Math Wars, GG Navier Stokes, Greg Brockman Joins | Dan Wright, Eric Seufert, Scott Wu, Sahir Jaggi, Andrew Borovsky, Harry Mellsop

0:00 You're watching TBPN. Today is Monday, 09/08/2026.

0:06 We are live from the TBPN Ultradome, the temple of technology, the fortress of finance, the capital.

0:12 You know what time it is, John.

0:13 Rent.com time. Time is money. Save both. Easy as corporate cards, bill

0:19 pay, accounting, and a

0:20 whole lot more all in one place.

0:23 I need my countdown. I need my countdown.

0:27 Doing some math.

0:28 There it is. Road to Christmas, hundred and seven days.

0:31 December 25. By. Today.

0:34 It's flying by. Made only possible with math.

0:37 Hundred and seven days. Time is flying till Christmas. Start. Road to Christmas is going strong.

0:42 Start scoping out Christmas trees.

0:45 Yep. Now's the time.

0:46 Once we once we get under a 100

0:47 Hopefully, you've already planned

0:48 open season. Open season.

0:51 Does math matter? That's the big question. I was debating this with Tyler today. Lots of lots of lots of tension on the timeline over Navier Stokes, the Millennium Prize math problem. There has been a Millennium Prize math problem that was solved before the Poincare Conjecture in the pre AI era.

1:11 These are very, very difficult math problems that all PhD mathematicians, all sorts of elite, the upper echelon of math has been grappling with for years. And AI is starting to knock these down. Remember, last year was the year of the IMO gold medals from both Google, DeepMind and OpenAI. We talked to Scott Wu about this, who's coming on the show later today. Huge raise for Cognition.

1:37 And Scott made the prediction on TBPN early in 2025 that he believed that artificial intelligence would be able to achieve gold at the IMO, the International Math Olympiad in 2025. And he was right. He nailed it. Progress continued but debate remained around cool calculator bro, basically is the is the critique. I mean people just they don't feel empowered by a system that's really good at math.

2:10 They just people already a t I 83 can do more complex math than most people can do. And so when you say, oh, there's another level of math that you don't know about and you don't care about and computers are good at it, a lot of people, their eyes glaze over. And so the bigger question here for me is all about the the public perception. What actually matters here? I think they're cool benchmarks.

2:35 I think they're interesting stories. Obviously, there's a lot of drama over this race. It feels like it was neck and neck. Now there's a bunch of debates going back and forth with a mathematician who works at Google and a mathematician who's independent who worked alongside an anthropic researcher who were both working on it. And they were apparently texting back and forth and going back and forth on did they steal the data?

2:57 Did they copy each other? Do they have different approaches? And so lots of different back and forth. And I think someone summed it up well by saying that these math models have discovered the the hardest problem of advanced mathematics which is authorship contribution. Who actually did the work? Who gets lead left on the paper? But yeah, there was a drama last year, 2025, during the IMO. OpenAI and Google DeepMind and I believe Harmonic, the math AI lab that we had on the show last week, were all working on solving the International Math Olympiad. Remember the six questions. Question six, no AI model could solve.

3:42 It was more complex. But both teams got a 35 out of 42 even though they only got five questions out of six right. That's good enough for gold. It's the bare minimum, but they did it. But there was a bunch of back and forth because OpenAI validated with former IMO gold medalists while Google's result was graded by the official rubric that year, which is internal to the organization.

4:05 We called it at the time, it's like they ran the fastest 100 meter sprint in the parking lot while the while the actual Olympics is going out in the stadium. And so a lot of these are just, you know, vibe wars, races, who does what, does it even matter. The interesting thing about Navier Stokes is does it matter, Tyler? You think it does? I'll debate you on this a little bit.

4:33 It seems like it doesn't because the equations themselves are known. They're just not fully proven. Engineers and physicists already solve them numerically all the time for a particular situation. So airflow around a wing, you might have to use some of these equations, weather models, water through a pipe. These are critical things.

4:55 They are boring. But you can imagine that there would be cause for optimism. You know, better weather models save lives during hurricane season. Increased airflow across airplane wing that could lead to more efficiency, longer ranges, cheaper flights, lower carbon emissions. There's a bunch of reasons why you could solve all that and be excited.

5:16 But that's not actually what

5:18 One of the reasons I'm excited, I I believe we agreed last week that if anyone solved Navier Stokes, Tyler would shave his head.

5:25 Oh, that's right. Didn't we do that? We said something No.

5:28 No. That was about the But

5:34 yeah, this is a big moment.

5:35 Yeah. I think it was actually a board it was if any if any AI progress happens at all, Tyler, should have to say.

5:41 Yeah. Yeah.

5:42 But do you have a steel man for why this matters?

5:45 So I I think I agree that like, I don't think that the breakthrough, you know, this new stuff that's going on is, like, practically

5:53 Yeah.

5:53 Super useful. I don't think you're gonna feel it when you're when you're on an airplane. You're not gonna feel, like, less turbulence now because But of I think, like, you know, these are, like, the most important open math problems. I think, like, these are more close to like, you know, a beautiful painting or something than like some like practical engineering Yeah. Like this is like pure math.

6:13 This is like the, you know Yeah. This is the peak of the mountain of like. Same month. Math.

6:18 Same month that AI solved this incredibly difficult math problem. I saw someone use Jaggi Poutine to book a haircut. The cool no. I mean, in terms of the actual economy, the actual impact of these things, like practically using these tools for something that people enjoy that's not just spinning their wheels, it's not just optimizing endless things, that's actually generative and moving things forward, not just defending against cyber attacks, doing

6:47 says, if he gets a millennium prize, you all have to get bowl cut.

6:51 Oh. Bowl cuts.

6:53 Bowl cuts are interesting.

6:54 That's really really ridiculous. You do get a million bucks. So, you know, there's that. That's not nothing but in the AGI race where trillions are being deployed, it is it is very minimal. But the it sounds like cause for optimism that Navier Stokes would improve airflow over a wing, something like that, but that's not really how it will work.

7:22 Proving this this these equations is unlikely to actually have any measurable impact and that's sort of agreed upon and physicists have sort of chimed in with this take that this doesn't this won't actually move the discipline of engineering forward at all, but it is a cool demo. Engineers already solve the equations numerically, as I mentioned. So clearly, it matters for hype and vibes, solving how our math problems is good heuristic to show how quickly models are progressing. But Yeah.

7:53 And and to to me, it just comes down to Yeah. Humans are able to use machines that humans built Yep. To further humanity's general knowledge.

8:02 Yeah. Yeah. It's good. And but I feel like benchmarks and if you're just trying to measure AI progress, benchmarks and raw inputs and raw metrics of the model sort of do the same thing. Like, the number of people who will be convinced that AI is progressing exponentially because they see this result.

8:24 And then you look at the number of people that that would be equally convinced of AI progress just by hearing, oh, there's, you know, 10 gigawatts coming online or the new model has 10,000,000,000,000 parameters or it was trained with a billion dollars of computer or something like that.

8:39 Like The that thing is the average person is probably more impressed by Sora than by this.

8:44 Totally. Totally. It's way more visceral to actually see something. And that's why the weekend was very much Blender and people showing three d renders and remodeling their houses and building small games and web games. And that feels much more like a okay breakthrough.

9:01 People are having fun. They're vibe coding stuff but they're also doing three d modeling which is like a new skill. We've had image generation, video generation but there's something much more, I don't know, like grounded in watching a from just a few images watching Blender model come to life. You can walk around inside of it and it can be something that you know the model doesn't exist. I saw a few demos that were like a train and it was like, oh, wow.

9:29 I modeled this amazing train. But there are perfectly modeled trains out there that you can just go grab from the Internet. That's not that impressive when you actually see your your specific house or your car or something that doesn't already exist out there perfectly modeled. You're like, wow.

9:45 Sorry to interrupt. Jon Anderson says, why are we only seeing the calendar days for Road to Christmas? Can we please get a weekdays and workdays breakdown for DieHards? Yeah. We need a we need a full screen graphic.

9:57 Let's work on that. Thank you. Good. Thank you, John.

9:59 Comment. Good comment. Let's Yeah. So in terms of field the AGI, computer use, definitely one of these. Astra, particularly faster at computer use for like really, really mundane things.

10:17 Also just interesting to just have a very different interface to your computer, like configurations, all the things where you set to dig through dig through a bunch of set settings, maybe set something on the command line, maybe find a control panel. You can just ask codex or any AI agent to go and do it, and it'll just do it and come back to you.

10:37 One of the use cases that I'm excited about is getting you know, when you get a parking ticket Mhmm. And then you're like, oh, I got this parking ticket. I wanna deal with it. Yeah. And it's like, well, the ticket is not in the portal yet.

10:49 Sure.

10:49 Being able to go into the chat Yeah. And say, hey, this once this ticket is live in the system

10:56 Yes.

10:56 Please pay it.

10:56 Parking super

10:57 And so

10:57 you can just forget about it. Yeah. Parking ticket super intelligence.

11:01 It it is It feels like we're close.

11:03 So, yeah. I I think that the Millennium Prize will get solved or proved or, you know, delivered probably this year, if not like this week. It feels like all of the different labs are like, put a million dollars of inference on this. Put a million dollars of inference on this. There is gonna be a back and forth and the labs are going to very quickly solve basically all of these problems.

11:26 And then there'll be a search for new problems. But the conversation will move back to curing cancer, almost Because that's the one that's so grounded, that's the one that's, you know, supposed to be tractable. It's been messaged from various labs for so long that people will want that, but there will be a longer flywheel from actually designing a cure for a specific cancer, running tests, seeing that it was effective. It's not, you can't do it with a million dollars of inference over a weekend, I think. But that's I I my my prediction is like the conversation will go back to biotech after we basically solve math, which is a crazy thing to say, but it feels like we're there.

12:10 What do you think?

12:11 Yeah. All the conversation

12:14 Yeah. People say lab leads to stop.

12:16 Yeah. Yeah. So one, the pharmaceutical industry has been curing forms of cancer or making progress against various forms of cancer for a very long time. Yeah. Credit.

12:32 Worst reputation of any category of businesses Yeah. On the planet. Yeah. So I don't think that that's like this, you know I don't think that's a solution to sentiment around the technology industry.

12:47 Yeah. I know. And setting up

12:50 this idea, like, imagine how underwhelming it will be if any AI lab comes out and makes like twenty percent progress towards One type specific

13:03 of cancer?

13:03 Type of cancer. Yeah. Right? That's not gonna come across as like, oh wow, cured cancer. Yeah.

13:09 It's like, okay, they sort of advanced understanding and maybe the treatment process. And this type of thing is happening all the time. Anybody who's had a family member that's suffered from cancer has heard about, hey, there's this thing in trials that's showing promising results, trying to get you into it or whatever. So that's happening all the time. Yeah.

13:31 Right? And it's not gonna suddenly change how people feel about the technology industry broadly.

13:36 No. It'll diffuse slowly and people will be back in like the what have you done for me lately? I mean, I just saw I just saw a real where's basically, the thrust was like, what has the the pharmaceutical industry done for us lately? And and I was like, GLP one's, like, sort of cured obesity? Like, it's And sort of a big for a lot of people GLP-1s have been like a really, really key intervention in helping them fight diabetes, fight obesity, fight a whole bunch of knock on effects that come from that.

14:08 But it's not so much as like the speeding train was coming towards them and the industry pulled them off the tracks the last second. It's something where, okay, if you're living at a healthier weight over from your thirties to sixties, you will probably be healthier at 60. But you're not seeing that like it saved a life today. Like a GLP-one has yet to get that type of credit. I don't know.

14:34 It's going to be back and forth forever. But anyway, in the meantime, somebody gave Astra a paintbrush, a robot and a camera and asked it to paint the Golden Gate Bridge in real life. It figured out how to control the robot and progressively got better throughout its attempts. The time lapse is sick. Here's a one minute time lapse of co I I don't know how to pronounce this, but his account painting the Golden Gate Bridge.

15:04 How would you rate this, Geordie?

15:08 It appears better than I could do.

15:11 Better than you could do. How long are you giving yourself?

15:14 I'm saying at least by the by the the later attempts.

15:17 Okay. The later attempts because it gets better. Yeah. The first one was not quite not quite there in my opinion. Still pretty impressive.

15:25 Yeah. Somebody was talking about potential for a three d printer boom with all the three d modeling. The idea that you could you can design. I saw someone else designed a a system where you can from a prompt, go from prompt to Lego kit basically in one go. So it it takes your prompt whatever you gave it.

15:47 Like, look at this, John.

15:48 That looks

15:49 pretty That's solid.

15:50 That's pretty solid. I think you could do better with enough time. I don't know how long that took, but it is impressive. It's a fun fun demo. And again, it like lives in the real world.

16:00 It's not abstract at all. So you can just watch it work, watch it do the thing. And it's very very interesting. A lot of times what's interesting is that when you're using these models, when they have to do a task like that, they will write a deterministic piece of software to execute it. Like I built a game where you, it was basically Tony Hawk, but you would play as a pelican on a bike and you'd be able to do tricks and back flips and the game worked really well.

16:26 It was fun. It built it in Godot, the open source framework for game development. And then I had to go and try and get the high score and instead of actually using computer use to go and play it, it wrote a script that would play the game flawlessly. And it looked really mechanical because it wasn't there was no variation in it at all, like watching a normal person play. But it did get a very high score.

16:50 So I I could see it doing the same thing where it generates the image and then maps the x and y axis and the plan into code and then runs that. And so there's like this process of what the model can do. The model might also write software to do that same thing. So there's this like iteration process there. Yeah.

17:08 And of course people are looking at Bach bench which is the the skill in writing a piece of music and it is the actual meme for my robot. Can a robot write a symphony? Can a robot turn a blank canvas take a blank canvas and turn it into a masterpiece? And I don't know if people are calling this a masterpiece, but they're pretty pretty impressed.

17:33 It was pretty good. I listened to this.

17:34 It was good. You're Yep. You're a fan of this lily pond four part corral in the style of Bach.

17:41 One of the best reactions to Astra's release last week was Josh, who's over at Century. His company was acquired by Century. He said after Astra's release yesterday scoring 99% on ARC AGI three, I think it's become clear that AGI is no longer some distant hypothetical. About fifteen minutes after the benchmarks came out, I booked the first flight I could find to Jackson Hole. And he goes on to say, he's also looking into purchasing a horse.

18:09 At this point, horses remain one of the only major transportation platforms with no API, no OTA updates, and no realistic path to MCP support. So glad to see more people

18:22 Getting into the horse.

18:22 Getting into the horse game. Bull market in the equestrian lifestyle.

18:27 Yeah. Seems like he was joking, but a lot of people were like this unironically. I want to, you know, touch grass, I suppose.

18:35 Did you see this Yes. Bojon posted, it's over. Our last line of defense has fallen. Astra beat all 48 levels of the I'm not a robot game. Yeah.

18:46 What is And this was crazy to watch. I I wasn't sure if this was sped up. It seems like it's maybe sped up a little bit.

18:54 I think you have play Tic Tac Toe too. It's pretty good. Yeah. What? Yeah.

19:02 The the capture defeat is gonna be whole new level of problems. At the same time, it feels like sort of remarkable that Pangram is as effective as it is given how advanced text models are. And so if you I mean there's a world where it's, you know, indefensible. Was that Waldo Bench? Was that find the way

19:25 It was finding Waldo, not creating.

19:26 Finding Waldo. Waldo. How how are we doing on Waldo Bench? That's really a test now.

19:31 For Astra.

19:32 Yeah. We gotta see. Because I I bet that there's a way to to have Astra generate like the the proper tiles. Because the problem when we

19:40 Yeah. You can like iterate on them more.

19:41 Exactly. When the problem is that when we try and use an AI image generator to generate a Where's Waldo, it always gets the scale wrong, I feel like, or there's like two there's like bigger characters in the foreground.

19:55 Yeah. Perspective is wrong.

19:56 The isometric display is what's what makes a Waldo. And but but there's probably ways to solve that with a little little sprinkling in of deterministic cogeneration or Waldo generation. But Waldo bench might be saturated now. I don't know. So it can't really be saturated because it's not real benchmark.

20:16 Lots of lots of back and forth in the benchmark world. It does feel like we're getting to the end of it and the and the results are either big challenge like a like a math problem or just a cool demo like this.

20:28 Another good demo from somewheresie. Astra, he says, Astra and I are working on making my contact form as difficult as possible. You actually have to fight battle in order to

20:41 This is hilarious.

20:42 Contact them.

20:43 That's amazing. This looks like a I might try and contact somewhere after this. This is pretty fun. I like this. Yeah.

20:51 The the defeat of all of the captchas is is somewhat linked to the the Instinct news. Did you see that this weekend, Jordy?

21:03 Oh, yeah. The resi bands?

21:05 Yeah. So Instinct is a personal assistant AI focused on consumer. You link it with your email, your text messages. And we've seen a lot of investors sharing posts about

21:17 For what it's what it's worth, I'd be surprised if anyone in the audience hasn't already heard of Instinct.

21:23 Yeah. And we talked about it before.

21:24 Thank you for reintroducing.

21:27 But but I but I think it's worth like explaining like how it works because it does seem like it triggered some bans. There was one report of an individual who got banned from Resy for because they were absolutely spamming to get reservations to restaurants.

21:43 And there was another example of somebody who was at the US Open was on the fan cam and Oh, that was really cool. As the user experience would have been absolutely incredible. Right? Sitting there and you're like, hey, I appeared on the screen Yep. For a moment.

21:59 Can you can you go and find this?

22:00 I've had that idea so many times. Like, oh, go find someone on the on the Yeah.

22:04 On the fan But I think what Instinct was doing in the background was being absolutely relentless and it seems like reaching out to hundreds of people aggressively, you know, following

22:15 the media contact. But they did wind up getting the video.

22:19 Yeah. Worked.

22:19 And it was mutually beneficial. And you can imagine if the other side has agent intermediation and filtering and can process things that are spam like, but there's actually a reasonable action to take, that could make it through a filter and not actually annoy anyone on the other side. So that was a very that that that was probably a glimpse into the future and like a cause for optimism. I could see I could see that being popular. But people are doing all sorts of stuff.

22:46 The the the classic example is like go save money, get refunds, do that sort of stuff. But people are are having fun fun with that. People are so many so many demos. This three d website that pulls apart the male anatomy into 200 2,234 model pieces. I shared this with a friend who I was trying to like explain in one video what to get like their wheels turning.

23:13 I think this is a good example. I wonder how much of this was like individually modeled versus pulled from an existing model and then just animated. But certainly works, certainly works well. Yeah. I wonder, oh, Anish made a game. That's fun. Astra's Opus 4.5 for games made Contra but photorealistic this weekend. Link to play below and it's it's up on Vercel. I wonder if we will get an actual boom in in Steam. It feels like Steam will be going through something similar to the the Kindle Direct Publishing store where they where they will be flooded with inbound.

23:57 Well, didn't we see this with the App Store too? Where there were like a disreport like like the number of of App Store submissions for Apple like Yeah.

24:04 Text. Totally overwhelm their systems.

24:06 But no real breakout apps. Like the top of the app store is still like JaggiPT and like free money or whatever. What a teamu.

24:15 Yeah. It's evolved a little bit. The the company that I'm confused that's been consistently in the top 10 Mhmm. Is this company Vinted.

24:22 What's Vinted?

24:22 Pre loved marketplace Good. By selling by secondhand clothes.

24:26 People are buying clothes. It's shopping app. Everyone needs So big TAM, you make it to the top of the store. But, yeah. Like the like you'll be able to vibe code a game.

24:36 You'll be able to get it on Steam or, you know, in the App Store. How do you get the distribution flywheel going? How do you innovate to actually break through? How do you make something that, you know, goes viral, gets traction? Because the, like, the the long tail is gonna get a lot lot lot longer.

24:53 What you do you think? Think?

24:54 I I feel like for a lot of these games though, you should just put them on keep like keep them on web because it's much more accessible. Not everyone has Steam. You have to actually download like

25:03 a I feel like web games are still like not that fun.

25:07 There were some web games I was playing this weekend

25:09 How long?

25:09 That you'll make. I played Did

25:11 you play more web games this weekend or did you play more Oculus VR, Meta Quest VR when you had the VR headset for an unlimited amount of time, but let's call it a weekend.

25:24 Okay. I was probably playing more VR.

25:26 VR?

25:27 I was playing Call of Duty.

25:28 You played Call of Duty in VR?

25:29 Yeah.

25:29 How how long? A couple hours?

25:31 Yeah. Probably a few hours.

25:33 Not bad.

25:33 But was playing well, when I was making a game too this weekend on Astra, I spent a lot of time. Yeah. So does that count as playing the game?

25:40 No. Can you drop it in the chat? No. I

25:43 think I think I mean, I I do think that playing a game that's more you designed yourself or for a small group chat, that's the same thing as the images. Like the AI images I generate are not broadly beautiful. They're not going in the moment. But like they will often be funny to me and three other people.

26:02 Yeah.

26:03 Right? Low TAM. Yeah. Low TAM. But that's the beauty is that with lower cost you can go after lower TAM opportunities.

26:10 And so I mean this happened what a couple months ago. Somebody send us, I mean Jeremy Gaffan simulator is the classic example. Someone sent us Jeremy Gaffan simulator version two, which was very low tam in the sense that it was just in jokes with me and a couple friends. And and and the the long tail of these things will continue to flourish. And then of course, will use them in, you know, real businesses built on IP, built with distribution.

26:36 You know what time it is,

26:37 What time

26:38 is it? Time for an ad. Console.

26:49 It's trailer time.

26:50 Okay. It's trailer time.

26:51 We got two trailers that did the timeline. Mhmm. Let's start with Nathan Fielder

26:57 Mhmm.

26:57 And his new film featuring none other than Elizabeth Holmes. Elizabeth Holmes

27:02 filmed before she went to prison.

27:04 I can promise you that.

27:07 Eight twenty four is not tear.

27:09 Deceive you on. Of course, I'm not deceiving you. Engaging with you as a human being. Why would I deceive you? There's no there's no reason for me to do that.

27:27 Okay.

27:29 The acting is so good. Okay. Good shot. How long is this You're being real right now?

27:48 You're always being real.

28:02 The team is stoked. Looks beautiful. Good sound design.

28:08 Saw post. Somebody saying it feels like I'm living in a simulation

28:12 Mhmm.

28:12 Because this media product is like perfectly with Yeah.

28:17 Oh, yeah. Yeah. Yeah. Yeah.

28:24 A 24 specifically because it feels like the TAM is just like Yeah. Like 20,000 people. But but in

28:31 The Theranos story was really big. Yeah. She was on the cover of Time Magazine, I believe, and cover of major magazines. It was all over the news when

28:39 it No. It's And it's a lot bigger than you would

28:41 The John Kerry Rue book is just a fascinating read. It's a real page turner. And so I think a lot of people read the book. The the the adaptation of the actual play by play story, not the documentary, but the what do they call it? Bio pick, I guess.

28:57 Yeah. That I think did not break through in the same way. But it certainly like continued to elevate her persona and then the posting has been like kept her relevant. Like, there's a Yeah.

29:09 There were some speculation that maybe it was Nathan that got access to

29:14 the account.

29:16 Completely disagree with

29:17 that. Yeah. I do not think that's likely

29:19 All I think that would violate some tenant of what it means to do a documentary. I mean, Nathan blurs lines all the time in terms of how he interacts with his subjects. So it's certainly possible, but I just don't I just don't think that's reasonable at all. I I I think they're her

29:35 much do we know outside of the Like trailer we don't know

29:39 Well, this was leaked something like a year ago from a journalist who got the story and it was widely denied that it was happening. And he was very upset that he wasn't getting enough credit for being correct here. He ultimately was. But, yeah, all we know is that they filmed the documentary in the lead up to her actually going to prison after sentencing. There was a pretty large gap between before when she had to report to the prison.

30:06 She gets out in 2030. So coming up three years, four years away. And so I believe even in a week, you can get a lot of footage if you're spending the full, you know, every day with someone. And then there were a number of visits to the facility by Nathan that were recorded probably talking through glass or talking at a table, something like that. So I think some flexibility there, maybe some phone calls, maybe some correspondence in the aftermath.

30:35 But we don't know what what?

30:39 SBF, Nathan Fielder. Oh. Sequel to this one. Yeah. This just becomes like a

30:45 The problem is that he's already in jail. So the the question is just what else does Nathan for Nathan Fielder have in the can? Because he's obviously thinking years ahead since he filmed that exact scene in I believe 2022 or 2023. And so who knows? There there's a rumor that he's secretly running a like a boy band right now.

31:10 There's this rumor that that there's a boy band that's There's popular on TikTok. This is real. That's popular on TikTok that's entirely invented by him. Boyfriend. Boyfriend. Yeah. Are you familiar with this?

31:20 Really?

31:21 I thought

31:22 you were gonna say he's Nathan Fielder's SD kid.

31:27 Yeah. The so yeah. I don't know. I mean, well, we these projects will emerge and obviously they're taking longer and longer because they're bigger films now. But

31:39 Yeah. After the rehearsal Yeah. It was hard to imagine how he would one up himself. Yeah. And it feels like he may have done it.

31:47 Yep. I think he a new tall mountain.

31:48 We're still stay with us in the trailer zone. When we come back, we're gonna be talking about artificial.

31:56 Let me tell you about the New York Stock Exchange. Why don't you change the world? Raise capital at the New York Stock Exchange.

32:05 Pull up artificial. Artificial. Trailer dropped just this morning.

32:19 Is the image of the future that you see?

32:23 Wait, pause. Was that a Valkyrie? Yeah. Okay. Continue.

32:32 So the visuals are already incredible.

32:36 They really are. I'm pretty sure it's a Valkyrie. It might be a little hollow.

32:41 Is that not the same Lucas singer?

32:42 No. No. No. No way.

32:44 No. No. Because you can see the the suspension Yeah.

32:49 Through that

32:49 that front section.

32:50 Okay.

32:51 Anyways, continue.

32:54 What is the image of the future that you see? We've created a machine that will solve the world's problems. We don't teach it.

33:06 Gate check. Gate check.

33:09 Nailed it.

33:10 Check? Oh, the gate. Yeah. Okay. For

33:14 sure. We

33:14 open Pandora's box together.

33:26 Voice is still Eduardo Saverin, but the styling and the crazy thing is I have that exact backpack. You get to be the one. I think I got it recommended by MKBHD. To

33:37 shepherd people in to this new world.

33:44 Lots of cameos. Not cameos, but

33:49 I'm so excited to see it.

34:01 Traumatic.

34:01 Dramatic.

34:02 Be interesting. It'll be interesting to see the the takeaway. People in the chat are back and forth. Is it too soon? Is it unnecessary?

34:12 Is it gonna be something that, like the social network, is divisive, but also, like, it's oddly inspiring to a certain cohort of entrepreneurs? We'll see. We'll see what the reactions are.

34:25 Yeah. Do you think that was Do you think Hollywood took away any lesson from the social network? And that they were trying to frame Mark Zuckerberg as this, like, generally, you know, troubled person. And then a lot of people came away feeling like, wow, I wanna be like Mark Zuckerberg.

34:44 Yeah. Sort of a yeah. But yeah. Unintended consequence of it. I don't know.

34:50 I don't know. I mean, they're going back for a second second scoop of the ice cream with the social network, the social reckoning. Is there a trailer out or just images? I think I think we saw that too. So it'll be interesting to see what happens there.

35:03 But get your tuxes ready. Going back to back, artificial social reckoning and then capping it off You can with

35:11 see everything. Yes. Then also the To Catch Predator movie. That's gonna be a good one.

35:21 Yeah. Outside of tech tech world. Yeah.

35:23 Guess you're right.

35:24 Yeah.

35:24 But still

35:25 Triple feature.

35:26 Yeah. It just feels like it's a it's a year of like niche movies that I expect to be breakouts, I guess. Like along with like Backrooms and Obsession and and these two movies and then the documentaries. There's a lot of things that are like oddly specific, but I would expect they do well. But I don't know.

35:46 We'll see. The social network did very well at the the box office.

35:50 And while we're at it Yeah. Let's stay in the trailer zone

35:52 What you got there?

35:53 And let's pull up the social reckoning. Okay. The official teaser.

35:56 Mhmm.

36:00 Tyler Culberson is calling it pure cinema.

36:03 Pure cinema.

36:06 But I

36:07 am. I am here to help Facebook, not hurt it. Okay? Alright. You send me a message.

36:13 What would you like to talk about?

36:18 The chairman gavels a session to order. You'll read your opening statement, which we'll skip past for now.

36:25 That's a

36:26 separate session.

36:27 And which And which one is this about? Questioning. I don't I don't think this is based on state your current the politics. It's not Cambridge and

36:35 a little

36:35 r k c c

36:36 c r Frances Haugen

36:38 your occupation?

36:39 Teenage girl whistleblower thing, which honestly looks a lot different in the backdrop of the addiction trials.

36:45 For a thirty percent I wonder if they were trying to get

36:46 this out before the trials.

36:48 Hang on through. Understand what you're saying. Aren't you a tech reporter? Ish. Ish?

36:53 These guys are counting on the next round of congressional testimony to make you likable, Mark. I'm happy to lend a hand, but I think you're doomed. This company and that guy are playing an unprecedented role in our lives. The fire hose of bad information you are injecting into the air supply is becoming jet powered. I'm a free speech absolutist.

37:16 I'm not the one who's lying, and I'm not stopping them from seeing someone who is.

37:21 Anxiety, depression of teenage girls got worse as a result of time spent on the platform.

37:26 Senior leadership knows and is doing nothing. I know there are easier enemies to make. The mafia would be an easier enemy to make.

37:34 So what would you need?

37:35 To stand up the story? The internal documents. This is a material of my NDA. We're twice as big as the biggest country on Earth. We're not frightened of Congress or post government around here.

37:44 Please.

37:45 Who is that supposed

37:45 to please?

37:46 Let me quote that.

37:47 We have a hundred and two hours to get everything.

37:50 She's gonna get sued into small pieces.

37:52 I don't wanna be made an example of by

37:54 a guy with unlimited resources. Heart, I promise you, is imminent.

37:58 Yeah. Chad is calling it out, but I don't think that I think this one I think this one's gonna flop

38:04 for sure.

38:04 Of a dorm room anymore.

38:05 Because of the lack

38:06 of supercars? Well, lack of supercars, and this is just not top this story is no longer top of mind. True.

38:14 Yeah. People wanna hear about, like, the talent wars. They don't wanna hear about, like, this kind

38:18 of stuff. They do. The talent wars. If there's no Alex Wang character, I'm not I'm not paying good money.

38:26 Says, oh, technology is so scary.

38:33 Yeah. I don't know. Yeah. I mean, they should do more tech movies. Maybe Seuss Popular

38:39 SES is Popular whole lot. No Brazilian jujitsu. Yeah. This was in the pre true. A lot of

38:45 making it in the script. Yeah. UFC.

38:48 If there's the the Social Network three and Yeah. All the scenes where Zach is talking But he's in the middle

38:54 of how is it 2026 and we don't have a Salesforce movie?

38:58 Seriously.

38:59 Like, the the the comeback, the sass pocket.

39:01 Or at least a documentary on the relationship that he has with the dolphins. Would be That

39:06 would be great.

39:07 And the and the wildlife generally.

39:08 Yeah. I'd like a CrowdStrike movie. I'd like a MongoDB movie. Let's do it. MongoDB.

39:14 What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI, own the data platform that powers it.

39:21 Here's a concept, Dolphin Force. How one man tamed It's a good story. Wild dolphin.

39:28 It's maybe the best origin story of a tech company ever. I don't know if there's a better one. It's the best. It's just pure bliss. You know?

39:38 Pure pure plot positivity.

39:40 And for those that

39:41 don't know. Company.

39:42 Benioff famously conceived a vision for Salesforce in 1999 while swimming with a pod of roughly 100 dolphins off the coast of Hawaii during a sabbatical. Feeling one with the pod, he envisioned cloud delivered enterprise software leading him to quit Oracle and secure 2,000,000 in seed funding from his mentor, Larry Ellison. Yeah. Incredible.

40:05 Be great. Be great. Anyway, in other news, lots of craziness in the AI world. Lots of new models. But the jobs apocalypse has been officially postponed by The Economist.

40:19 An AI jobs boom is here. And people have a wide range of takes on this. Is it just postponed? Is it permanently postponed? Kevin Bryan summed it up well. He said, I am shocked, shocked. So he's being sarcastic, of course. Shocked to find out that a productivity enhancing investment supporting technology is good for workers. Knowing nothing else, this should be your prior because this is how productivity and competitive markets almost always works. A new technology comes out.

40:49 It supports investment. It enhances productivity. Everyone hires to go after all the all the different battles. But The Economist breaks it down in a lot more detail with a bunch of charts and a bunch of data. So they say, perhaps AI will eventually make many humans unemployable, but there's no sign of it yet.

41:06 On September 4, the Bureau of Labor Statistics reported that the American economy added a 162,000 jobs in August, far above expectations. The unemployment rate, just 4.1%, lower than almost 90% of months over the past fifty years. Young workers, cast as AI's first victims, are holding up remarkably well. The gap between unemployment among twenty to twenty four year olds and the overall rate is close to a multi decade low. Some companies and workers are being severely disrupted by AI.

41:40 Hiring professional and business services is running 10% below the average from 2015 to 2019. Tech giants like Microsoft and Meta are trimming headcounts as they reorganize their businesses around the technology. Smaller firms such as Block, which is sort of a crazy mog. Smaller firms. But it is smaller than Microsoft and Meta.

42:00 Block, the owner of Square and Cash App, and Intuit, the maker of TurboTax and QuickBooks, are replacing people with bots. American companies have announced some 16,000 AI related job cuts a month on average so far this year according to Challenger Grand Christmas unemployment consultancy. What a funny name for for a consultancy. But AI related layoffs get lost in the churning job markets where so you remember that number. So 16,000 AI related job cuts a month.

42:31 In a typical month, The US workforce just churns 1,700,000 jobs. And then they add 1,800,000 jobs, so there's net gains, but you're looking at less than 1% of the layoffs that happen in the economy are job losses or firings, less than 1% is AI AI related, which is what The Economist is pointing out here. And the evidence so far is that AI is already creating a lot of jobs to replace those as as to replace it has destroyed. The vast sums pouring into data centers and power generation have set off a race for construction and infrastructure workers. AI startups are hiring like there's no tomorrow.

43:14 Incumbents are racing to keep up with new AI roles and making some workers more productive. AI may be increasing the demand for their services. Add it all up and The Economist estimates that AI so far has created around 1,000,000 new jobs in America. That easily exceeds the roughly 200,000 layoffs attributed to AI since mid-twenty twenty three, and it appears more than enough to offset weaker hiring in many back office roles. America's AI infrastructure splurge has created many of them.

43:43 The spending on the kit needed to make AI run from chips and servers to data centers, cooling systems, and power is roughly $500,000,000,000 a year above what it was in 2022 when the world got to know Jaggi Pit, calculates Goldman Sachs, a bank. Thank you. The Economist. Data center construction alone is proceeding at an annual rate of more than $75,000,000,000, nearly 60% higher than a year ago according to Census Bureau data. That building spree requires armors of workers, electricians to wire them, HVAC specialists to stop racks from overheating, grid engineers to hook them up to the power supply and technicians to install and maintain the machines.

44:20 The hiring boom is visible in the numbers. The Economist tracked five industries at the heart of the data center build out from electrical contracting to equipment manufacturing. Since 2023, employment in them has risen by roughly 320,000 more than broader construction and manufacturing trends would suggest. The BLS expects utilities to be the fastest growing big sector between now and 2035. Not all of those jobs owe their existence to AI.

44:47 Grid upgrades and other factory building matters too, but lots of them do indeed. A jobs website finds that data center vacancies have more than doubled in two years as job postings overall have fallen. LinkedIn, a social network for strivers. These are so funny.

45:05 Just firing. Goldman

45:07 Sachs, a bank. LinkedIn, a jobs website, a social network for strivers, estimates that nearly half a million data center jobs were created between 2023 and 2025 in America. The scramble for workers is showing up in paychecks too. Indeed finds that installation and maintenance jobs at data centers advertise wages about 40% higher than comparable work elsewhere. Official wage data tell a similar story.

45:32 In the year to June, average hourly earnings rose more than 13% in electrical equipment manufacturing and nearly 8% among electrical contractors. Even with the pay gains, workers are still not easy to find on a recent visit to a transformer factory. Donald Livins of the National Manufacturers Association asked the company's chief executive what help she needed most. Can you come in here and help run one of our lines for me? She replied.

45:58 It's not just hard hats that are proliferating. AI is also creating a new class of white collar jobs. Engineers build the models, data annotators. Tyler's clapping for the white label for the white collar job creation. Data annotators label their inputs and judges and judge their answers. Forward deployed engineers adapt them for customers. Newly minted heads of AI decide what companies should do with the technology. And there's some interesting reporting in the journal about Google and Accenture teaming up to do more forward deployed engineering. They they they're deploying some I think there's a thousand people on this one team. That's just an example of like new roles for AI diffusion.

46:39 Some of these roles barely existed until recently, says The Economist. Many are quickly growing in number. Postings for heads of AI, AI engineers, and directors of AI have roughly doubled since 2023, 2024. The numbers are starting to add up. Preliminary research by economist Gad Levnon at the Burning Glass Institute uses the research outfit's career history database to identify jobs that would not exist without AI whether at AI native firms or because they are AI specific roles at other companies.

47:13 He reckons roughly 1% of professional jobs are now AI jobs on the order of 1,000,000 positions in America. In computer occupations and life sciences including researchers using AI to discover new drugs, the share is four to 5%. LinkedIn's own analysis points to roughly 640,000 new AI specific jobs between 2023 and 2024. So Jevan's paradox. The Economist tracked employment in professional occupations closest to the AI boom, engineers, software developers, mathematicians, and data scientists, and compared their growth since 2022 with professional employment overall.

47:51 These roles have added roughly 730,000 jobs above the trend in recent years. AI will not have created every single one of them, but it almost certainly created quite a few. Third source of job comes from productivity gains. AI allows lawyers to draft contracts faster and analysts to comb through financial filings in minutes. If higher productivity lowers the cost of professional services, it is possible that demand for them can rise enough to create more work overall.

48:21 Nikita Beer was talking about this this weekend. He used Astra to create three different potential deck remodels, redesigns, building expansions to his home. And it was interesting because it's a lot of work that you could say is like, that's going to displace an architect. I don't think the models are at a place where you would trust a deck that's built in Blender from even the best model. I was finding in my blender modeling test a lot of misaligned beams and not things that weren't quite right.

48:57 But in terms of visualization, in terms of getting excited about a project and getting to that next stage of saying, hey, okay, I have a really solid vision here. I can imagine what I want to do. Let me actually go and do that. There's this odd diffusion that happens in the economy. I was talking to Sahir and Jetty about this, the credit card effect where credit cards had a big effect on the economy in a very, very boring way.

49:23 They just slightly lubricated the machinery of of the global Global commerce. Global commerce, basically. So you're just like, okay. If I if I send a wire, I'm not gonna be able to get the money back. But if I put down my credit card and it doesn't show up, I can probably get a charge back. And so that instilled just a little bit more confidence that what you're buying is accurate and and it and it enabled more commerce. And so the same thing is is starting to be true here where people can say, okay, I'm ordering shoes. Let me find the shoes that fit me perfectly, that are the right price, that will come on time, that are from a reliable manufacturer, aren't gonna fall apart. And if you get a very solid report back that gives you confidence, you make that purchase just a little bit sooner. And if everyone is doing that, it winds up speeding up the machinery of global commerce which is incredibly boring but incredibly valuable.

50:22 Totally. Because you're Yeah.

50:23 The one of one of my favorite examples is you you were looking at a number of different houses Yeah. All of which were not Yeah. Necessarily in the condition where you're like, great. I wanna, you know, move in here. Yep.

50:34 And you just dropped the Zillow links Yep. Into Codex Yep. It made you an entire website that you could walk through with before Yep. Before the current state Exactly. Then exactly how they would evolve Yep. Based on your stylistic preferences.

50:48 Exactly. So oftentimes you'll like a a fully remodeled house will command a 20% premium, but you can get sort of sucked into, okay, well, the newer the one that's already been through the remodel, it's just visually more striking. So you're like, oh, I want to necessarily pay up. When in fact you might want to buy the cheaper house, do the remodel and then wind up with the same end product and maybe capture some of that value yourself. And so there's a lot of different places, you know.

51:15 And a lot of times the the conclusion is very underwhelming. I mean, we were talking to Nick about this where he was taking a picture of his room and I saw a lot of people doing this online. Take a picture of the room, make sure I've decorated it properly, that the bed is in the right spot. And I think the result for you was like, yeah, it's basically as best you could do. Good luck, dude.

51:35 Which is hilarious. But at least it gives you confidence that, okay, yeah, like I really shouldn't order a bigger bed. I got the bed I got is the right one. And if you do that on the way in, you you you wind up moving things faster. We covered we covered Navier Stokes. I don't know if there's more to cover there. The chat's asking about it. We we covered it at the top of the show. I mean we're we also have Greg Brockman joining the show. So we'll ask him about advances in math.

52:03 What's going on with Navier Stokes, the back and forth there. He's joining at 01:10PM Pacific. So stay tuned. And we have Scott Wu as well who is an IMO gold medalist and a fantastic mathematician. He can of course comment about this because he at Cognition is in a unique position of both being incredible at math but also like Cognition's whole business is not in the theoretical math space.

52:30 It's in AI diffusion. It's in getting work done in the enterprise for real companies. And so I think he's in a unique position to both ground the conversation around what it actually means, how impressed is he by these solutions, and then also what does it actually mean for, you know, Boeing. Like, will they be able to make more planes more efficiently, more reliably? Like, that is a valuable thing that we actually want to see in the real world.

52:59 So.

53:00 Something happening in the real world.

53:02 What's up?

53:02 Former f one CEO Bernie Ecclestone Oh, yeah. Was stopped in Portugal at the airport for bringing in a shotgun from Switzerland. Woah. He's 95 years old.

53:15 Mhmm.

53:15 And this is actually the second time he's been No way. Stopped at an airport. In 2022, he was arrested in Brazil for illegally carrying a gun while boarding a private plane to Switzerland. Mhmm. And was ordered to pay a thousand euro fine to local authorities.

53:31 So rough weekend for Bernie, but at this point, I I think he's like, sorry, I'm I'm 95. I'm not gonna stop bringing guns. Everywhere I go on on on planet Earth.

53:47 I gotta do it. Well, let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents. Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex helps you move projects forward from start to

53:59 Dean says, just like grandpa carry his gun. That's all. Think In other news, Wimbledon does not plan on providing credentials to influencers next summer in hopes of avoiding the issues that affected this year's US Open.

54:12 What were the issues? We didn't influence

54:14 Players have called for spectators to follow the sports etiquette after matches were interrupted.

54:19 Did they specifically call out taking a picture of the court and putting logos all over it using AI and then posting on on X and getting a couple 100 likes.

54:26 Looking at you, Nick.

54:27 He's locked in.

54:28 Looking at you. Working.

54:30 He's working. We're just making jokes at your expense. Don't worry about it. You're good. What what was there anything specific that was tied to this?

54:42 I I think it was a number of issues. There were also complaints of of one of the players was complaining about the smell of marijuana

54:50 Oh, really?

54:51 In the stadium which is interesting. Yeah. But but yeah, next year Tyler will be going to the US Open in disguise. So Full Hollywood makeup? Yeah.

55:03 You can look forward to that.

55:05 Yeah. I wonder what what what are they gonna do? Just not give free tickets to influencers or actually block people who are influencers?

55:12 I think what was happening is they were probably giving like media slash trespasses to influencers.

55:16 I think the influencers were like doing too much stuff Yeah. Like getting in the way of fans Yeah. Creating scenes because they're sort of like running around with like Gonzo filmmaking equipment as opposed to like something that's a little bit more cordoned off and like It's different when you have a red carpet and you're doing, proper interviews. That sort of lands a little bit better. Anyway, Shiel Monat is calling for a revisitation of the Setrini piece.

55:43 Agents become the default demand side interface and transfer economic rents from incumbent intermediaries to whoever controls the agent, which does seem correct. And so with instinct, who should you be long, who should you be short. There's an interesting dynamic where anytime something happens in AI, there's a lot of people that go. I I think Buko posted pointed this out that everyone will go and ask AI, okay, what are the longs, what are the shorts? And so that creates an amplification in the market.

56:14 Oh, piece together this new trend. I want exposure to this trend. And so things get overheated much more quickly. But, yes, this idea that if it's a restaurant booking platform and your AI agent can call the restaurant and get a booking just as easily as they can go through a platform. Now Resy has exclusives, so that creates more of a complex scenario.

56:36 But by default, if you can just write an email or make a phone call or send a text message, an AI agent should be able to disintermediate those platforms. But that's not true for Uber. That's not necessarily true for DoorDash where they're like, it's very difficult to find a person to go and pick your food up or pick you up personally because they need to be very close if you want the car right then. So, Shiel says, how after you in an agent disintermediating world completely depends on how much you control how much control you have oversupply. To answer your question, Buko, you may not have a choice in whether you allow bots.

57:14 Expedia can book agent can block agents, but it doesn't control the hotel inventory. If Expedia stops letting agents book, they'll go through booking.com, Google or Direct. Resi has more leverage via exclusives. I'm sure Expedia has some exclusive program, I imagine. But if agents become a major source of diners, restaurants will want to be bookable via agents and reservation systems.

57:35 The support agents will win supply slash someone will come along with a cheap direct booking sell. AI can disintermediate discovery and customer ownership, but Resi can survive. But can Resi survive as the restaurant's reservation infrastructure? But that's a much smaller rent pool unless the inventory network remains differentiated. Uber has much hard is much harder to disintermediate.

57:56 It actually controls a valuable network of drivers, dispatch, pricing and payments. Why would they allow agents? It's basically they can afford they can't afford not to. After an AI agent takes over your consumer interface, how much economically essential stuff are you doing that only you can provide? For Expedia?

58:12 It seems tough to me. So, Sheila is worried about Expedia. Anyway, was there anything in the chat you wanted to run through? Because we have our next guest joining. In just a few minutes, have Dan Wright coming back on the show from Armada building modular AI supercomputers in a shipping container. A supercomputer in a shipping container. Apple's coming out with new phones. We're getting a folding phone probably Finally. This week. It's gonna be exciting.

58:37 I think it'll sell well. And it's been a while since there's been a visually Yeah. Differentiated Apple product.

58:42 So Yeah. I think it's a pretty easy sell. Hey, you use this thing for three to eight hours a day. Yep. Or if you're Taylor Lorenz, eighteen hours a day.

58:53 It should be some

58:54 crazy Would you like a bigger screen? Yeah. I think a lot of the answer for a lot of people. And would you like to have some differentiation versus your peers that have just one simple screen? Yeah.

59:04 I think a lot of people are gonna opt into that.

59:06 I wonder how it's gonna look. I've seen a lot of like leaked renders.

59:08 It looks and A lot of the renders look squarish.

59:11 Yep. Not good.

59:12 Just throwing me off.

59:13 Yeah. Not I I feel like they they're gonna surprise us with something. They've they've it feels like the the news and the facts and things that can go over text message sort of leak out and Mark Gurman is like I saw a stat from Eric Newcomer that Mark Gurman has an order of magnitude more scoops than the next biggest journalist on tech meme or something like that. Yep. He's like the power law tech

59:39 You hear that?

59:40 Ghost freak goat. Seriously. But it feels like people A

59:45 much journalistic force. Discuss things with

59:47 Mark and say, oh yeah, we're not working on our car anymore. Keep that off the record, but I'm letting you know and he'll report it out. But people will stop short of sending him an image of the new phone and I think he's been pretty good about that.

1:00:02 Yeah. It was actually Josh over who says, we don't talk enough about how Mark Gurman is the most cited, most market moving tech journalist by nearly an order of magnitude. Yeah. The whole industry of Apple reblogs and day traders live off of his scoops. It'd be like if LeBron won 40 championships.

1:00:22 That's a that's a really good post. We love Mark Gurman. Well, there's one more story we gotta cover before we go into our guests. A man drove a Ferrari Purosangue from South America to Alaska. This is a crazy, crazy story.

1:00:41 This, the Transamerica race has been talked about a long time. It's really, really difficult with the Darien Gap, the gap of like wild forest that exists between North And South America. You usually have to put your car on a boat to get around that part and I think it still counts. But it's still a very treacherous road, a lot of weather conditions. If you break down, you might be very far from anything.

1:01:03 But a man named MJ has taken a Ferrari Purrissong way far beyond where most owners would ever consider driving one. His goal was to travel the length of The Americas, taking the v 12 Ferrari from South America all the way to Alaska. He's now attempted the journey three times, putting roughly 800,000 kilometers on the Purosangue in the process. His dog joins him for the journey. The first attempt ended after an accident.

1:01:29 He brought the car to a dealership in Santiago. On his second try, they made it all the way to Alaska. For the third, he reversed course with an even bigger goal attempting to set a record from Alaska to Argentina before a major road closure forced the run to end in just two days. Through dirt, snow, and thousands of miles of pavement, the Purosangue has effectively become a long distance expedition car with Rico, his dog riding alongside him through it all. What a

1:01:58 Heartwarming. Heartwarming story And a beautiful ad for this naturally aspirated v 12.

1:02:04 For sure. For sure.

1:02:06 At that shot. So good.

1:02:08 Oh. I didn't realize it was Chihuahua. That's funny. That's probably a more convenient dog to have in a in a car for a really long. You don't want a big dog that needs to stretch his legs.

1:02:16 You want

1:02:17 the dog to be able

1:02:17 to just run around the cabin, I think. If you're it's it's a good good road trip dog, I think. Anyway, we have our first guest of the show, Dan Wright from Armada in the waiting room. Let's bring him in to the TBPN UltraDome. Dan, how are doing?

1:02:32 There he is.

1:02:32 I'm doing great. How are guys doing?

1:02:34 We're doing fantastically. Welcome to the show. Welcome back. What is new in your Walk us through the latest. And maybe just, since it's been a while, of reintroduce the the shape of the company, the mission, and, where you guys have gone in the last year.

1:02:51 Yeah. So Armada is the hyperscaler for the edge. We build the infrastructure for the 70% of the world that doesn't have the big hyperscale data centers today. Mhmm. The mission of the company is to bridge the digital divide and make sure that we have AI everywhere, wherever wherever we need it.

1:03:06 But first, you got to have the infrastructure there. And so that's what we're doing. We've been very busy. We just raised earlier this year 230,000,000 at a 2,000,000,000 pre. We launched Galleon Forge One, which is a factory with our partner Johnson Controls in Gilbert, Arizona, where we're now continuously manufacturing these modular AI data centers called Galleons that we built.

1:03:29 Yeah. How how is power usually solved for when you deploy these?

1:03:34 Yeah. So a lot of times we're deploying them where there's already stranded power.

1:03:39 Sure.

1:03:39 A good example of this is a couple of weeks ago, I was in New York, and we did an event at the New York Stock Exchange where we were sort of fast following on Jensen's announcement that compute is now an asset class. It's a new asset class. And the New York Stock Exchange is actually making that now like something that you can invest in the same way that you can invest in electricity or other types of commodities. And we were there with our customer in Norway called Fossifall

1:04:09 Mhmm.

1:04:09 That has tons of distributed sites all over Norway. I'm actually going be in Norway with them next week, but they also have them in Finland and in Sweden. And it is largely renewable energy, so it's 99% hydroelectricity. They can just plug in our AI factories and then scale up quickly. And one of the things that, you know, makes that easier is that we've recently announced some larger form factors.

1:04:37 Last year, we announced Leviathan, which is two megawatts per unit. And when I was in New York, we announced Orion, which is our newest form factor. That's 10 megawatts per unit. When I was, you know, launching Leviathan, everybody said that's great, but how do we scale up even faster? And so now Armodic incredibly say, wherever you have power, we can plug in and we're the fastest from zero to 200 megawatts anywhere in the world.

1:05:02 Talk about why, for example, you know, one of these companies needs to have that compute actually at the edge and why it matters across different industries?

1:05:13 So one is latency, and this comes up in a lot of conversations. I'll give you another, real world example. We work with the state of Alaska Mhmm. And they don't have the big hyperscale data centers there. So the first data centers deployed were ours.

1:05:29 We did this last year, and we're still working closely with them and kinda scaling up with them. They had twenty eight hours of latency to process data from drones Mhmm. For emergency response, for avalanches and floods. We also deployed last year with the Navy in the middle of the ocean, very similar types of use cases where if you have a large distance between the source of the data where the data is being generated

1:05:54 Mhmm.

1:05:54 And then where it's being processed, the data sort of becomes worthless. Mhmm. Because avalanches and floods and threats and battlefield scenarios, they don't wait days or even minutes. You have to be able to use the data in real time. Mhmm.

1:06:08 Another big drive driver of this is sovereignty. The the shorthand for our value prop is the three s's, speed, scale, and sovereignty. And the sovereignty piece is really important. There's this global trend that's going on around sovereign AI. Everybody wants to be able to take the latest models, but they wanna be able to fine tune them to really sensitive data sets that they wouldn't send to the cloud.

1:06:31 And then, you know, have, like, a sovereign AI infrastructure that they actually own, and that's what Armada enables.

1:06:38 So, yes, speed of delivery. Yeah. That makes a lot of sense. Is hydroelectricity under discussed right now? Is there an opportunity there in America or abroad?

1:06:48 I mean, we talk a lot about solar where it feels like there there aren't as many, like, big winners yet or big, like, hot startups. That exists in nuclear. That's very exciting. I've you know, we've heard about wind, but no one is really talking about, like, let's just do another Hoover Dam or something. Is that possible?

1:07:08 Are you optimistic about

1:07:09 I think so. I think I think it should be talked about more. And I think in general, stranded energy should be talked about more. We're doing these projects all over the world as an example. You know, the the Niners are playing in Australia this week, and we're doing an event there with another partner called WindDC that we work with there.

1:07:30 And they've got a huge amount of stranded wind and solar energy. Australia is big in wind and solar. Norway is big in hydroelectricity, but this is happening all over the world. There's these stranded pockets of energy. And they, last year, had to curtail 7.2 terawatt hours of energy in Australia because the grid's completely overloaded, but they have this stranded power that we're just bringing the infrastructure directly to, and then you can scale up into the hundreds of megawatts.

1:07:57 And then ultimately, same with the fossil fuel, they'll they'll scale to over a gigawatt over the next few years, and we can scale with that.

1:08:03 How flexible do you want to be around, like the actual chips that go into the systems? Because when I hear latency, I feel like Cerebras, Croc, these like faster systems. There's Talos, which I think AMD just bought, where you're baking the weights on. And you can actually get to a an inference speed for certain AI workloads that would actually benefit from saving a hundred milliseconds. Whereas if you're putting a bunch of NVL 70 twos in a data center, you're going to be waiting while it's cooking.

1:08:38 Right.

1:08:39 And the last mile is going to be negligible.

1:08:42 Yes. So again, it's all about speed, scale, and sovereignty and then giving the customer the choice. So we say, okay, well, what workloads are you trying to run? Yeah. And then based on that, you can sort of right size the the infrastructure to the the actual need and where they expect it to go.

1:08:59 And a lot of what our customers want to do is they want to try different things. Mhmm. They want to try, you know, obviously the GBs, but now they're looking at the Vera Rubins. They're looking at things from other, you know, chip companies as well. And so the nice thing about our our galleons, again, we're manufacturing these.

1:09:17 It's not like construction. And so as you want to try different things, we can just make, you know, quick iterations on the design and ship them out. So they can try different things, and then what they like, they just order more of it and we scale up with the demand versus with the traditional data center. Yeah. The downside is not just as we were talking about the power.

1:09:34 You gotta figure out the power situation. We take advantage of the power that's already there, but what we also allow them to do is scale up with demands that they don't overbuild or build the wrong thing.

1:09:44 How are you thinking about, you know, staying aligned to, I guess, like, why and and how the company started and the focus around this edge compute versus the natural pull from the market where if somebody comes to you and says, like, okay, now I want the one gigawatt data center and we think you guys can can do it. Like, just running the numbers on that and and resource allocation, I imagine there's a natural pull to go bigger and bigger when and and anyways, I think that's kind of

1:10:13 Yeah. Interesting. Was actually

1:10:14 problem good problem to have, but will be interesting.

1:10:17 Actually super simple for us because one of our company values is heal the customer's pain first. We're, like, obsessed with customers and what their what their pains are and how do we solve those. And so the the kind of evolution of our Galleon product line has all come out of conversations with existing customers where they've said, okay. I want to use your edge galleons, the smaller ones, say sub sub a megawatt, for inference. But then they came to us and said, I also wanna do fine tuning of all these models, latest open source models, closed models, my own models on these proprietary datasets without sending them to the cloud.

1:10:59 And so what we enable is Sovereign AI factories where they can do both. They can take the latest models, open source models. You probably saw what, you know, NVIDIA just did with Hugging Face and with Poolside. They were gonna see a lot more around open source. They can take models from the OpenAI's and the Anthropics of the world.

1:11:17 We can help them fine tune those to the sovereign data sets and then push them to all the edge nodes to run, and then you can do what's called federated learning

1:11:25 Yeah.

1:11:25 Where you're actually fine tuning that model on that data at the site, and then you're using it to improve your core model and then pushing the updated model out to all the sites Yep. Which has the benefit of it's more cost effective Yeah. It's more secure, and it enables you to take advantage of lots of different models, experiment with different things, and see what works.

1:11:48 Yeah. Well Very cool. Congratulations on the progress. Thank you so much for coming on.

1:11:51 Great update.

1:11:52 We'll talk to you soon. Great to

1:11:53 see you guys.

1:11:53 Great to

1:11:54 see you, Dan.

1:11:54 Congrats on the Let

1:11:56 me tell you about Shopify. Shopify is a 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. Who do we have next, Jordy?

1:12:08 The Seufert.

1:12:09 We got Eric Seufert, founder of Mobile Dev Memo live with us on TBPN, on YouTube, not on Netflix yet. Should we be considering switching teams?

1:12:21 And Eric, just just just for for what it's worth, every time you you you come on the show, I call John afterwards. I'm like, my top three all time guests on the It just fires us up to We

1:12:34 love talking

1:12:34 to talk

1:12:35 with you every time.

1:12:35 Good to see you.

1:12:36 And it's great to

1:12:36 see you. Yeah. Good to

1:12:37 see you. So I I told my wife Mhmm. I'm going back on TBPN. Mhmm. And she said, is that the show with the two handsome hosts?

1:12:45 Oh. I said I said, yeah, but when I'm on, there's there's three

1:12:51 handsome Yeah.

1:12:52 That's right.

1:12:53 And it was like that meme with Natalie Portman where she's looking at the guy and I was having to say, right?

1:13:00 Right. There's like three

1:13:02 handsome guys on the show, right?

1:13:03 Right. Right. I I

1:13:04 think you're ready for Netflix. I think Netflix would to have be

1:13:08 But what would the what would the economic dynamic be? Because Netflix and YouTube are both going for exclusives now. What's actually playing out? How much are they at each other's throats? How much are they going to converge in the way that like TikTok and Instagram and YouTube are all converging?

1:13:25 Is there gonna be another convergence or is there or is there actually a stress point there where YouTube is trying to be both Instagram and Netflix and they maybe can't do both?

1:13:34 Well, yeah. I mean, they are converging. So this is a really fascinating dynamic right now in streaming. I've been following it with a series called Netflix's YouTube opportunity for roughly a year. Right?

1:13:43 And so what I read about first was they brought miss Rachel on. Right? And so what they did was they brought miss Rachel on, but they didn't give her they they didn't they didn't buy new content. Right? They just paid her to bring her existing content over. So you've got this proven base of content that has however many billions of views. I know my children probably account for some several billion. But you've got this proven case of you've this got proven base of content where there's essentially no risk. Like, you know there's an audience, you know it's popular, and you pay some amount of money to bring over this existing content. So what's actually really great for the content creator too because they don't have to produce anything new.

1:14:17 They just have to chop they chopped up the existing content into like a season. So they packaged it like a season. And they also did that with Danny Go and they also did that with Mark Rober. And so my point was, look, they've you know, Netflix had been opt they've been executing tremendously well for the past several years. They had had they had commanded extreme pricing power.

1:14:35 Right? The the premium service is going at almost $30 a month in The United States now. They've been they've been the the advertising tier price has been tracking with, I think, is the ARPU differential to make it equivalent to the next highest tier. So they've been doing a good job of maintaining that pricing power. My sense is they did too good of a job.

1:14:51 Right? And so they made it really hard to compete in the space, and that pushed everybody to the bottom. That made everyone take the exact opposite approach. It's like, okay. Let's go fast. Let's go free ad supported television. Now Netflix has an even bigger problem on its hands. Right? Because they may have reached a ceiling with what they can charge. Certainly, premium.

1:15:08 I don't know how much higher you can go than $30. Mhmm. There's not much they they probably extracted all of the net additional subs they can from password sharing crackdowns. Now, they're faced with, okay, we just have to bring in a lot of live events and sports, things like the Beyonce Bowl, they've you know, and and the the the January NFL games. It's really expensive.

1:15:28 Right? And so are you gonna compete on content, or are you gonna go the opposite direction, race to the bottom, and try to get u UGC on there? And I think that's probably what they're they're they're trying to sort of thread that needle. Because if you listen to Netflix's leadership, they say, look, we can never go pure fast because that would have a deleterious effect on the brand. We're seen as a premium service.

1:15:47 But how can you maintain that when you see Roku and Tubi joining forces? Right? When you see all of the pressure that Amazon is putting on your business. Because Amazon is the identity spine for advertising across all of fast. All of fast CTV is Amazon with these data deals doing all the all the identity for the buying.

1:16:07 Right? And so I think they're facing like this opposite problem now where they had too much pricing power. They pushed everybody in the opposite direction to go fast. And now I think, honestly, if you believe the reporting that New York Times did that they're gonna be offering up these subscription bundles, how can you not go fast and do that? Are you gonna be willing to accept paying $30 a month for a premium Netflix tier, seeing content in your Netflix app on your TV, and having it say, and if you want this, you have to pay even more to subscribe to this other channel?

1:16:33 I wouldn't. I think the only way they can do the bundling is if they adopt fast.

1:16:37 That ARPU gap from the premium ad free tier to ad supported, it feels like are they is there a reason why the gap doesn't make up the full $30? Is there some economic reason or did they just get stuck in this weird place because it feels like if they got even a little bit more juice out of the ad model, they could just offer a free ad supported tier. But maybe that would create more churn because mentally you're going from $30 to $10 is different than going from $30 to $0 But do you understand more of like what they're grappling with? Did they get unlucky or is there something about the the structure of Netflix that would would would really make it difficult for them to do just a fully free and ad supported tier?

1:17:29 No. I think it's just a reservation they have. Maybe it's almost like a superstition, right? Mean, keep in mind, they were saying for the longest time they would never do ads. Yeah.

1:17:36 You know, Reed Hastings said like They always do.

1:17:40 Yeah. They

1:17:40 always Until until until they change it. Like he I mean, Reed Hastings said ads were like a blight. He said personalized advertising was a cancer. Yeah. And so I mean, know, then they they changed the tune.

1:17:50 But I mean, I think he said just that, if I'm remembering the wording correctly. But but the thing is like, I think the so there's there's three tiers, right? There's there's the it used to be basic. Now it's standard with that standard and then and then premium, right? So the standard is what they have to make up their ARPU gap with.

1:18:04 Premium includes a lot of stuff. And so maybe you can make the case. Right?

1:18:07 For a

1:18:07 high income household, it's worth paying for the higher streaming quality and the more the more devices that you can onboard. My sense is with standard though, that that revenue that ARPU gap is just they they just they just index it to the ads ARPU. But the the thing is like, the question is like just internally, what is the resistance to going fast? And my sense is it's that perceived quality that they would lose if they went fast. But I mean, what can the market moved in that direction.

1:18:31 They pushed the market in that direction. And so I think they're going have to capitulate.

1:18:34 Have you been surprised that Amazon Prime Video hasn't seen more UGC? Because I was looking up randomly what does it take to publish a film? And you can submit to Amazon Prime Video that goes through a review process but the cost to submit is de minimis and you can get a video product up there but it hasn't actually seen a groundswell. And I'm wondering if that's because Amazon specifically has partnerships that scratch that itch.

1:19:09 Yeah. I think that might might be the case. I mean, keep it like like I said, I mean, Amazon has all of these identity partnerships Sure. That essentially make it the data spine and the identity spine for fast,

1:19:19 like broadly.

1:19:19 Right? And so if you think about I remember people saying, oh, I bet Amazon's gonna buy Roku. And I remember thinking like, why would they? They've already got this identity partnership with Roku where they get access to the best impressions. If they bought it, they just be getting everything else that they don't already buy.

1:19:31 Mhmm. Right? And so my sense was always that, like, well, they they're already picking over the stuff that they're not buying. Why would they wanna own it? Mhmm. And so, like, my sense is, yeah, maybe it competes with the fast channels that they're partnered with or maybe maybe they have similarly like, you know, some sort of perceived quality bar or hurdle. But but I I think like the the UGC thing is is is a little bit of a distraction because, you know, I don't think Netflix is ever going pure UGC. Like, they are very selectively curating the YouTube creators they bring over. But they brought over a lot and they went on a shopping spree this summer. Yeah.

1:20:01 They bought a lot of stuff this summer. Or and and now now YouTube's pushing back.

1:20:05 I saw some post about how the results of that of that shopping spree that Netflix went on with YouTube creators had a significant power law where there were some creators that were really performing well on Netflix. Is that because of the way the Netflix algorithm works? Is that true? Or do you think that there's a Okay. If somebody's getting a billion views on YouTube, we can pull them over and get x views and there's some multiplier.

1:20:30 Or is there something special about the way Netflix is, like, actually rolling out these deals?

1:20:37 Well, I think there's a there's a couple pieces there. I think like they've got YouTube certainly has like more Rex's adjacency. So they have more opportunities to push people into this content than Netflix does. Netflix has less stratified, less deep catalog, right, than YouTube does. Yeah.

1:20:50 So that's that's that's one piece. But then you do you just you do just see, like, even with miss Rachel, like, two performed substantially worse than season one in terms of view hours per per minute of of of content. And so, you know, it's maybe it's just consumer preferences where they had other stuff to prioritize. I mean, Netflix has invested a lot into its own recommendation algorithm. And so it just it depends on like what they they they feel the need to promote at any given point in time and they they have to, you know, they they just have to exclude stuff that doesn't fit that purpose.

1:21:18 Yeah. Have to imagine that someone like Miss Rachel, if she's doing a deal with Netflix, she as part of the deal is not just going to be money, it's going to say, hey, are you going to actually give us impressions so that we have the chance to be successful here and get real viewers that go into the rest of the funnel, show up to the live shows, buy the merch. Like, it's not enough for you just to pay and stuff us off in a corner. If we do this, we want to

1:21:38 do it I wonder I wonder if Netflix has an opportunity to just basically take all the best UGC from YouTube because I think YouTube has a massive Yeah. That's that's seemingly what they're doing. But underappreciated maybe how big of like an AI, like AI is probably like seemingly a net good for YouTube right now because there's more

1:21:58 You mean specific all

1:22:01 the the long tail. Like something will happen Sure. And I'll get served a video and the voice sounds good now, maybe the script sounds okay, and then you realize like, okay, this channel was just created two months ago and it's not really. And and YouTube has an incentive just constantly be serving new creators.

1:22:18 In theory, Netflix and human

1:22:20 I review generally want to consume content from creators that have been are are either new and extremely passionate and dedicated. I even saw a creator over the weekend that that started years ago using like AI voices and just stopped and said Oh, interesting. That YouTube is so flooded with AI content now. He's just And he's he's not American, but he he makes all of his English isn't his first language, so he has an accent. But it's appealing because you're like, okay, this guy's actually making this content.

1:22:50 It's not just like fully generated. Right? And so I think YouTube or Netflix's opportunity is to like try to carve out all like actually have a filter again and like be a curator and take the the best content that historically would have just been on YouTube and and try to bring it over.

1:23:09 Yeah. But it's still big.

1:23:10 Well, yeah. But I mean, so YouTube absolutely does not want that to happen. Right? So I mean, keep in mind, like, the dominant platform for YouTube in The United States is the TV.

1:23:17 Yeah. Yeah.

1:23:18 Right? And so so they they are I've called it a CTV behemoth. That is their dominant platform by by Viewtime. And so they they compete directly with Netflix for engagement. It's actually really problematic if Netflix is able to just poach their best creators, the the top, like the the sort of like the the cream of the crop, and then push them into Netflix and and then to justify the subscription price.

1:23:37 Right? So like, what they've done is they've they're now inking deals to keep people exclusive for a period of time. And they're also saying, look, we're gonna punish you if you move. Mhmm. We're gonna deprioritize you in our recommendation systems. We can't promote you if you're on Netflix because we wanna promote stuff that's exclusive to us. And so, you know, and you're not gonna share in brand revenue. So it's actually what Netflix has done is they forced YouTube to apply a lot of the curation pressure and incentives that YouTube always didn't want to do. They always wanted to resist being that kind of channel. They said, look, know, it's just look, this is just an open market.

1:24:09 Like, people compete and you get you get views if you outperform. But now, they're having to sort of put their thumb on the scale in certain ways to keep the best creators. Mhmm.

1:24:17 Yeah. Mask very much a mask off moment for the for the platform that wants to position itself as like, we're just this friendly platform for creators and we just want to support creators. And they're like, if you even put one of your videos over here, you're not getting any more money and you're not getting any more views. Yeah. And we will end your career.

1:24:35 Like that's basically that's basically Be a shame if

1:24:38 something to your audience.

1:24:39 Yeah. Be a shame if if something were

1:24:40 to happen

1:24:41 to your reach.

1:24:42 What is going on at Apple around services, ads? Give me a little bit of the history there. Apple's obviously had some some sharp words about advertising in the past, wound up building

1:24:55 a great

1:24:55 ad platform.

1:24:57 So John sent me a screenshot. It looks like It's from MobileDevMemo. It's from MobileDevMemo. And the screenshot says the Shiller exit is a bit more notable. Apple's app marketplace is loved by many consumers but often criticized by developers and subjected to increasingly onerous regulations, but there's a bit more to the story.

1:25:14 I'm told Turnus and service to cheap at EQ wanna make even more money from the App Store and figure out how to ways to raise margins and squeeze additional recurring revenue from the platform. Shiller on the other hand seems to believe that such moves would only further irk developers and governments. And John said, first piece of good news from Apple under Turnus, Cupertino will finally be focused on App Store margin expansion. I've been pulling my hair about out about this for years. Feels good to be vindicated.

1:25:41 Better late than never. So obviously Yeah.

1:25:45 That that quote's from German. That was quoting German. Got it. Yeah. So that was from German's newsletter, Brock.

1:25:49 But yeah, we got a new a new era, right? Yeah. Shill it out apparently. Yeah. I stepped away from the App Store. Eddie Q is taking over with Turnus. Yeah. And apparently, I want to squeeze more money out of the app store, which makes sense because it's under monetized.

1:25:59 But why not ads? That is that can't be

1:26:03 anything but ads. Okay. No. It can't be anything but ads. What else could it be?

1:26:06 I asked this question on Twitter yesterday. Yeah. What else could it be? Yeah. What could it possibly be besides ads? What other opportunity is? Are they gonna increase the commission? You nuts?

1:26:14 Yeah. No How

1:26:15 could they

1:26:15 increase the commission? The the

1:26:16 the commission direction. You said it's under monetized, but you believe it's under monetized purely on the ad side. The commission side is people's totally at their limit. Totally.

1:26:25 Yeah. Yeah. Commission can't go anywhere. Certainly, it's not going to go up. They'd lucky if they can maintain the commission.

1:26:30 Yep. I mean, they're probably going to

1:26:32 have to know, so basically, I I think what I think is going happen, so they replied with their in the Epic v Apple drama, which is interminable apparently. They made their latest proposal, which is that they'll apply a 15% commission on the link out. Right? So they're they're having so in if you remember, just go back a little bit, the Epic v Apple case. Epic essentially lost on every 10 of the 11 points.

1:26:55 But what they did went on is that Apple does have to allow link out. Right? So I mean, you're in the app. The developer can put a link in the app to a website that allows them to monetize there. Right? So they have to be able to do that. Now what Apple responded with was saying, okay. Yeah. But we're gonna apply a commission, that when you add in the Stripe fee is essentially just as much as 30%. And also, have to do all this reporting.

1:27:15 And also, if you have someone clicking out and going to a website, they're probably not gonna be as likely to convert. Right? Yep. So apply all this commission all these frictions that, like, okay, it's never gonna there's never gonna get any traction. Now, what they've proposed is like, okay, we'll accept a 15% commission on this, but you still have to go through all the reporting.

1:27:29 Yep. That makes it interesting. Now, we'll see if the judge accepts that. Right? They she might not. But like, Epic was fighting this. They want zero commission on link out, but Epic wants zero commission on anything. Yeah. But so like, if if that happens and the commission drops even further, you're gonna push a lot more revenue outside of the App Store onto the web, which actually a lot of it's already fled to the web. I mean, like, go to any subscription app.

1:27:50 Go on the Facebook library, find your favorite subscription app, Strava, whatever. Look up their ads and see where they link to. I guarantee you it's to the website. Mhmm. Almost every subscription app spends the vast majority of its of its advertising on web web destination ads. Yeah. So they're sending you to the web, you you register on the web, and then you download the app, you log in. And so all of the monetization has happened on the web. There was never any monetization in the app in the first place.

1:28:13 Mhmm.

1:28:14 Games have started adopting that too. And so they've already lost a lot of this in app monetization to the web. The commission's already under a tremendous amount of pressure. And it's not you know, you've got the EU with the DMA. They seem to have come to a resolution there, although, you know, Epic also thinks that's not true.

1:28:31 But you've got Japan, you've got Brazil. I mean, you're gonna see increasing cases where government step in and say you have to offer alternative in app monetization and you have to offer alternative app stores. And so if you think about what does the app store have that can't really be taken away, that's probably under monetized, it's a lot of engagement. It's if I remember correctly, like the last touch point where they released any data was 600,000,000 no. It was 900,000,000 weekly active

1:28:58 users. Mhmm.

1:28:58 So they Jet GPT reached parity with that at some point a couple like roughly a year ago. 900,000,000 weekly active users. That's a lot of engagement. And what do you do when you have a lot of engagement, a lot of eyeballs, lot of attention? You monetize it with ads.

1:29:10 I think they have

1:29:11 But with that get Yeah. With the app store, every time I search for something, I'm seeing a seemingly seeing a bunch of ads.

1:29:18 Yeah. But can see way more. Like imagine you go to a website, you go to an app in the app store, you're browsing, you don't purchase the app and then the next morning when your alarm clock goes off

1:29:27 notification.

1:29:28 No. Just replace the alarm clock sound with an ad that says please download this app or you're ready to go. Retarget me the day. Open up the iMessage where

1:29:40 I was going with this is at what at what point does Apple, if they actually want to grow their ad business, not start to compete with an app loving and offering like bringing their bringing their ad network into the apps. Right?

1:29:55 Well, I mean, they do, essentially. I mean, they've got ads in the search results, right? They just added a second slot, the second placement in the search results recently You a couple months know, they've got like recommended apps where

1:30:07 you can

1:30:07 find the placement in an App Store page.

1:30:09 They could do like you have to pay to get organic results. Like you search and then it's just 10 ads.

1:30:14 That's

1:30:14 an And then unlock, unlock, unlock Oh, premium tier. Unlock organic results. It's a dollar and 99¢ to see what you're actually searching for. No. But what what I was saying is like, you know, like basically a user comes, they search for a game, let's say.

1:30:30 Yeah. Then they see an ad, but then they end up downloading a game. And then I'm saying like all the surface area that's potential like the real opportunity is every time someone opens up that game, they're getting an Apple app.

1:30:41 And No. I think they will. I think they should. I think they will. I think they will.

1:30:44 I don't know what else they could do. I mean, I think they're gonna do that. I think they're gonna have some sort of e comm ads offering. I mean, I predicted a year ago that they put ads in maps cause it just made sense. They've got a lot of engagement there and they they just did that a couple weeks ago.

1:30:56 But keep in mind, I wrote about this when they so they took their ads service down for a while. Usually, they do that when they're making a change. Right? Mhmm. Right before they did that, and that's when they implemented ads in maps, but they also they they they created a whole new campaign optimization API to accommodate Maps now and all the other places that they have.

1:31:18 So this unified campaign optimization API. Right? Now, it's all that's extensible. You could extend that to anything. Right? So they're already putting in place a scaffolding to support that, but they're also putting in place the policy scaffolding. So they also made all advertisers, like, sort of recertify their approval of the of the Apple ads service agreement. And that gives them permission now to serve ads on websites and apps apps they don't own. And I think that was the big sort of signal here. Now, they've done a lot of other things too.

1:31:50 They rebranded SKAdNetwork to the ads attribution kit. They renamed the whole thing from Apple search ads to Apple ads. I mean, all of this sort of points towards a more generalized ads product. But I think opening up that services agreement to say we can place ads on third party properties probably does signal that they intend to do that.

1:32:05 Yeah. Yeah. I mean, that seems like a huge way to grow services revenue. I I don't know if you have the exact size of the product.

1:32:13 Ads come for everyone. About Eventually.

1:32:17 What about new Siri? I mean, looking at the Chatty Pity ads hitting a billion dollars, you've been very optimistic about the potential of ads in LLMs and that it's a logical end state. Do you think that they'll do Siri? What what else are you tracking in the development of AI and ads?

1:32:38 Well, I mean, we've talked about chatbot ads. Sure. You know, I have a lot of thoughts on that. But I mean, I could just go back to Siri. Siri's tough because it's only voice.

1:32:44 Right? And so I think you need There is an

1:32:46 app where you can talk.

1:32:47 Yes. That's true. So That's true. That's true. I think once you have the chatbot interface and do have that.

1:32:51 I think once you get a lot of usage there, ads become an opportunity.

1:32:54 Sure.

1:32:54 I just do think you need a visual component to make ads work. Yep. It's also why I just don't think ads to agents is ever gonna take off. Yeah. Think there's like a lot of incentive conflict there.

1:33:02 I just don't think that you can show ads to agents because you you wouldn't know who to trust. Yeah. And who's getting paid in that case? And who's paying? Right? So that's there's there's like a lot of incentive conflicts there. But I also just think there's a there's a there's a visual component to ads that's necessary. Right? Because that's, you know, that's what ad creative is. That's the that's the principal way that you do messaging, that you do brand positioning, that you you create affinity.

1:33:23 And so I think if you're just doing voice, it's really tough. I think in the chatbot experience, you can. My sense is how they wanna have I think how they wanna monetize AI generally on their hardware is is is through essentially a a deal similar to what they struck with Google Search. Right? Like, you have to pay to be the model that gets, by default, gets attached to these services.

1:33:45 And I think that that could be very, very lucrative to them. And they've already sort of, again, they've already set up They've already kind of created environment for that to happen with this sort the core AI framework, the access to the models going through private cloud compute. They've already created the the conditions for that to happen.

1:34:01 Yeah. And I think I think Demis at DeepMind was saying that like there won't be ads in Gemini or like the core Gemini models but he's out and over time you could imagine that the ad gets baked into the actual Gemini response and then that is monetized on the Google slash DeepMind side which then justifies the pool of capital that gets traded to Apple in exchange for that entry point. Makes a lot of sense. What do you think OpenAI needs to do to get from 1,000,000,000 to 10,000,000,000 on the ads product? They have billion users.

1:34:36 The technology does not seem that complicated to me. I know it's deeply complicated but it feels like when you have the machine that can, you know, solve math and write endless code like just writing a matching algorithm, serving up the UI, like that feels tractable. But is it a supply side thing? They need to get more small businesses like what Facebook did where Ridge Wallet and every small company and medium sized company is on there? Actually, do they need to go after e commerce gaming?

1:35:05 Like, where do you see this going?

1:35:07 Well, it's first of all, famous last words that it doesn't seem that complicated.

1:35:10 Do we get to build? No. No. You're right because because you and Ben Thompson were talking about like, oh, they they they've said they're gonna do this and then it still took months to actually get out.

1:35:19 Well, yeah. But they're doing it. Right? I think they just need to keep doing what they're doing. I think the opportunity is vast and immense and I think they're executing at a at a blistering pace.

1:35:26 I I I So they just opened up to more countries. Now, there are more than 40 that they're available. And they just opened up to 31 more countries. This was like two weeks ago.

1:35:33 Okay. A

1:35:34 week ago. So this is like I mean, so they're they're expanding. I mean, it's just take it takes time. Look, I think what they will be I think you you start you start to see the growth inflect when they integrate through conversion optimization. That means you're bidding a specific amount for a specific outcome.

1:35:53 Like, you're still doing CPC optimized conversions. Once you are bidding against a specific outcome Mhmm. Then the growth inflects. And I think, you know, they'll be off to the races. That's the gap from one to 10. Yeah. And then 10 to a 100 is just continuing to onboard, you know, every SMB possible.

1:36:08 Yeah. Where do you sit on Instinct and the personalized

1:36:12 agents Breaking news. So Meta just released an Instinct clone Yeah. Oh, Muse. Oh. And it's a new standalone app.

1:36:19 It's not the Meta AI app. Okay. It's a new personal agent app that's in the app store. Muse from Meta, it says your personal agent that takes things off your plate, approve what gets sent or spent, track ticket prices and book reservations, connect all your apps, get ideas for what your agent can take on

1:36:38 Okay.

1:36:38 Which I think is is smart because a lot of people just don't really fully understand what agents can can do yet. But, yeah. Let let's talk about Meta's current AI strategy in a moment. They're they're, you know, have started a pricing war. It's unclear what how much the Meta AI app matters to the whole strategy.

1:37:02 It's they're doing Muse now, they're doing coding, they're throwing a lot at the wall. But what's your view? I

1:37:11 think I So I like to sort of just I think I just I'm just I'll hone on I'll hone in on a thing. They're they're yeah. They are throwing a ton of stuff at wall. I'll hone in on the things that I think are like where there's there's there's like a a true like thematic strategy. So a couple things.

1:37:26 Right? Muse agent, interesting. Open Claw kind of thing, like, we'll see instinct kind of thing. We'll see where that goes. I think there's gonna be a lot of those, and I think those get more and more domain specific over time. But, you know, we'll see that's interesting. It could get more consumer adoption on desktop. We'll So see. It's going be the standalone app. Was just reading Vaz's tweet about it before we hopped on.

1:37:44 Yeah. It's it's interesting to me that they own WhatsApp, which has billions of users, and they wouldn't just try to clone instinct in WhatsApp and just like try to get like actual crazy adoption there versus launching a standalone app.

1:37:58 Yeah. Like button inside of WhatsApp as opposed to a separate app. This is always

1:38:04 You can interact with it that way. You can interact with it from what's up.

1:38:07 Totally. Totally. But the the friction of getting people to download a new app instead

1:38:11 of It was always different when it was just like, if you open Instagram, you get stories now.

1:38:16 Right.

1:38:16 Right. Right. There's no separate, there's no step. It's coming to you whether you like it or not because we believe in stories and we are going to get that to a billion MAO very quickly and they did.

1:38:27 Yeah. I mean, you have to you have to keep in mind that there's still a lot of reluctance to this on the consumer side and there's still a lot of mistrust from Meta on the consumer side. So I think, know, if you just integrated this as a forced download on WhatsApp, you might get a lot of churn. Sure. You might get a lot of telegram adopters.

1:38:42 Okay. My sense though is like if you look at a couple different things, there's there's a there's a there's a thread that you can sort of parse. Like, one is Meta AI. Right? So, okay. Here's here's an interesting enterprise use case. Mhmm. What if you had the Go to AI for managing all ad campaigns within your company?

1:38:56 100%.

1:38:56 The to AI app for managing all campaigns within your company. Because I'll tell you how people do it now. They use Codec. Codecs not purpose built for that. They use Claude.

1:39:03 Claude's not purpose built for that.

1:39:04 Yep.

1:39:05 Meta AI now does that. They've added that as as a as a whole feature set. Yep. I think that gives you a strong indication of where they wanna go with this. Who knows ad campaign optimization better than Meta?

1:39:15 Why do they not have the right to win that? So even if you just said, this only applies, the enterprise use case here only applies to opt

1:39:23 to to

1:39:23 optimizing Meta ads, that's still a massive opportunity for revenue.

1:39:28 Totally.

1:39:28 Not and and that's just on a on a on a on a first order perspective, not even considering like, well, that actually might result in more ads.

1:39:34 Well, that's why I was I was somewhat excited about Manus because I was like, okay, they're buying this enterprise agent, they can point it they can point it at Meta Ads and if the agent is only good and it's actually not getting like, if Meta's saying, yes, we want you to spend we want you to optimize ads with this product. And we're fully endorsing this. We're supporting it. It's not computer use or any of these other things.

1:39:57 Yeah. But I mean, like I mean, yeah, that was exciting. You know, got rolled back. But but I mean, Meta AI now does it. They've introduced that functionality to Meta AI.

1:40:04 Now if you're an if you're a performance marketing team, you're saying, well, I need a tool to optimize my Meta campaigns. Mhmm. Which one am I gonna use? Am I gonna continue to use Codex? Again, it's not built for Yeah. Like, I don't think that they've devoted resources to making that a primary use case for it. And so Meta AI has done that. And if I even if it was just for Meta. Now now imagine that it's the go to for all ads optimization. Yeah.

1:40:24 Okay. Well, now it's an even bigger commercial opportunity for them, an even bigger enterprise opportunity. Right? And keep in mind, Meta launched the Robin Media Mix model as an open source framework. So it's basically like a measurement apparatus that helps you do, like, probabilistic measurement across all of your campaigns that you're running, like online, offline, whatever, out of home.

1:40:40 The reason they did that, think, was because they felt they were being under attributed. Right? Now, if you if you thought if you had this if you had this sort of same mindset with the AI enablement layer, the tools that people were using optimizations. Yeah. Then it this makes a ton of sense. This might end up with more revenue flowing to them because the the optimization ends up preferencing them. Not with the thumb on the scale, but just because it was actually under optimized before. So that might be a reason to do this. If you genuinely thought people were spending less than they should be to be optimizing their ad spend on your platform, then you would do this. And if that's the case, this could have this could have like dramatic second order effects to you.

1:41:14 So that's one thing. The other thing is business AI. And business AI, think, is like underappreciated. I've been following this for like quite a while. I had their VP of business on the podcast maybe a year ago, and we talked about this. But they launched a business AI, which basically was a chatbot that you integrate on your website. They said, okay, that's kind of boring. Who cares? But what people were doing is they were going they were clicking an ad, going to someone's website, and then they were they were using the chatbot to say, like, what's the what's the best selling product? And stuff like that.

1:41:39 That has a lot of opportunity there. Like, even if it's just surfacing basic data like that, there was no way for a lot of SMBs to do that. Yep. Now, what they've done is they've introduced like an AI enabled pixel, which can automatically, you know, sort of tune itself, requires no very little optimization from the the advertiser side. Imagine all the other stuff they could do if they have access to your landing page.

1:41:58 Imagine landing page optimization. Imagine personalization. It's all driven by Meta's own systems. And the advertiser could say, look, I'm an SMB. I don't have time to a b test or I don't have the resources to do, like, constant a b testing, constant experimentation on my landing page to to optimize conversion.

1:42:14 I know if my conversion was better, I'd be able to spend more money on Meta and get more sales. So if Meta can offer that to me, I'm okay to surrender that capability to them. Imagine they penetrate even deeper into the customer experience on the website. Certain brands will say, no way. But a lot of SMBs, which is that big bulk of their advertiser base, will say, yes, please.

1:42:32 Do anything you can do to optimize conversion for me because that's gonna result in more revenue for me, which I'm then gonna reinvest in more ad spend. So I think, like, Business AI is really really valuable to them, and they've integrated that deeply into WhatsApp. So that is going into the WhatsApp experience where people are able to communicate with businesses directly, get a chatbot, you know, understand these things about the the the catalog, understand these things about the business, and the the the core product offering. I think that stuff has a ton of potential, and I think that's probably undervalued if you look at if you look at Meta's AI initiatives. I think people focus too much probably on like the output Mhmm.

1:43:04 Of, you know, TBD. Yeah. But but like I think and that's important too. I mean, look, the the the model that they just released is competitive. Like, no one people left them for dead six months ago, but now they're actually producing competitive models. I think that's important too. But if you think about the integrations in the surface area, they could just apply even commodity AI to, it's there's a lot of value left to gain.

1:43:24 Yeah. We never book in

1:43:26 our No. No. No. Yeah. And and what you're saying to me it's like it would be I think it would be thrilling for shareholders if they saw MSL actually focused on applying AI Mhmm.

1:43:40 In the business and not like Yeah. Oh, coding's a hot category. We should we should introduce a coding tool or or oh, let's let's try to ramp Yeah. Muse API revenue to to a $100,000,000,000. Right? Or or whatever these things are that they're trying to do. And so, yeah, it's like it's products. It's like net new products that are consumer products like Muse that I think are interesting and it makes sense that they'll take a crack at this category because it will probably be, you know, multi trillion dollar category. It's aligned to their existing business. And then and then some of these other things of like actually applying

1:44:13 AI Yeah. Instead of personal super intelligence, it's small business super intelligence. That would be a huge opportunity.

1:44:19 And small business owners like small business owners, I feel like are people that generally they may have a little bit of a love hate relationship with with Meta. Mhmm. And that they're they're like, I I'm dependent on Meta. If I turn off Meta ads, my business, you know, revenue drops 50%. But at least they're like, I need Meta.

1:44:36 Yeah. Like, they're dependent on it. Whereas average consumers, you look at the comments and people are like, okay, Meta's talking about privacy Right. With their new agent. Right. It's like, no one nobody's like buying the the Meta and suddenly is privacy focused narrative. Right?

1:44:49 Yeah. Mean, I look, lot of the SMBs, they understand that they wouldn't exist before meta. I mean, there's not there's not a lot of animosity towards the company that provides your right to exist. It's like the oxygen that they rely on. I don't think there's a whole lot of hostility to that.

1:45:02 I mean, it's it's what always bothers me about, you know, these discussions is like, just not recognize that d two c would not exist absent meta. If if if meta's ads platform hadn't like evolved in the way it did, there would be no d two c category. Right? And so it's like, it's it's ridiculous to say first of all, I think one one I mean, let me know if you're up we're up on time here, but like I know where that like, a lot of people look at the index. Like, so if you look at, like, the the advertising a share of GDP, it's basically it it sits within like a narrow band, like, over time, historically, going back really, like, hundreds of years.

1:45:31 And people say, look, that proves that advertising is not driving the economy. It's almost like a drag, you could say, because it's just this cost base. But like, the thing is when when an advertising format when advertising enables new business, the fact that I'm still maintaining a share of my revenue as as my ad spend doesn't mean that that ad spend would exist or that my revenue would exist without the ad spend. Yeah. Right?

1:45:55 And so if you create these new opportunities for commerce, like, doesn't matter that the share that you reinvest back in advertising stays the same. That's not an indication of the value of the economy. It's creating that chunk of the economy. And so I think that you you get that gets missed in these discussions. Like, if you it it actually impacts economic growth.

1:46:12 It's it's it's endogenous there. And so it's an input to the economic growth. And so you can't say that the fact that remains within this narrow band as a percentage has is some way to evaluate its importance to the economy. And when you are creating new opportunities to transact that are entirely derived from advertising, then you are creating that section of that segment of the economy.

1:46:31 Totally. Totally.

1:46:32 Do you think that what is the what is the state of affiliate marketing? Because when I think about a product like Instinct or this new meta product, an agent that will go and buy things, it feels like that would be easier to bootstrap in theory than an advertising platform that you need you need scale and a and a demand side or supply side for. And so you could in theory when even on day one of instinct I asked to go order a pair of shoes and it uses an Amazon affiliate link and they're making money. Is is that going to persist or is that like a dying a dying monetization path? Or is it just small?

1:47:10 No. I mean, I think there's some upper limit now, but I think it's something that'll persist

1:47:15 Mhmm.

1:47:15 You know, indefinitely. I mean, you look at there's big companies that do essentially affiliate. Rakuten is one of them.

1:47:20 It's a

1:47:20 successful company. I I think the thing though is like, is that subscale? Yeah. Affiliate makes sense. But this is the argument I made when when ChatGPT introduced instant checkout.

1:47:28 Like, that's not gonna that's not the optimal way to monetize that attention. The optimal way is to do conversion based conversion optimized advertising because that introduces the auction mechanic. You actually deliver the value that a person's bidding. And then so the more value you give to people, the more they bid and the more revenue you make. Right?

1:47:44 You don't get that with affiliate. Affiliate tends to do the opposite. It tends to preference the lowest cost but highest converting goods. And so it's kind of you're promoting like the worst stuff, the cheapest stuff. Right? The Shane stuff, the tamu stuff. And the thing is when you unlock sort of like the the latent value with the auction mechanism and bidding, especially second price bidding, then you get you get growth with with performance. Yeah. And so that's how you scale the platform. You onboard long tail SMB advertisers.

1:48:12 Mhmm.

1:48:12 Well, thank you so much, Jordan. Do have anything else?

1:48:16 No. Always a pleasure.

1:48:17 A good Always always a great time hanging out. Thank you so much for coming on the show.

1:48:21 Great to see you.

1:48:21 Have a great week.

1:48:22 The third technology brother.

1:48:23 And we'll see you soon. Let me tell you about public.com. Investing for those that take it seriously. We got stocks, options, bonds, crypto, treasuries and more with great customer service. We ran long there so we're shifting Harry Mellsop from Antioch to the end of the show.

1:48:40 We're gonna be joined by Scott Wu from Cognition in just a minute or two. Huge While we yeah.

1:48:46 So While we were live with Eric Mhmm. Meta came out with Muse Agent. So Tyler, download it. Got it. You got it.

1:48:55 Already got it. Okay. Look at that. Look at that. Excited to see what you think.

1:49:02 Built me an equestrian simulator.

1:49:05 I don't know. Don't think it's focused on vibe coding. It's like it's the whole book me reservation, book me a flight. Okay. I tried to you imagine

1:49:15 Table in the Beverly Hills Hotel, please.

1:49:16 Can you imagine being like resi or OpenTable and and having you know, trillions of dollars of watching trillions of dollars of CapEx and people are like I've been one day, you're gonna be able to just tell your agent, you know, book me book me this reservation. And they're sitting there being like, it's two clicks, sir. I guess, sir, it's two clicks. Cause you go into resi, I mean, it's it's it really is it really is quite quick.

1:49:44 Apparently, have to ask Scott about the number 46. Do you get this reference? I don't know. Okay. We're gonna ask him because he's here in the waiting room.

1:49:53 And let's bring in Scott Wu, founder, CEO, Cognition. Welcome to TBPN, Scott. How are you doing? Great to see Tell us about

1:50:02 the numbers. News day, I was going to say. Huge news day.

1:50:06 You guys to talk about.

1:50:07 Congratulations. Yeah. Crazy crazy news day. We're to talk about math. We're going talk about AI.

1:50:11 We're going talk about fundraising But first, the number 46. I was I was told to ask you about this. Why?

1:50:17 Oh, that's funny. Is That's funny. Well, saw the my Twitter handle is Scott with forty six.

1:50:21 Okay.

1:50:21 The the reason my my handle my my I'm I'm like Scott with forty six on basically all these platforms is because when I was in, like, elementary school, I the the biggest thing I knew of was this middle school competition, math counts, which is now made a bit more famous because people have seen these videos of math counts and stuff. And in the math counts, there's, a written proportion, and then when you do well enough on that, you go into the actual, like, head to head. A But perfect score in the written portion is 46.

1:50:46 46.

1:50:47 And so then what ended up happening is with this particular round, actually ended up pricing it at 46 pretty partially for the meme, partially because it, you know, ended up being the right

1:50:55 kind of

1:50:55 thing You did the meme. And that, but yeah.

1:50:58 Yeah. You you have to imagine that certain VCs are like, okay. We've agreed to this number. But are there any insider meme references that would save us even 3% here? Let's I know.

1:51:09 Do a deep dive to see if we can get the founder to give us a little advantage.

1:51:12 Yeah.

1:51:13 Yeah. Know. But congratulations. Tell us about the fundraising round. You got every investor in the world on the cap table at this point.

1:51:20 Is that right?

1:51:21 Yeah. Look, you know, it's it's an exciting time for Cognition, obviously. And I mean, I think that the the biggest thing I would just call out is like agents are just getting really, really good. You know, and I think two years ago by the way, we're not that old, like two and half years ago. So two years ago, you know, was when we kind of did the Devon launch, and back then, as you remember, was like, it went pretty viral and so on, but but it was really just like a prototype I

1:51:44 would not say. So much skepticism. People were like

1:51:46 There was a lot

1:51:47 of They are coining a buzzword that will never exist. Agents, they're just hyping this up, and now it's like everything is agents.

1:51:54 Yeah. I remember thinking about this because I was

1:51:56 just like, man, it's great that Devon actually works because if it didn't,

1:51:59 we would just look so dumb right now for me. Yeah. Oh, yeah. Totally.

1:52:02 But so so that was two years ago. Yeah. And like, didn't have customers, you know, it's very much just like a prototype and and like, here's what we're building towards and so on. Right? One year ago, I would say it was very you know, the the the product existed.

1:52:15 It worked. It was it was for very specific use cases, you know, that that you would kind of have it work and have it do it end to end. But I think most people in the world hadn't really gotten across to the idea of using a background agent. Now I would say it's it's I mean, especially in our circles, it's like, you know, it's becoming a more and more commonplace thing, and a lot of it is just that the capabilities are just so good that, obviously, you should go and delegate to something that just does the task entirely, right, and be able to manage these and so on. So that's been the biggest thing.

1:52:43 We Yeah. I mean, the agentic era arrived. The importance of harnesses like Devon arrived. How do you think about orchestration? It feels like we saw a glimpse of that with Gastown that went viral now.

1:52:58 Whenever you fire off a prompt in any modern AI system, you can see the harness and the lead agent sort of talking to sub agents and delegating things. Is there is orchestration a meaningful leg up? Is it another exponential in the sense that we went from LLMs to reasoning? That was an order of magnitude. Then we went to the agentic era.

1:53:26 Have we always been in the orchestration era, or is this a new era? Is this a new meaningful change? Are there new disciplines that need to be explored in the world of orchestration?

1:53:35 Yeah. No. So so it's a it's a really important question. I think there's there's definitely a lot of, like, what's the word? There there's there's definitely a lot of kind of, like, you know, misconceptions, I think, out there about it.

1:53:47 I will say that there's a sense that people get of, like, oh, like, if you just say the magic words to the model, it will suddenly become 15% smarter or something like that. That's much less of a thing, especially now because the models are all RL's on very specific capabilities and these kinds of tasks and so on. I think what is much more the case is if you number one, if if you go and bring in all of the context and the information and the systems that you need, for example, you know, encoding, a very simple example is like a model and an agent that can go and test its own code and click through its website by itself and then go and look and say, okay. That was right. That part was wrong.

1:54:20 Let me go fix that. It's obviously gonna be way more capable than something can one shot it on its own. Right? Mhmm. A model that can go, you know, an agent that can go look up at in in the Datadog what went wrong in the logs is like much more powerful than than something that can't. So so that's one is just like being able to bring in all the tooling, the context, and so on. You all this kind of, like, messy real world stuff. How do you navigate a big code base? How do you get through, you know, all all the the secure systems that are you know, that a company will have on its software? And then number two, I would say, is combining the strengths of the different models and the things that they're good at.

1:54:50 Right? And so, you know, I I mean, everybody's talking about price performance of model. Everybody's, you know, showing the charts of the Pareto curve and so on. And this model is, really cheap, but is good enough for this percent of tasks, and this model is, like, the way expensive one and, you know, this much percent better and whatever. Right?

1:55:05 But, obviously, what that means is that by combining the the different models, as long as you know what use cases to route to each model, then then you can do a lot better than than any one alone.

1:55:14 How permanent is that task, that job of picking the right model at the right price point? Because we see there's a lot of attention from the from the big labs and it's very clear that they're that every big lab is going after building beautiful slides, front end, back end, cybersecurity, bio, like math. They're all working on these and there's this trade off between a certain model might be really good at something else and then the new big model comes out and it's better than everything at everything at every price point. But how Yeah. And I think the risk is that you can you can wind up in a situation where you're like, oh, well if I just wait two weeks I'll just be able to use the major hammer because everything will look like nail.

1:55:58 But it feels like there's also always going to be this cost optimization, speed optimization. So that optimization problem, is that sticking around forever? Is that what you want to build the core of cognition around, nailing that?

1:56:11 Think I it's sticking around. Yeah. I mean, I think it's if anything, it's going be a bigger thing as time goes on for a lot of these use cases, the intelligence is is is really not the bottleneck anymore. Right? And so so obviously, it's been been been been another hot day for the last few months, but but I would I would say it's like, your ChatGPT, for example, new, bigger, smarter comes out, you know, smarter model comes out.

1:56:30 It doesn't necessarily change their retention metrics all of a sudden because most of

1:56:34 the things people ask ChatGPT, you know,

1:56:35 it turns out the model is good. Yeah. They they already go and get them right. Right? And like now what you care about is is is not just pure IQ for for any task, you know, whether it's coding or or legal or or customer service or whatever.

1:56:46 You know, it's it's it's it's not just, okay, what is the raw logical intelligence of the model that I'm working with? It's like, okay. Well, does it does it know all of the details of what I'm dealing with? Does it do things in the style of how I want it to do? Is it fast? Is it cheap? Is it effective? Is it is it trained specifically on my subset of use cases or, you know, all all all of these things like that? And basically, what that means is I think that there should be this frontier where people start to much more aggressively use all these different models rather than, you know, relying on the single biggest one, right? Like, think a year ago, that that was more of a phenomenon of a of a year, year and a half ago because there were a lot of use cases that were just on the cusp of possible, you know.

1:57:25 And when that was the case, of course, you want to go use, you know, the very smartest model that you have up there, Sonnet 3.7 or whatever it was, you know, a year But and a half now, because all the models are really good, you know, what that means is is you can be a lot more judicious.

1:57:40 Kind of a question going around, I think, on different people's minds. Coding agents are now incredibly capable, but is the quality software around the world actually increasing? And so I wanted to get your view on, you know, you guys are working with a lot of the biggest companies in the world. Do you feel like their the quality of their software is increasing or are they just doing more or are they doing back end migrations that consumers don't even experience? Mhmm.

1:58:08 Because you've had this I I I guess question for a while which is like, okay, when I when I download an app for like at a big airline Yeah.

1:58:16 Was chirping at ruin about this. So was like, if the AI models are so good, why is it annoying to use the United Airlines app? He was like, have you tried to use it recently? It's actually pretty good. And then I did try it and I was like, yeah, actually maybe it is better.

1:58:28 But I don't know. Where do you sit on Yeah.

1:58:30 I think short answer is definitely yes. It is better. Mhmm. I think to the extent that that there's more to do, a lot of that is is much more like it's much more a function of these practical problems like going out and getting distribution, going and doing that.

1:58:43 And Yeah.

1:58:43 I mean, I think I think people, you know, people in our ecosystem understand this intuitively, but it's worth kind of calling out out loud that, you know, a 50 thirst fifty fifty thousand person software org is not going to figure out how to use coding agents overnight the same way that, like, a three person YC company is going to. Right? And so so there are steps in the process that need to go happen. There's like a lot of kind of like, you know, onboarding and education that has to be done. There's figuring out the right systems.

1:59:08 There's, of course, getting through people folks like, you know, the the the security guardrails and the and the reviews and the walls that people have to make sure all of that is tight and and something that they wanna have operating in their ecosystem. But but, know, once that's there, we we very much see that that's that's clearly the case across all these industries. And I mean, I think, know, it's like Yeah. So between like Yeah.

1:59:29 Go ahead. So so so I can imagine if I'm if I'm selling cognition, there's a few different pitches I could give. One is I go to a company and say, look, you don't have an ERP or you don't have an e commerce system. We're going to come in and build that and it's going to be done by the time we're complete you're and going to have this. It's going to live forever but it's a new capability.

1:59:50 Others, we're going to give every person in your organization a copilot we will help And transform the then third might just be like, look, we're just giving you access to dev and it does good stuff. Like, you know, go and you couldn't get it at this scale, but now you can because we're working together. What's resonating the most these days?

2:00:13 Yeah. No. I mean, I think folks are rightly, by the way, are are very focused on at at this point, on on just, like, what are the actual use cases that it that it's gonna drive and what is it gonna mean? You know? Sure.

2:00:23 Think there was a there was a period where it's very much, alright. We're in the token maxing world. Yeah. How many tokens are my engineers using? How that was you know, that lasted for all of, four months or so. You have great times. But now,

2:00:36 you know, I I think there's much more kind

2:00:38 of like there's much more clarity of thought, would say, from from folks in the industry about, okay, well, AI is great. Let's talk about what use cases that we actually care about. Like, what what are the top three, four priorities we care about as a company? Like you said, maybe it's this this killer new feature that we wanna put out. It's this app that we wanna make way better or it's whatever.

2:00:54 And let's talk about how we actually go and measure that and improve on that. You know? And I think the the biggest thing, if anything, I think is is is kind of like, yeah, like, how how can you show it in the concrete results? Right? Like, the the value should be there, obviously, because AI is so good, you know, and it's so smart.

2:01:08 But but, like, unless you're tracking that, you're making sure you're using it for the things that are effective versus aren't effective, you know, you're you're looking at the the productivity on a case by case basis, obviously, it's like you you'll never actually know which things you're doing are working versus not. Right? And I think that's been a big big theme for folks.

2:01:24 How do you predict that the router market is gonna evolve? We've seen a lot of moves recently. Ramp has a router. Stripe, you know, bought OpenRouter. There's a bunch of other players.

2:01:34 Everyone wants to be in that token stream. But how do you see this sort of like category evolving?

2:01:41 Yeah. No. Look, I mean, I I think it kind of makes sense. I mean, it's it's you know, a lot of these companies that you're for example, are they think of themselves as like the center for finance on the Internet. Right?

2:01:50 And, what you are people going to be spending money on in the Internet in five or ten years? I think a lot of it is going to be tokens and models and agents and whatever you call it. So I think from that perspective, I think it's super reasonable. And I think the routing part is a big piece, to your point, I think the entire kind of like infrastructure around how you do payments, around how you do spend management, all these things, I think, are still going to exist and basically need to be redone in the world of agents. And so I I I frankly think that there are a lot of products to go and build in that space.

2:02:20 What has your reaction been to the recent progress in math? When you Yeah. I think the very first time you came on the show you predicted correctly that the IMO gold medal would fall. Google OpenAI both scored just barely gold, not 46. I think it was 38 or something.

2:02:41 It was like they missed the sixth question but they did achieve gold.

2:02:46 42 by the way. 42 is a gold in in the the math Olympiad which is slightly

2:02:51 different. Slightly different. Yes.

2:02:52 Have to

2:02:53 remember it's it's different numbers for every competition. Yes. Yes. So so I'll tell you

2:02:56 what my my honest thoughts on it Please.

2:02:58 Which are, look, first of I think it's it's it's actually insane.

2:03:02 I I I think there's there's a lot

2:03:04 of controversy and there's Yeah. Discussions, arguments, whatever, but I feel like the most important thing to call is that guys, we just solved Navier Stokes with AI. Like, is Insane. It's absurd. And and and I don't think it's gonna stop, obviously.

2:03:16 I mean, I think the like you know what's funny? I actually have money. This is my this would be my only AI bear view that I ever expressed to you guys. Okay. I actually had money on a bet that the rebound hypothesis will not be solved by the end of twenty twenty six. Okay. We'll see as, know, three and a half more months through. But I'm pretty sure if not '26, it will be solved in '27. I think it might get done in '26. Okay.

2:03:37 But but

2:03:37 I think it's less than 50%.

2:03:39 We'll see. But but no, I mean, I I think all of this stuff is just gonna get done. And I I mean, I think it's like it's it's kind of insane. It's it's hard to put it to words. I think if you're if you're not familiar with the the the kind of scale of this, but, you know, Navier Stokes, for example, is, like, very fundamental problem about, like, fluid dynamics and such and it's been around forever basically and lots of people have sucked lots and lots of time into it.

2:04:03 You know, the fact that this can even just be done in like a week.

2:04:06 Yeah. Eighty eight hours.

2:04:08 Just going and working around in the system and orchestrating. It's just, it's insane. Yeah. I think on the point of the controversy, I mean, it's kind of funny, but similarly, I think a lot of people who aren't as familiar with the math academia world might not know that. You know, this is actually just always what happens in math academia.

2:04:25 Like, literally, Newton versus Leibniz on the founding of Callaghanos Oh, was the is still, by the way, is one of the biggest arguments that people have, you know, this is back in like the sixteen hundreds or so. And so, in practice, think like, you know, I'm I'm sure look, I I believe that both sides meant well. I think the accomplishment itself is going to be the biggest thing that we all remember from this. Mhmm. We could debate, you know, what what exactly shapes up and and what what it means for folks.

2:04:53 But but but but it's it's, you know, the the rest, I I would say

2:04:56 it's kind of the the AI the AI accomplishment itself is

2:04:59 the biggest thing by far. The rest kind of feels par for the course, to be a 100% honest. Yeah.

2:05:03 So it it seems exciting for the the world of math. It's certainly exciting as sort of like a researcher recruitment to this is where progress It's is also sort of useful benchmark. Like I could imagine in the future like two weeks in the future but just a few weeks in the future if you're training you know Muse Spark 1.5 you want to just throw Navier Stokes at it and say don't search the internet, don't look at the result, can you solve it because that's a good benchmark, right? But I want to know about your philosophy because you went viral with the Devon launch. Like what matters to your customers, your recruitment because there's a world where you're like, oh, I want to put my throw my hat in the ring and duke it out for different math challenges.

2:05:51 But that doesn't seem critical path to the growth of your business, certainly hasn't been because you're growing very quickly. But how do you think about the value of when you're talking to your actual customers, what is important to get across?

2:06:04 Yeah. For sure. And it's kind of by the way, it's it's kind of a hilarious thought also that, you know, we're just gonna be

2:06:09 like, I I actually still think that one of

2:06:10 the things that's that that I've never gotten over is, like, one of the token benchmarks that people do when they're going and training models and doing runs and stuff is AIME. Yeah. AIME. And I think people don't necessarily know what it is, but basically, the AIME was it's a high school math competition. It's one of the the hardest math competitions and it basically selects like the top few 100 kids in The US Yeah.

2:06:31 To go qualify for the next level to actually compete for the US team and everything. Like, I took all this, you know, every year as a kid. And it's kind of funny because it's like, yeah, dude, your bottle can't even get a 15 on the Amy. It's it's out of 15. You know? It's like, what are you even doing?

2:06:44 But like, no human can do that,

2:06:46 you know? And we're already someday, it'll just be like, wow. You can't you can't even solve the Riemann hypothesis. Like Yeah.

2:06:51 What does your model even do? You must totally

2:06:53 messed up your retraining run

2:06:54 or something.

2:06:55 I get it. You're trying to save money, so you're using a model to cancel Riemann. Like, look, not everyone has token budgets right now, so you got to pinch pennies. It's cool. It's probably good in Photoshop.

2:07:08 But yeah. No. So for for us, what I would say is, obviously, like, a lot of what matters is show show you the the the real bench for don't get me wrong. Be sick if Devon went and solved the hypothesis. I don't currently expect that that's what's gonna happen because it's much less, you know, this this kind of, like, fundamental, you know, basic science research is much lesser focused as

2:07:26 opposed to to kind of going and doing real

2:07:28 world use cases. And so we see you know, it's like when we show benchmarks, it's like, okay. Here's a benchmark on how it does at finding security vulnerabilities in real world code bases. Here's a benchmark on how it does yeah. Right.

2:07:37 And and a lot of it is basically just like making sure we're speaking to the thing that Sure. That that folks care about and folks need.

2:07:44 So, yeah. I mean, related to that last question, how big is the cybersecurity side of the business? How much demand is that driving? Obviously, that's been a huge story this year.

2:07:54 Yeah. Yeah. No. I mean, it's been it's been a massive thing for us. Obviously, everyone that I think is is is really thinking about this and thinking about I I mean, everyone's kind of freaking out, guess, is is is, like, the honest way to put it, which is probably correct.

2:08:05 I I think cybersecurity is going to, you know, it's I I I think there's going to be real threats that happen. I mean I mean, again, you know, when we say it's like these big orgs take time to adapt and to to use new technologies and so on. Obviously, these hackers out there, like small teams that are going and using the best of the models or doing their own, who knows what they're doing, like, they're not waiting, you know, and and a lot of these capabilities keep getting way better. And so, no, it's a small but meaningful part of of our business. It's probably in the neighborhood of, like, you know, around 10% of of the dev and sessions and or the dev and ACUs that get spent today are on security, but but we see it growing pretty quickly.

2:08:42 I mean, our security product's only like two months old. So

2:08:45 I love it. Well, congratulations on the fundraiser. I gotta ring the Gong.

2:08:48 Amazing update. There we go. Great to see you, Scott. Please go. Please go take out some open problems just for fun.

2:08:59 Don't don't distract the team. Like, don't don't rope them into it, but just

2:09:02 Back spin up to the next prime.

2:09:04 Yeah. Yeah. Exactly. Spin up spin up a swarm. Just you and the swarm.

2:09:09 Great to

2:09:09 see you.

2:09:09 Great to see Cool.

2:09:10 Cool. Thanks for having me.

2:09:11 Alright. Cheers.

2:09:12 Well, we were talking about security. No better time to tell you about CrowdStrike. Your business is AI, their business is securing it. CrowdStrike secures AI and stops breaches. We will be joined by Greg Brockman, the co founder and president of OpenAI in just a minute.

2:09:30 In the meantime, the Descartes the Descartes acquisition, there were talks that Anthropic was buying Descartes and it seems like they have walked away. This is an exclusive in Bloomberg. But

2:09:43 Say it.

2:09:44 They put Descartes before the horse. That's the funniest quote tweet by Shashant Rahman here on the timeline.

2:09:53 We love Yeah. Stories come out that that maybe one of the stick is totally, you know, I I don't I don't know how true this is but it sounds like one of the sticking points was maybe Relocation. Descartes happily building in Tel Aviv not wanting to move over to The US. Who knows? Dean There's

2:10:11 lot of

2:10:12 Jacart is incredibly talented.

2:10:14 It was a very fun demo. He came on the show and used his AI image model, his video model.

2:10:20 Yeah. His willingness to do just a live demo of technology that seemingly was better than anything else Yeah. That we have seen.

2:10:28 I mean it was it was it was low res but it it showed you a glimpse into the future. He would, you know, be prompting it while he was on the call with us talking about what's behind him. Now I'm in a wizard castle. Now I'm in a sci fi, you know, cyberpunk city. And all of that was very, fun to see.

2:10:48 So I'm sure that they will continue cooking, and we will have to check-in with them soon. But we have Greg Brockman, the co founder and president of OpenAI with us. Welcome to the show, Greg. How are you doing?

2:11:00 Doing great. Thank you for having me.

2:11:02 Thanks for hopping on. Huge day. Can we start with the with the math advances? What is what what is happening? Why is this important?

2:11:10 There's a lot of back and forth in the timeline, but I'd love for you to just set the table for us on what actually happened with Navier Stokes today.

2:11:19 Well, it's always a huge day in AI and modern scientific progress, I would say. Yeah. Today, we announced that our model had solved the Navier Stokes problem, that we found a counter example or a sort of proof that you can actually

2:11:38 Mhmm.

2:11:38 That these theoretical equations do have a singularity or kind of breakdown under certain circumstances. Interesting. And this is a problem that has been open for a very long time. It's one of the seven millennium problems, which are kind of some of the deepest, most important problems in mathematics. And, I think that the problem itself is important, right?

2:11:57 This is new knowledge for humanity, that the proof itself is actually very elegant and beautiful, and I think that there's a lot to learn from it.

2:12:03 Yeah.

2:12:03 The equations have lots of application in fluid dynamics in other areas. But to me, what's even more important is about what this represents about where we are in terms of model capabilities. Yeah. And the fact that we can actually generate new knowledge, that we can learn from these models to have them help us solve problems that are otherwise outside of reach or would take us very long to solve.

2:12:24 Yeah. I mean, where should I actually go with this? Is this going to help me book a flight? Is this going to help me cure cancer? Is this just going to help you recruit researchers who are fascinated by this stuff?

2:12:35 Because I think that there's this has, you know, taken over the technology world. But I imagine that this will not be something that gets talked about at backyard barbecues with friends and family that are three clicks removed

2:12:48 from Outside of SF. Yeah. Yeah.

2:12:50 Well, look, I think that there's, first of all, the applications of this specific result or the equations themselves, right? Which are things that let us better understand phenomena from ocean currents to airflow around aircraft.

2:13:01 Sure.

2:13:02 Around turbulence and things like that. But it's really about the broader insights and methods that can help scientists and mathematicians further accelerate their research. And I think that, again, representative of if we have models that can help solve this kind of problem, then what happens from here? Like, what other problems that are immediately applicable? And I think that talking about curing diseases and new medicines we're going to be able to develop, all of that starts to become much more real when you have models that are at this level of assistance and capability.

2:13:34 And I do think that there are going to be real changes to think about in terms of we can have so much more ambition with the kinds of challenges that we can hope to tackle now.

2:13:44 Yeah. I I mean, I think even even if this doesn't break through to the broader the broader world, the weekend definitely will because it seemed like everyone was talking about Blender, talking about Astra, building stuff. Take us through the launch of Astra, what the feedback has been, what you've learned. It seemed like there were a couple resets, the model scaled very well. How was this launch different than previous launches?

2:14:14 Well, first of all, I've just been blown away by the community reaction to Astra. It's been really amazing and very humbling, honestly, to see all the creativity and the different ways that people have been applying the model. Mhmm. And I think that it's very clear that we've reached a new threshold of computer use, so this model is able to really work with different kinds of applications in a way that was not previously possible. And people are taking full advantage of that fact and really thinking about how to create.

2:14:42 Lots of people showing off three d creations and mapping out physical locations and turning them into into these three d models and thinking about can you use this for design of physical parts. Someone talked about how they were designing some mechanic for catching hair in a shower drain and that they were able to it was super cool. Right? That that they're now able to actually manufacture that.

2:15:04 Shower hair

2:15:05 super intelligence. Yeah. No. That stuff's so mundane, but it's so important. I I feel like a lot of this stuff gets lost.

2:15:12 Right?

2:15:12 Exactly. And I think there's a core there that's very important, is that we are talking about these grand challenges sometimes or very esoteric applications. But really, the everyday, the number of problems that you have in your life

2:15:24 Yeah.

2:15:24 That you would love to solve, it's now possible, right? Yeah. That we're really trying to empower the individual to make it so that you can be you can have superpowers, that you can accomplish more. Yeah. And I think that that, you know, really trying to benefit people, empower people, build tools that can really help you and help you in your your daily life.

2:15:42 Yeah. That's all part of what we're we're we're working on.

2:15:45 So, yeah. I mean, it seems like a huge number of people in in tech effectively rebuilt their entire house in blender and plan to remodel this weekend. Thanks to Astra. And but I am wondering about the merge and how you bring together codex, Chatuchi PT work, Chatuchi PT on desktop. I have a gaming PC now with a NVIDIA card in it.

2:16:09 I have a Mac Mini. Like I have all these things and I can imagine that I'm just I'm doing that unnecessary or like it's fun for me but that early adopter work of going and unhobbling it a little bit here and there. But in the future this will all just be tucked in one, you know, prompt box and it might build a three d blender model for to answer my question of should I remodel my house or not. But how do you see the capabilities that we saw on display from sort of light power users over the weekend actually making their way into consumers who might not even know what Blender is?

2:16:44 Well, I think you're exactly right that we really want to shift these tools from requiring kind of a low level

2:16:52 sort

2:16:53 of Yeah. Access or guidance to really having the human be able to fly, right, to really be empowered, for you to be able to set the goals and the objectives and that you still should feel like you can get into those details and you can understand them. You can provide that oversight because, ultimately, you should feel accountable for the outcomes. You have this absolute amplifier, right, like a trampoline or like a rocket ship for the mind, like however you want to analogize it. And I think what that means at a practical level So, of all, this year we've been really seeing this shift from just pure chat use cases

2:17:27 Mhmm.

2:17:27 To Jentic use cases. But I would also keep in mind that chat is alive and well. Mean, we're we're now well over a billion users every week that you can see that the market share of ChatGPT is starting Yeah. To climb once Because we've been investing so hard in so many use cases that are important for people in education Yeah. And health, and a variety of other areas.

2:17:46 And then at the same time, these productivity deep knowledge work use cases, those are really taking off. We've had this like almost vertical wall of agentic adoption since we launched ChatGPT work. Yeah. And I think that the fact that these are two distinct modes, that that is actually a point in time. That is something that we're continuing to unify and merge, and that we're starting to see that there's a new emerging form factor for how people want to consume AI.

2:18:08 And I think that it's almost like that the promise of AI has always been that you have something that you can talk to and really delegate work to that's proactive and persistent. And that I what we were promised, if you were to rewind five, ten years ago, is never a low level language model that you have to think about context windows, and you have to select thinking strength and you have to select different models. Like, none of that none of that is is the future. So, I think that we're moving towards real amplification, real giving you time back, real having computers that are able to operate according to your goals, to your desires. And I think that that is the core of it.

2:18:46 We're developing this safely. That's one of the core commitments that we make and how we think about this. We really see the power of these tools starting to really start to increase in terms of what people are capable of and that under the hood, utilizing tools like Blender so that that is almost a detail that fades into the background is absolutely the direction of travel.

2:19:08 It feels like in AI particularly, there's been a almost like a first mover disadvantage in that billions of people have tried ChatGPT and and some percentage of them tried it for the first time and have a certain impression of the product and what it can do. And then, you know, even in the last few weeks, there's been new agents and products that come online and people try it and their mind is just completely blown. And I think it's funny because I'm like, well, as somebody who's like, you know, trying to get the absolute max out of Chad GPT, I'm like, well, I've been running like, you know, I've had like an agent running that, for example, will tell me every time a SpaceX launch is gonna happen and if it gets delayed. Right? These sort of like persistent agents that are running the background.

2:19:50 And but but strategically, I feel like it's a it's a new kind of challenge because you have this as capability has been been scaling, first movers need to be almost like constantly reminding the the market of all these just like new ways overhang.

2:20:08 Yeah. We we think about this a lot. And I think that there's this discovery problem that we as a field need to really encounter in a first class way. And we haven't done it fully yet, but I think we have a real shot at solving it better than any product before. Mhmm.

2:20:24 Because the the thing that that right now we kind of rely on, you think about chat, gbt, chat, gbt work, these are both text boxes.

2:20:32 Mhmm.

2:20:32 Right? And it's like, well, this new text box is way more powerful than the old text box. But, oh, there are some reasons that you still want use the old text box. It's not You know, it's like far too confusing. Yeah.

2:20:41 People just want something that can help them solve their problem. The whole point is to get your time back, not for you to have to go and become an expert in all these these internal details. But at the same time, we also have a model that understands what you're trying to accomplish, right? That you're explaining to it, Here's what I want. It has a lot of context on you, and so it should also be able to proactively say to you, Hey, actually, if you ask me this other way or if you added this connector or if you authorize me to do this or if you hook up your credentials in this way, I can go and do this other thing for you.

2:21:12 And so, we're thinking a lot about that self knowledge, that onboarding process. And I think that is a huge, huge opportunity. And I think that there is both the disadvantage that you cite of people tried it, they form an impression, and it's changed that it's something new. But there's also an advantage. I mean, ChatGPT, like over a billion users every week, like that is unique.

2:21:31 One has that kind of use on these models. I think that the number of people who have tried ChatGPT before, I think it's another billion, billion and a half, something like that. That's a huge opportunity as well for us to go back to those users and say, Hey, can now solve the problem for you. We can now help you in ways that you didn't see before. I think that it's true.

2:21:53 It's real. If you look at how many people use chat for health, 300,000,000 people every single week with health queries, right? And that that's really making a difference in people's lives and that of their loved ones. So we have such

2:22:02 opportunity Yeah. One you one stay there with health and explain where One these

2:22:07 note before that. Something that I think is really interesting and I think something that OpenAI can can do a lot better is, like, when people talk about like, everyone in AI wants to be like the apple of AI from a marketing standpoint. And when they when you think, like, Apple marketing, you're thinking, like, Mac versus PC or you're thinking 1984, these big branding campaigns. But the actual thing that Apple does really, really well with marketing is they just hammer really, really specific details about their products. Right?

2:22:34 They're like, they're advertising the new camera. They're advertising Memoji. Right? They're advertising like certain features in Safari. Right? And so it's like with AI, the surface area of like things that you need to communicate is actually like an order of magnitude greater because it could do so many different things. So I think that there's such an opportunity for the company to focus advertising. Yeah. The brand campaigns are awesome and like the the launch video for Astro was amazing. But it's like, there should be billboards running of like very specific things that Chatch BT can do to give you back your time.

2:23:06 Yeah.

2:23:07 Yes. This this has actually been a real sort of realization or just like something that I have really come to over the course of of of this year. And if you look at even for example the, you know, we just announced ChatGPT images 2.5. Mhmm.

2:23:20 And if

2:23:20 you look at the launch video there, the thing that I love about it is it shows here's someone creating an image, and here's like a bunch of different variations of it. And then here's them taking their favorite one and having it in the world. Like someone said, here's a cool like old sketch of a candle holder. You see an awesome visualization of it. And then you see the physical candle holder and you're just like, that's what you want.