort of trying to quantify the the background that leads to taste and it was basically saying that like Steve Jobs was incredibly high IQ but also had like varied training data in that he had like been homeless and like traveled the world and like got all these like bizarre experiences.

2:45:09And I'm wondering if that like resonates with you or you think that it's, it's like too difficult to even quantize what makes for a great designer or great taste. You know, I think that the great designers I've met through my life, um, have come from so many different backgrounds. You know, a lot of, a lot of folks have come through rigorous design training. Like they just, just went through a system and other people, wow, they went and created their band poster and that got them into graphic design. Somehow they showed up at the right house and crashed on someone's couch and then became a product designer and then everyone went back.

2:45:47Everything in between. There's not really a pattern. Some people have traveled the world and gone on the meditation retreats and some people are as buttoned up as you could get and just totally straight-laced. I think that the nice thing about designers is they're so unique and there's no pattern matching. Yeah. How do you think about variability of design going forward? Are we at sort of a narrowing period because most people come with just a generic prompt or because I saw your Figma make examples and I was like, I didn't even know that was possible. There were so, but I was like, I I would never have gotten there.

2:46:33So are there going to be tools to maybe help people like be inspired by your background, which has a whole bunch of variation on it? I don't know. Yeah, I'm just wondering because, I mean, we were just talking about a company that it feels like the prompt was like, copy Jordi's website. Maybe it wasn't, maybe it's just coincidence, but it felt like, okay, like there wasn't, there wasn't like a twist here where it was like, oh, just do something completely different that's never been applied to this particular category. at least like steal from a different niche instead of from the one that's like very, very similar.

2:47:08Um, but I'm, I'm wondering how you're, how you're seeing like variation in design, uh, take, take, take hold in the world. Yeah. This is the moment I think, uh, where we're going to see the pendulum go from, I mean, let's look back. Right. So we had the flash era, geo cities, like I'm not saying it was high quality era, but there was a lot of variation. And then iPhone comes out as skeuomorphism and then Swiss minimalist design. And nothing wrong with Swiss minimalist design. That is a really cool and storied field and lineage. But it is just one part of the greater aesthetic realm. And we can go into such interesting places and try so many interesting patterns on the UX side too.

2:47:57It's not just UI and the visuals. It's also the structure, the IA, the way that people navigate through these things and interaction patterns. And I think there's innovation that's going to be flourishing on all of it. I'm just so excited to see the Internet get really dynamic and really visual. And people try things that we haven't seen in a while and things that people haven't seen ever. Because I think that that's what it's going to take to stand out now. There's been so much there. The exponential curve of software is just taking off in a way that it's always been exponential, but now it's vertical.

2:48:36We talked about this before. And in order to stand out, you guys have a show that you're managing to actually break through and get people to be watching this. This is not normal, right? But you have worked very hard, very diligently to create the conditions under which you have this audience. Well, this is the thing that everyone has to figure out now is how are you going to actually get any attention in a world where there's constantly new information, even beyond America winning the ice hockey match? yeah since i've been i've been thinking about uh uh how gen ai is impacting marketing it's allowing somebody has like a unique idea and then it used to be like you had like at least a month to kind of like run with that unique idea maybe three months if it was like you know a specific campaign or something like that and now somebody can literally fast follow you and your idea almost immediately so good ideas always get copied yeah but they get copied even faster now another thing that that i was thinking about when you were talking about kind of the explosion of ui uh john had a had a post yesterday that he was sort of jokingly making like a buzzfeed style listicle like five five features that could explode your llm's usage or something like that uh and one of them was like caching and people have had this idea of like generative UIs, UIs that's being made like on the fly.

2:50:10And I feel like we may get, we may like have products where that's happening, right? Like you're in an LLM and a UI is being generated, but it will still make sense. Like if, if there's something that happens all the time, like thousands, millions of times a day, billions, like it does not, it really stops making sense to just generate it on the fly. It's like, Hey, this happens a lot. Let's make the best version of this and we don't need to like light. I had a poster on this actually, exactly this topic. I can send it to you if you want. Basically the post was about how the length of time that an artifact will exist for is inversely proportional to the likelihood it's going to be generated.

2:50:54And I think this is true across media. So basically if you are writing a book, unlikely that you're going to have AI just generate your book. yeah uh amazon booksellers would like the word yeah you just you can just say you don't read adult romantic fiction we get it uh no no no i think i think i think uh continue yeah but like i think um you know something that's like a ad that will live for you know a few hours totally like yeah probably you're now an agent and it's going to generate that. The difference is you're making an asset to respond with a meme format to news. It makes a lot of sense to put that through a Nano Banana or a ChatGPT, but if you're making a billboard, maybe you're still using AI to some degree, but you're going to invest significantly more time.

2:51:47I think if you're going to make a billboard that's going to cost money and it's going to be up for a month and a lot of people are going to see it, I think you're going to have a human touch. We actually, with Weavey, now Figma Weave, we've been looking a lot at these kinds of use cases. And one thing that we think about a lot is how you have this first prompt. But then you actually want to go and have it go through a process where you can transform and mutate it. And almost like clay that's being shaped, you can treat it like a medium to get to a final output that's amazing. So in our earnings call, we talked about how NVIDIA, which they've actually done this as a public case study, they put the entire making of this keynote online and they used Weevy with it.

2:52:32And what they did was they had all these robots and they needed to basically get to a 20K image of this whole scene. Like how do you create a 20K image with perfect lighting throughout? I mean, it's really tough. So they had a whole custom pipeline with Weevy they did. and just like the amount of work to go do that at scale uh of that pixel density is immense and i thought their workflow was super cool so i'll share with you too i love it uh yeah please send us that uh and thank you so much for taking the time to come chat with us thank you for having me always a good time we were we were so fired up seeing the results from yesterday really a testament to how locked in the team.

2:53:19The team is amazing. Yeah. They're working so hard and they're just an incredible team. I'm so grateful. Yeah. And I think it's a proof point. I mean, I think that we're going to see it over and over again, whether it's the taste discourse, just customers with Figma. But design is a differentiator. Like, figure it out. Go learn design. Hire your designer. Otherwise, you're going to have a hard time. That's my last message. Well, thank you so much for taking the time. Great to see you. Have a great weekend. We'll talk to you soon. Up next, we have Peter Morales from Code Metal, and we are kicking off our Lambda Lightning round.

2:53:55I believe Peter is in the Restream waiting room. He's about to be in the TBPN Ultra Dome. But the Lambda Lightning round has begun. Here we go. What's happening? Thanks for hopping on. How you doing? Great. Good, good. Great to be here. First time on the show, please introduce yourself and the company. Hi, I'm Peter Morales. I'm the CEO of Code Metal. It's great to meet you guys. Excited to be on the show. Extremely metal name. I'm sure you've heard that before, but I had to do it. I love the name. Yeah, what got you into this? Talk about your journey, how you got here, and then we'll get into the business.

2:54:36Yeah, so at least for me, I started my career in defense. I graduated with a physics degree, and this was one of the areas that was like, hey, we want people with that background. And it turns out it was an awesome area to jump into. You got to apply first principles, but then also build systems that had impact. And so spent some time developing, basically taking software that was written at a high level and bringing it down to low-level hardware systems. and the reason being was I was a big programming nerd but also kind of understood the math and it was a sweet spot of translating software to hardware.

2:55:09So I got to work on some cool AI algorithms on the F-35 and that kind of inspired me to do more work in that place. So I went over to MIT Lincoln Laboratory and started doing research in AI, got to be one of the founding members of their AI technology group and then ended up going to Microsoft where I was hoping to put all these pieces together and get to build more for commercial industry products that took sort of state-of-the-art algorithms and brought it down to hardware. And so I was working on the HoloLens and hoping to learn kind of these like industry best practices. And it was the same as defense, man.

2:55:42It was just as like hard to push the latest and greatest down to hardware and, you know, inspiration behind Code Metal was there's got to be a better way to do this. Makes a lot of sense. Before we go further, I don't know what or how much you can share around AI in the F-35, but generally for those types of exquisite systems, like how, you know, back when you were actually doing that work, where is AI actually being impactful in an exquisite system like that? Yeah, so you got to view it from like our perspective, which is like, hey, we're going to take these at the time, kind of really cool algorithms, and bring these down to ancient processors.

2:56:23So exquisite system, but awful hardware systems to actually get the algorithms running. And this was part of the DARPA arc program. So it's like adaptive radar countermeasures. So the idea being is, hey, I've spotted you in the sky, and I'm going to go ahead and try and take you out now. But maybe I don't know the complete parameters behind what that system's doing. And so I need to kind of on the fly come up with a strategy to defeat those missile sites on the ground. And so applying some different techniques to that, I can't talk about, was sort of the gist of the game. Fascinating. Give us the quick history on CodeMetal then.

2:57:03When did you actually start the company? You're coming on today with some big news, but give us the history until now. Yeah, so it's been a real rocket ship. We started just two years ago. Wow. Great success. Nice. so um yeah we've we've had really really great partners j2 ventures did our pre-seed round and um you know their roll your sleeves up kind of like help you founder friendly fun and we jumped right into it at shield capital who really understood the dual use mission and then you know once we started seeing this takeoff and sort of for us and i'll get into what we do exactly um you know in some of the mission critical industries that we operate Excel jumped in and it's just been sort of milestone after another in terms of the real problem that we're solving, which maybe I should give some background on next.

2:57:52Yeah, please. Yeah, go for it. There's this explosion of basically AI-generated code. And it's great. You can build an MVP. But I'm sure all of you have been working with Cloud or OpenAI. And maybe you've asked it, like, are you sure? Or like, is this right? Don't make mistakes. Yeah, exactly. Build me a brake system. No mistakes. You say you're 100 % sure on this. Yeah. Yeah. Turns out to be wrong. Take off from the aircraft carrier. You're good to go. Yeah. You're not crazy. So, yeah. So, the idea here is like we want to answer that question. So, as much as we are an AI company and we're an AI verification company, and that means inherently these systems are stochastic, right?

2:58:40And so you can't necessarily rely on AI to check AI. And so we're leveraging a lot of techniques both from sort of traditional test verification and formal methods to basically prove as you work your way through these hard software challenges that we have done the identical thing that you put in. And in this case, we can't rely on natural language. Like writing a prompt is not a good way to define how like an exquisite system needs to operate. And so we're not like a low code or no code, we're a high code, you know, shop. And we focus on basically solving software problems that are the real tall tent poles in production, namely moving from MVP to production for some of the biggest industries.

2:59:18And defense being obviously one of those based on our backgrounds. Very cool. Take us through the funding round. What happened? Yeah. So exciting news to announce today. We have announced our Series B,$125 million. Wow. Congratulations. And tell us about the new president and COO. How'd you meet? What is he going to be doing? Why is that an important role to fill? And I think that's the second big announcement for us. We've been really heads down, a bunch of us engineering nerds and also starting to grow into business folks like Laura Shen. She joined us from National Security Council, had a background at Uber, and really was sort of like my main teammate on the selling side.

3:00:04But we've grown a lot since then. And we've got operational needs. We need leadership on sort of working with some of these bigger clients that we started to talk to on the technology side. And, yeah, Ryan Atte from CEO of Tableau actually joined us as president and COO. And, yeah, he's been awesome already. Like we learned about him through our network and sort of trying to find the right person who had the right sort of curious attitude, wanted to first principles come in and do something actually really new. and yeah he's been an awesome force just experience working with some of your NVIDIA's of the world the different partnerships that he's had at the places he's deployed he knows our customers and so having him on the team has just been awesome.

3:00:47Assuming you guys execute your roadmap perfectly which I'm sure you will what does it do to hardware development timelines things like that is part of CodeMetal's job to bring the acceleration that developers are feeling when they're building enterprise software, maybe they're building AI apps, et cetera, to bring that to the hardware level? Yeah, you're nailing it. So a lot of these industries have had, they're not one generation behind, and we don't want to use AI tools. They're many generations behind, even when you get if you're a cloud developer. So I look at it as three different things.

3:01:30The first is like getting actual features, acting like you're a vertically integrated software hardware company, because all of a sudden you can push things directly to hardware and see those things run on whatever system. So that's sort of our rapid development vertical. Then you've got code portability. So there are huge supply chain problems. So, you know, we've talked to, for instance, like the Army and they've got, you know, hey, we wrote all this code for NVIDIA GPUs. Like it's ours. Like we've developed it. we maybe have supply chain issues here. Can we run on a different piece of hardware?

3:02:02All of a sudden, that's going to gum up the work. So you have to pay every contractor to rewrite the code. They could translate that. That solves programmability. That solves supply chain issues. That's another huge area of growth we're seeing. And the finally is like legacy systems. Like there's been so much investment in code that thou shall not touch. And we just put a box around it and talk to it. Being able to update those and deploy those has been another big area of growth for us. That's awesome. Very important work. Thank you for everything. Road metal is accelerating. It's heavy metal.

3:02:29I love it. Customers are accelerating. It's very metal. Yes, and congratulations on the funding round. Yeah, congrats to the whole team. Thanks so much for coming on the show. At this rate, I'm sure you'll be back on in Q2. We'll talk to you soon. Have fun out there. Have a good one, Peter. Cheers, Peter. Let me tell you about Plaid. Plaid Power is the apps you use to spend, save, borrow, and invest, securely connecting bank accounts to move money, fight fraud, and improve lending now with AI. Hi. Up next, we do not have our guests. We have Eric from Freeform joining in just a few minutes. But in the meantime, we can talk about, friend of the show, well, he sounds like he's here.

3:03:07So we are going to bring in Eric from Freeform. He's been on the show before. Welcome back. How are you doing, Eric? What's going on? Yeah, good to see you guys. Very cool background. Kick us off with an explanation. These are all squat racks that we're seeing, right? Yeah, these are just fancy looking things, you know. They don't really do anything. What you're looking at behind us is our next generation manufacturing platform. We call Skyfall. It'll be the fastest and most automated laser melting platform on the planet. I think last time we talked, we had talked about how we were getting ready to design and build this.

3:03:42And now we're almost done building it. Yeah, amazing. What else happened? There's some news. Oh, man, lots of amazing news. We just closed a big Series B,$67 million. You know, we're... We got to hit the gong. Amazing. Amazing. We've talked about getting a gong in the office, too. I love it. You should. We know the finest gong manufacturers in the country, so happy to give a reference. Yeah, we'll have to get... You'll have to pass those along to us. It was led by Appandeon, Michael Linz of Linz Capital. NVIDIA participated, Founders Fund, Threshold, Two Sigma Ventures. We've got an amazing cap table of investors with deep expertise and operational scaling and deep tech space.

3:04:34It's been great. Amazing. Where's the biggest source of customer growth coming from? What products are actually getting made, at least that you can talk about? Yeah, for sure. I would say there's growth across a lot of different sectors. Demand is outpaced capacity, which is amazing, and we're building a substantial backlog. Yeah, it's been fantastic. Obviously, there are a lot of critical industries out there, the geopolitical climates of the world today where flexible manufacturing is becoming much more and more important, shortening iteration cycles, being able to scale up quickly. These things are just super important in that space.

3:05:19And then, you know, you're just seeing, I think this, you know, the advancements in AI are driving, you know, advances in engineering, in design, in access to information that is accelerating product design, which then people, you know, want to build faster and scale faster. So there's all these kind of forces that are driving advancement in all sorts of different industries or problems that these industries are looking to advance manufacturing technologies to solve. Talk about aerospace or maybe building airplanes generally. I imagine that we're not using freeform to make the fuselage, but are we talking about like, you know, the seatbelt buckles or blades inside of a jet engine?

3:06:08like at what what scale are we operating here yeah for sure great question um no we're not making the fuselages i think uh you know we're not a hammer looking for a nail sure um or a hammer looking for a face that's right that's right uh the you know we're generally talking about parts that are of the scale you know the platform behind us is roughly 600 millimeters by 600 millimeters by a meter tall. So something that can fit into that envelope. Obviously, we can make extremely high-precision things. We can make them out of virtually any material, any metal. And, you know, intricate features, big features, small features.

3:06:53We can do just about anything. Any metal. You know what I'm thinking? Coffee maker made purely out of lead. Lead-based coffee maker. Yeah, there's a very viral coffee maker trying to be free of plastics. And so, So, John, I mean, a lead based coffee maker would be free of plastic. Yeah, clearly that wouldn't have any plastic. No problem. No problem. Yeah. How much have you scaled up the team since we last chatted? Where is that going? Yeah, it's fantastic. And I want to add on to my answer to your last question, too. We're flying mission critical parts on multiple frontier programs today. We are shipping out of the factory many hundreds of parts per month that are going on.

3:07:35very important vehicles and systems. So, for the team question, we're about 60 people, 55, 60 people now, so I think when we spoke last, we were maybe 35, 40-ish. And we're trying to hire about 100 people over the next 12 months. Where are those jobs going to be based? All in the same place? Everything here in Hawthorne. cool amazing for now yep very great amazing uh on the on the parts side like what what's the what uh tons of demand but where why are customers actually choosing you guys is there kind of a speed up around kind of like order timelines things like that like where where uh what's making customers the most happy yeah i think you know honestly we've cracked the code i mean the Let's go.

3:08:35Let's go for cracking codes. We love cracking codes. When we talked before, it's just interesting how far we've come in a relatively short period of time. I think the rigidity and inflexibility of traditional manufacturing systems is just no longer acceptable. And you've got this massive advancement in the digital world that's occurring, these transformations that are happening. And the way we make stuff hasn't really changed in decades, not fundamentally. And so, you know, I think customers are looking to design more and more advanced things. They're looking to control physics in new and different ways.

3:09:19You know, you can, I saw over my decade plus at SpaceX, you can design, you know, the most advanced things. But the real problem is when you go to make it. And, you know, closing this gap is becoming really important. and what do companies want to do? They want to focus on what they're good at. And the code we've cracked is that we are the most successful and fastest growing company in the advanced manufacturing space that leverages printing ever. And because we have developed, we've solved all of the major problems that were fundamentally preventing the technology from being scalable. And so we're no longer in the proof of concept phase where we're proving you know that the theory of it we are this capital is being deployed to massively expand capacity to meet the the demand that is we're building backlog and it's continuing to accelerate amazing last question for me what's the biggest lesson that you learned from spacex that you find useful applying today oh man i think you asked me a similar question last time there's so many lessons i'm so fortunate to have had such a an amazing experience but maybe a different answer now maybe a different answer now than when we last chatted you know you've the the codes have been cracked yeah yeah i mean i think that what comes to mind is you know the the relentless uh pursuit of the thing that you know is possible you just you have to fundamentally believe in what you're doing.

3:10:54It has to be based on, you know, first principles, like this is a possible thing. And then I am just, we are going to relentlessly pursue it until we get it across the finish line. And I think that's why we've been so successful here, you know, combined with the willingness to, it's not about how hard something is, or if I know how to do it or I don't, if I, if I don't know how to do it, but it needs to be done, I've got to figure it out. And we're so vertically integrated across everything we do from software to machine learning to advanced compute to electronics to the physical platforms that we build ourselves that this is something I don't think there's any other company out there that is like us on this front.

3:11:37We don't use third party products at all. Nowhere in the stack, top to bottom. Well, thank you so much for taking the time to come chat with us. Congratulations on the progress. Congrats to the whole team on all the progress. I'd love to hear that you guys are shipping parts. In Southern California. Let's go. I like the Hawthorne, Gardenia, Gundo, Long Beach rivalry. Yeah. Just who can produce the most mass in the world is fantastic. Yeah. Well, have a great rest of your day. Great to see you, Eric. We'll talk to you soon. Thanks again for having me again. Good bye. Thank you to see you. Let me tell you about Phantom Cash.

3:12:12fund your wallet without exchanges or middlemen and spend with the phantom card up next we have lubisha basik from talus welcome to the show how are you doing what's going on i'm doing well how are you did i did i get even close to the name first name great second name close okay give it to us by each by each okay well powerful first time on the show please introduce yourself and the company. Hey, so you introduced my name. I'm kind of a long-time, I guess, chip designer more than anything else. My latest endeavor is Talos. What we do at Talos is, you know how I'm sure both of you have used ChatGPT, right?

3:12:59And when you ask a question, you see that words are kind of rendering on the screen. It takes a while. If you use it a lot, it costs a fair amount of money. So for people that run code through it or something like that. So what we do is we basically make those responses show up in milliseconds so that you can ask whatever you want and you get like 100 ,000 words, it pops up in a split second immediately. The web UI is slower than our chips for the most part. And it's also kind of close to zero cost. The web UI is slower. That's crazy. Super fast, super cheap. You're like, this HTML, this JavaScript just doesn't load as fast as the tokens.

3:13:40What a world to be in. That's right. That's happening today. Yeah. That's happening today. And I guess the way we do this is kind of the opposite of what your previous guest was saying. So, you know, everybody says you should be flexible. We say you should be really inflexible. Okay. But you should be hardcore. Okay. Hardcore. So, yeah, take us through the technical story here. Is this only possible post-transformer, post-MOE model? How much are you calcifying, and what technical decisions do you have to lock in before you went into tape-out? We're not so much focused on a given model type. Whatever model is popular today, it's transformers with MOE.

3:14:23If it was something else, very likely, most likely, we would have no problem consuming it. really the main step is that we take that model and we basically cast it straight into silicon. It's not a piece of software. It doesn't run on a processor. At least I, as a kid, I never imagined AI as an executable, like Excel or Word. The general mindset was like Dave from Space Odyssey, or Hal from Space Odyssey. right like and uh so we're kind of you know going to that idea and basically making models straight into hardware once we make this chip it's it is a model that's that's basically uh that's all it is if you want a new model you have to chuck this one out again it's kind of a throwback to um to the past like where it's kind of like video game consoles yeah you know there's a receptacle uh you unplug a cartridge you plug in another cartridge and you have a new model what are the different trade-offs when you go to the fab are you thinking wafer scale large die how do you think about all the trade-offs and what is available at modern fabs so our approach is that you know we we basically made this one trade which is commitment or flexibility for kind of everything else given that we tried super hard to make everything else that simple which means no interposers, no 3D stacking no super high speed I.O.

3:15:55no HVM the chips are pretty big but they're really really simple they plug into a normal server they don't consume a lot of power you can cool them by air you don't need anything but old school data centers so it's essentially we made it such that there's this one sort of compromise that we make you take having taken that it's better at everything and it's way simpler than anything else Yeah. What is the what's the sales process and look like for this? And what are your customer relationships actually going to look like in practice? Are you I imagine, you know, you see massive progress on the model side every day.

3:16:37There's a new model. Are you like, how are you going to scale up kind of like physical production to sort of match that progress? Or is that the wrong way to think about it? You know, it's hard to be sure. So I'll answer all your questions briefly. So on the first one, our default plan is that we're bringing out like an inference service and we're going to let users basically get to tokens directly. Obviously, we aspire to having large customers like the big AI labs or hyperscalers. It's just that, you know, like if your default plan involves closing Microsoft, it's a bad default plan. It's not an easy step.

3:17:23So it's not that we're not doing it, but we're doing something that's going to bring us to customers and allow iteration and it's fully within our hands. It doesn't involve a big uncontrollable step. To your second question,

3:17:39our approach starts winning at about three months of holding the model. Even if you swap them every three months, we still win. At a year, we win by a lot. So at this level, it's really hard to call, but I feel strongly that there's applications that can certainly tolerate having the same model for a half a year at a time. What exactly those are, I don't know. Yeah, someone was saying yesterday that GPT 3.5 still gets like a shocking amount of So much money, yeah. So the other day, GPT 4.0 was retired. There was a bunch of sort of negative remarks on the internet. People are sad and angry that it's gone and so on.

3:18:22So, I mean, we definitely want to be at the leading edge. We think that it's actually quite doable. But in the end, like where the line is, where sort of people are comfortable with the commitment we need, time's going to tell. How much is AI accelerating the translation from the model weights or the model architecture into an actual chip design? In our first generation, not a lot at all. We use it, but lightly. For our second generation, we're actually trying our best to maximize the amount of stuff that AI does versus what people do. So we're aggressively moving that boundary, and the cutting-edge models have gotten good enough that they can really help a lot.

3:19:15We started about two years ago. That wasn't the case back then. It's a new phenomenon. Yeah, makes sense. Kind of a wild card, how are you thinking about what we expect is an explosion of new AI hardware devices? We're hearing about pins and in-ear things. Is this something that could end up in a consumer electronic device at some point, or is it better suited for more of these data center or enterprise applications? I think it's suited to both. But we're initially targeting data centers, and we made these big chips for that reason, as opposed to something that's smaller form factor, could fit in a device.

3:20:02Really mainly driven by the fact that this kind of avalanche of PIN and such has been announced now. You know, it's been five years away for kind of multiple sets of five years, potentially, right? Yeah. Yeah, let's not think about VR. Yeah, and you can obviously make a hardware device. that can deliver AI without having the chip locally. Yeah. Well, Jordi, anything else? No, let's hit the gong. Okay, yes. Tell us about the news. How much did you raise? We actually, the special thing about today is that we've launched this product to the public, right? We did raise some money that we didn't announce.

3:20:43Actually, we raised a bunch of money. We hit like kind of nirvana where we can live off of interest. with like 220 million dollars raised total but what we're really proud of is that we spend money super slowly and that we we really are living off the interest right that's amazing well congratulations where where can people go uh do you have a chat jimmy.ai if you want to demo it yes right that's that's right chat jimmy.ai i can send you guys the link if you don't already I think it's Llama 3.18B? That's right. Yeah, baked into a chip super fast. It is quick. It's fast? Yeah. I love it. It is instant.

3:21:28It almost seems faster than my eyes can process. That's the point. I think that's the future. Very cool. That's what we're shooting for. Well, congratulations. Awesome. It was great to meet you. Thank you for speaking to us. Great meeting you both. Thank you very much for having me. Congrats to the team. We'll talk to you soon. Cheers. Goodbye.

3:21:46Are there any other stories you'd like to go through, Jordy? We could talk about the F1. We could talk about semi-analysis. We could talk about the AI doc, or we could talk about all that tomorrow. The AI doc, we're working on getting the team from there on. We can skip that. Yeah, Dylan Patel is allegedly mulling, raising a venture fund. Sounds like he invested some size into FluidStack already. Yeah. This is funny because this is framed as a scoop. And I suppose it is. But they were openly discussing this on a podcast like a month ago. So it's not that big of news, but it is exciting. We love Dylan Patel.

3:22:30We love some analysis. And it'll be very interesting to see what they do. I think that there was some mention, I mean, they weren't talking about it in fine detail, but just a discussion over whether or not this would actually be a full venture fund or if it would be more of an RIA, more public-private. There's a variety of different strategies, and Dylan knows a lot of different folks in the ecosystem. Obviously, he works with a lot of hedge funds, and so there's a lot of opportunity up and down the stack, not just in the private markets, and every venture fund has sort of a different strategy when you have this level of analysis, rigor, data, understanding the markets.

3:23:11There's a whole bunch of different ways to express that financially. And so I'm very, very excited to see what happens. And good luck to the Seminassus team out on the fundraising trail. Well, last but not least, a new California law, which is FIPVCC, is forcing VCs to ask portfolio companies about their sexual orientation, disabilities, and more. Also, it can be reported to the state. The team at AngelList built a new system to help funds in AngelList comply without ever seeing individual founder responses. This seems very important to preserve founder privacy everywhere. We've decided to make this system available to any fund for free, regardless of who they work with for fund admin.

3:23:50So obviously a bunch of funds are running on AngelList, and we'll have this built in, but important that everybody be able to use this, seems like. I remember when this law was originally being proposed. Did it actually pass? Sounds like it. Oh, okay. Which is surprising, but nothing is surprising in California. Well, DoorDash is surprising to me. Over a billion orders a month. Yeah, well, soon. They're doing 900 million right now. Oh, 900 million, okay. 900. So let's hit the gong for, oh, that was a, yeah, I can hear it. You probably can't hear it at home. Lock in DoorDash folks. Keep going. We'll wait.

3:24:31Call us when you hit a billion. That will impress me. Now I was listening to Keith Reboy talk to Eric Newcomer and he was saying like his whole DoorDash thesis was just the DoorDash was the get food button or the I'm hungry button on your phone. And then there's a whole bunch of hard work that you need to do to actually deliver food, but the consumer impulse was just, I'm hungry. And you push the button and you get food. And it has worked to the tune of 903 million orders per month. I wonder if there's more around Christmas. Do they always see a December bump? No, not really. Huh. December was a good month for them.

3:25:06They jumped. They actually jumped in March in 2021. I don't know if something happened, but they've been on a turn. Have they ever had a down month? I don't know. I don't think so. They had a rough time in September of 21. Coming out of COVID, it went from 345 to 347. Barely any growth. But I don't think they've ever had a truly down. This is quarterly, I guess. Wait, is this? Oh, no. This is monthly reporting, but you see every four months. Ben is getting us fired up. He says three and a half hours. Let's go. Let's go. We're not quite at the three and a half hour mark. We've got two minutes left.

3:25:44We've got two minutes left. There was one more story we didn't get to. Blue Owl permanently halts redemptions, a private credit fund aimed at retail investors. People have been talking about this as kind of like a layman moment. But I looked at the size of the fund. It's$1.7 billion. Fund for ants. It's a private credit fund for ants. Interesting. No, not great. I remember we talked about this story last year. They said they kind of leaked that they were considering halting redemptions. And they were like, actually, don't worry, we're not. And then now they actually have. So not great. What do you think lock-in meant?

3:26:23Getting locked in? You just got locked into your investments. No more redemptions. It's the great lock-in. Again, don't call it a code red. Call it the great lock-in. You're good. Don't call it we're suspending redemptions. Call it your lock-in. A Baja blast. We're going to Baja blast your returns. Yes, exactly. To the stratosphere. Anyways, folks, thanks for hanging out with us today. Thank the bomb. Let's do it. Playing at the bomb. Great show. A lot of fun. Always fun having Sagar and Jetty on the show. He is... Really gets people riled up. He's a professional. He's just a professional. And so he shows up.

3:26:57Sounds good. Looks good. Great takes. Back and forth. Definitely our most controversial guest. Just judging by the chat. But we have a lot of fun. Yeah. A lot of fun. Anyway, leave us five stars on Apple Podcasts and Spotify. And we'll see you tomorrow. Have the best evening. Nice work brothers. I'll see you on the next one.