The next next thing in AI and AR

27:29Outside of, to your point, Heath, the Apple TV Plus app on like Roku's, right? Which is where Apple ships an app. And this is the split is kind of the same. It's like, okay, we're going to give you an asset package of Apple things to run locally on the car for things that we cannot risk if there's a disconnect. The speedometer, the turn signals, da-da-da-da-da, right? and there's some differentiation in how that stuff works. Like the truly critical stuff cannot be touched. They call that the overlay layer. That's your turn signals and hazard indicators. The speedometer and stuff like the needles, those assets can get refreshed over time.

28:04So those run in a different layer called local UI. And then there's everything that runs on the phone. And then in the middle, there's like, I don't know, you want to like mess with your sports settings in your fast car? Like that'll just be the weird automaker UI shining through CarPlay. And it's like, oh, this kind of feels like a mess, right? It's unclear who will take this and who will use all of CarPlay, like redesign the entire car to be CarPlay, to be Apple, and who will say, okay, we took next generation CarPlay. We fund cool new speedometer. Every other part of the car controls will be the existing UI because we can punch through CarPlay.

28:43And honestly, based on the feedback we get on the show, it's like some of our audience will pick the car makers that use all car play. Like they will make an affirmative decision to be like, you know what? That car maker, Toyota, uses all of car play. We're going with them over Honda, which is like, here's your weird heated seats. Well, the third path is Rivian or Tesla, right? Where it's totally controlled by the automaker. And like Rivian is a great example of, I think, you and I both test-driven Rivians, and Eli, like, I think their interface is great. It's very snappy. Obviously, I missed some of the iPhone connection, but I could also see more automakers going that direction.

29:22Yeah. But they have to make good user interfaces. Rivian's is good. That's a big difference between... Yeah, no. Like Rivian and Tesla made good user interfaces for the most part. Tesla... Everybody else... Tesla's getting weirder and weirder, I would say. For the most part. Tesla's like you turn the car on and it's like, do you want to play a game on Netflix? You know, it's... that's a lovely experience so Rivian just upgraded the generation 2 R1 trucks just came out they've changed the computing architecture of those is different but then they also updated the interface to be more cel-shaded I mean they have an Unreal Engine and they like using it that said I looked at the parking lot at Apple Park when we were at WWC a lot of those people have Rivian's and there's all these rumors about Apple and Rivian doing stuff together.

30:12Apple Music just came to the Rivian interface. If Tesla's never going to break, I feel confident about that one. Just a prediction I'll make based on nothing. The Tesla UI is going to be X.com in a few years. That's the direction Tesla's coming. It'll be Brock fully, just telling dirty jokes when they try to drive a car. Rivian is still in the place where they might try some other stuff to claw back some market share, to win some new buyers. and there are an awful lot of Rivians. It's California. People like trucks there, but it's something I noticed. All right, moving on. See, it's like a fast round, not quite lightning.

30:53We're just moving. Kranz, tell me about this remote because you're very excited about a home remote. You know I love a home remote. I'm still sad about Logitech. I still have the fanciest of the Logitech remotes and it had soft touch plastic, which means it's starting to rot. I have a TiVo remote in the corner of this house. It's not a good situation. It's so bad. The rot on that soft touch is so gross. But there's a Kickstarter happening right now for the Haptic RS90 and the RS90X. And they promise all the stuff all the big universal remotes promise, which is like you can control stuff via Bluetooth.

31:29You can control it via IR. You can control it via Wi-Fi. Usually that requires some sort of hub that you have to plug in somewhere in your house. this appears to require no hub and also it's just got a big screen on it so it's instantly the coolest remote you can have in your house uh and they can it's supposed to be able to control all their smart homes and stuff and like i said logitech did this i've got a control that does this doesn't have the cool screen and it required a hub so this is instantly cooler and this also is going to come soon rather than discontinued it's a kickstarter it's yeah kickstarter there's The Kickstarter part of this is the really big heavy caveat because there's a whole bunch of integrations that they're promising.

32:13No. Not there yet. There's no Matter support. There's no support for some of the other smaller radios and stuff like that. So this is like cool. Neat. We're like two minutes away from Samsung being like, Bixby AI and your TV will not do everything for you. I don't want it. We are, though. We are. It's coming. Yeah, like AI is probably, you know, the reason Logitech got out of this business was because they said nobody wants these. And me and four other people were like, oh. I want my phone to be my universal home remote. That's it. I don't want another remote. I want fewer remotes. Can we make that happen?

32:52Yeah, I just, that's how I use my phone. But like, I'm just getting, like, who is this for, I guess, is the question for me. It's me. What show are you on? Alex. It's me and four other people. It's the four people who are still sad that our Logitech remotes have killed. Whenever I see a screen like this, I just think about how slow it will be. I think about touching a button and nothing happening, right? I love it. I'm like, yeah, I want a screen like this. But you know that thing you do with a touchscreen where you touch it and the button looks like something happened, but nothing happened? That's this product.

33:25Almost like a guarantee that that is this product. It is, I will say, I love that it's called the Haptic RS90. Great name. Great name for a universal remote. Fully good name. I just worry that actually people just use the apps on their smart TV. And what this thing is trying to solve is not. I mean, that's why Logitech got out of business. We can't control the apps on your TV. We can't see your TV. Yep. Yeah. So we'll see how this does. It looks like it's going to be funded. It looks like enough people have invested in it. It looks like it's going to be funded. This is like hot on the heels of that company Brilliant going out of business.

34:01Yeah. Which Jen covered. and they were like, we're going out of business, but the thing we put in your wall will still work. So Brilliant put a touchscreen in your wall that can control everything, which I was super hype on because the idea of having an in-wall Sonos controller screen seemed sick. You walk in a room, you're like, I'm going to pick a song. Boop, I'm not using my phone, the whole thing. And then they just went out of business. And it's like, that is always the worry I have with all of these things. Yeah. This is going to be like almost$400 probably when it comes out. That's too much.

34:34All right. Speaking, let's do one more gadget, and then I want to talk about what's going on in ARNBR with Heath. We mentioned RISC-V at the top, and we mentioned companies that might go out of business. Framework is shipping the Laptop 13, which is fascinating because they're going to have a RISC-V chip in it. This is the competitor arm, right? This is the open source competitor arm that has the performance per watt characteristics. I'm kind of dying to try it out. I also have no idea what software I could run. Also, I think one guy pointed out that it runs, it'll run about as fast as a Raspberry Pi 4.

35:09Yes. You got to start somewhere. Baby steps. Baby steps. Baby steps. Yeah, I have no idea what can run this, but I want to, like, this just is cool. This is the kind of thing I just want to have one. I like the actual line from the company in the PR is, the main board is extremely compelling, but we should be clear that in this generation, it is focused primarily on enabling developers, tinkerers, and hobbyists to start testing and creating on risky. Like, don't use this. The peripheral set and performance aren't yet competitive with our Intel and AMD powered framework main boards. It's like, yeah, this is just to play with.

35:49I will say the nerds in our audience have been very excited at risky for a very long time because arm is a like dominant quasi monopolist now like the arm architecture license is the thing and there hasn't been anything to compete with it which is why everybody is moving to it including microsoft because if you want performance for what you end up on this architecture which means companies like qualcomm are able to dominate the market right next to apple and you need an architecture license from arm to expand beyond it and arm isn't giving this up anymore, which is part of the reason they're suing Qualcomm, because Qualcomm is saying the Nuvia chips fall under its architecture license, and ARM says no.

36:34So there's all this stuff that's wrapped up when you have one dominant architecture that is owned by a company, and RISC-V is the thing, the open source. Was it NVIDIA? Not quite as fast as a Raspberry Pi 4, but it's the thing that might break it. Was it NVIDIA that tried to buy ARM a couple years ago? Yeah. Yeah. It was, wasn't it? Arm was, I mean, the finances of Arm are hilarious, right? Like it's a weird English company that started licensing these designs. It got, SoftBank bought it, right? It was part of the Vision Fund. In the heady days of the SoftBank Vision Fund, where Masa-san was like, everything I touched will be a monopoly.

37:15It didn't work out. You might recall WeWork blew that entire business up. They started divesting of assets. They were going to sell Arm. NVIDIA wanted to buy it. basically everyone yelled. Like, everyone yelled. Apple yelled. Like, all the architecture licenses were like, you cannot let our competitor buy this company. They said no. Arm is now public instead, which is a good outcome. Largers are bad. Fun fact, Arm going public made up for all the losses the Vision Fund has had for the last 10 years. That's crazy.

37:48Somewhere that we were, people are like, God damn. They're just launching a new product. Don't worry about it. I could have changed the world. It's a lifestyle. I watched one episode of We Crash. I was like, I can't live this again. I cannot be a part of this again. All right, let's wrap up by talking about AR and VR. Heath, you've got a scoop. Meta just reorganized Reality Labs. We're talking about that. The Vision Pro team is trying to make a cheaper headset. There's a lot of action here. What's going on? Yeah, maybe we'll start with Vision Pro. The information had a good story out, confirming and putting more details on what Gurman and others have reported, that Apple is indeed working on a cheaper Vision Pro.

38:28They hope to release it by the end of next year. And they have shelved a future iteration of V2 of the Vision Pro as it is today. So another high end, very expensive one. They're trying to get this cheaper one more into the ballpark of a you know, premium top of the line iPhone price. So we're talking 12 to$1 ,500. I had heard even before the vision pro came out that Apple was already working on this cheaper version. Um, apparently when Mark Zuckerberg was talking to me and others and basically saying, I'm going to scorch the earth with cheap headsets forever and not really care about making a business, which we will get to in the next story.

39:08Um, Apple was like, Oh, okay. We need to like really work on a cheap one and get it out. So they've been working on it for already a couple of years. I do expect we'll see it next year, whether it ships next year or TBD. And I think, you know, I've only used the Vision Pro once in that initial demo that we all had at WWDC last year, but I would think that, you know, if it was in the ballpark of$1 ,500, I would maybe pull the trigger even if the os and the software hadn't meaningfully um gotten better just because of you know the immersiveness of it i kind of it's just at 3 000 it's just like it's it's out of touch for most people and it's still out of touch at 1500 but it's much more of a almost impulse buy for some people at 1500 um so yeah and it will get hopefully more developers interested um but yeah apple's plugging away on and i think what we're what the the other part of this the fact that they are canceling the more expensive one or shelving it just goes to show that they packed so much tech into this thing that it may not actually be a good product.

40:18And they may kind of like I found that out already. Neelai, you talk about how you don't use the Vision Pro anymore. It's a great demo. Honestly, it was probably top three demos I've ever done of technology. And it's something that I consistently hear people do not use after the first couple weeks. Would you say this was a Homer car? No. I mean, it's not like that ridiculous, right? Like it had some focus. They wanted it to be something. I think about the fact that it was a great demo. So much of technology right now is like great demos to fall apart as a product, right? And that was the Vision Pro to a T.

40:53And every AI demo so far has been that from, unless you have some very narrow use case where AI can solve a problem consistently for you. Every other demo is like, what? I don't know if I would do that. The Vision Pro, they have to make a big decision. And here is the thing that I noticed at WWC that I've been thinking about. In that opening keynote video when they were jumping off the plane. Yeah. And they were doing the introduction. And everyone was wearing their flight suits. Do you remember this? Vividly. Ridiculous, bombastic opening. They had Mike Rockwell on the plane. He's in charge of the Vision Pro.

41:29and he turns to the camera with the eyes on and they played it for laughs. The eyes were a joke. Those eyes aren't shipping anymore. I can tell you that. This is what I'm saying. If you remember the initial Vision Pro stuff, it was like the eyes will place you in the environment. They were so serious about the eyes. So serious. Like dead serious about the eyes. I asked in one of our briefings, because the only question I asked anybody during our Vision Pro briefings were have two of you ever looked at each other with the eyes? Like have two people ever existed in the same room wearing a Vision Pro and when you looked at each other with the eyes did you laugh?

42:11And they were like that's not funny. Because they were dead serious that this thing would bring you into the environment. They meant it. It was sincere in that Apple New Age sincerity way. Like when you go to Apple Park and you walk towards the theater and they play New Age music at you and they're dead sincere about it. And I'm like, this makes me feel like I'm in a cult. There's a gap there between how sincere Apple can be and then the world. And they were super sincere that these eyes would be the thing that enabled you to wear a headset all the time. No, let's be real. They did the eyes because they wanted to be able to not call it a VR headset.

42:46Well, so this is what I'm getting at. This next version, this cheaper version, they have to decide if this is a VR headset or an AR headset. And playing the eyes for laughs was the thing that signaled to me, okay, this is going to be a big deal. I will bet strongly that there will not be the eyes in any future Vision Pro product. And I'll tell you why. It's just a bunch of costs for a reason, right? No one's taking you on that bet. We like our money. I'll take the bet. The eyes stay. It was a Johnny Ive thing. I know this from talking to people who worked on the Vision Pro. It was Johnny Ive who insisted that the eyes ship, I don't know the exact dollar amount, but it added an absurd amount of cost and manufacturing complexity to the Vision Pro.

43:29Because they had to do that curved glass display. Well, the curved glass was... Or that curved glass in the display on the inside, right? It was having the front-facing OLED. I think it's an OLED, right, Neal? With the eyes. Yeah, it's a very low-rise. It added an absurd amount of cost. No one wanted to ship it except Johnny and his people. And they've all left. And they're making a joke about it in keynote videos. So, yeah, I don't think the eyes are long for this world. But so if you get rid of that and you're like, you're not, now you cannot be perceived by other people in the world. The thing is a VR headset.

44:01Yep. Right? And that to me is concession in one way, right? Like they talk about this thing as AR, AR, AR, AR. But if you can't wear it with other people, like it's a VR headset. And then everything they've announced since is immersive, right? New immersive videos, new experiences, new immersive games. And it's like, oh, this is becoming a VR headset. And a$1 ,500 VR headset right now, the state of the art there is a pretty good product, right? Like you can make something great for$1 ,500 VR. Yeah. And I guess that maybe brings us to Meta, right? Should we go to that? Yeah. So Meta obviously was first to this, bought Oculus a long time ago.

44:41They restructured Reality Labs, the division that does all of their hardware and Horizon, that cursed platform. this week I had the scoop and command line. They are forming two new groups. So Reality Labs is now Metaverse and Wearables. So you will notice there is no VR in those two phrases, right? The Rift, I'm sorry, God, look how the Rift, wow. The Quest is now under Metaverse reporting to Vishal Shah, who used to run ads at Instagram and is a software guy. So you basically have the thing that created this whole org that the company really bet its future on is a subdivision of a software org.

45:23Meanwhile, wearables, I'm told, is leaning heavily into the Ray-Bans. I published this internal memo from Boz, the CTO, basically saying that the Ray-Bans have been a much bigger hit than any of us anticipated. Zuckerberg and others have been alluding to this throughout the year. I've heard that it's somewhere in the ballpark of a million-ish units that they've sold, which is 3 to 4X, the first version of the Ray-Bans, and they're seeing really good retention on the product. So people are actually using it consistently. I have them. I have the AI turned on in them. I wrote in command line a few weeks ago.

46:00I'm not trying to pat myself on the back, but I will just saying that Meta is quietly winning the AI wearable race here because the Ray-Bans are something that Meta has not done in a while, you could argue, maybe ever, but shipped a really good consumer product. Yeah, you're 100 % right. It's something that normal people like, something that tech people like because it's got the AI and yeah, the AI still hallucinates, but it's very fast and it's right like maybe half the time, right? So like I pointed at a tree in my backyard and I say, what is this tree and how do I water it? It got it right.

46:36So like, that's not, it's still a demo, right? It's not like I'm, it's not gonna like change the world, but in a couple of years, you can see where that's going. And it's like, oh, if I have reliable visual AI in Ray-Bans, I may use my phone less, which is the reason Meta is doing all this, by the way. So they are putting way more resources behind the Ray-Ban line. And they also have these full-fledged AR glasses that they've been working on since 2018, codenamed Orion, which are insanely expensive. They cost thousands of dollars to make. I have heard from people inside Meta that it's a, again, incredible demo.

47:13Um, they really feel like they've gotten there on the, on the hardware. They're not sure what the software use cases will be yet. Um, they're doing a lot of reviews of it right now. I expect we'll get a tease of that later this year at connect their conference. Um, so they have these two lines that they're doing with glasses. They've got the AR glasses, which I leaked the whole roadmap for Meta's, uh, hardware, uh, a year or so ago. So we know that the Ray-Bans are going to be updated, uh, next year with a little heads up display. So think more Google Glass, less full AR Magic Leap. And they're going to come with a wristband that uses neural interface technology from the startup metabot called Control Labs to control the display and to control some of the other inputs.

47:59Because you can also play podcasts through the speakers and the frames, right? That's shipping next year. That's going to be interesting. and then the AR glasses, the full-fledged ones that are going to be more expensive and also have the band will be 2027. So this is the path Meta is on. They're going full on glasses and really deprioritizing. They hate when I say this, but it feels like that, the Quest, because the Quest is still really struggling. They have some early product market fit with gaming, but the retention is still not very good, I've heard. and the Quest 3, they got a lot of really good things about the mixed reality, right?

48:39But it's just not, I don't know why I want to put it on. Whereas the Ray-Bans, it's like, I get it. I have audio on all the time. I can take calls. I can take photos and videos. And I have AI that is starting to work more and more. And by the way, they look like normal glasses, right? So it's just obvious that this is the direction they should go. They seem to now be realizing that and they're putting all their chips there and laying off some people as a result. So that was the news out of meta this week. So what's really interesting about Quest 3 is the rate of innovation on the Quest 3 or improvement, I guess, in the Quest 3 skyrocketed after the Vision Pro came out.

49:15They're like, oh, we should ship a bunch of these features. Like, I guess there's some competition. Like, here's travel mode. We took it away, it's back. They increased the fidelity of the pass-through, right? Like, just stuff that they should have done. But they're still not, it's not winning. and even Apple, if they come out with a$1 ,500 version of the Vision Pro, they still have to make the case for VR. Right, which, like, Meta's been trying for a long time. And now if Meta's saying, all right, it's these glasses, it's the thing that everyone always knew was the thing, right? It's augmented reality glasses where actual light passes through the display and you're looking at actual things instead of screens and then we can put some information near or around or over those real things.

49:56turns out that's the thing everybody has wanted the whole time. Why would Apple continue to boondoggle its way towards the Vision Pro even at$1 ,500? Well, because they're on this path, right? Apple was working on AR glasses like Orion at Meta, and they basically shelved it semi-recently in the last year or two for the headset route. They think that if they can keep iterating on the headsets, they'll figure out glasses eventually, whereas Meta is like, no, we are going to have glasses out in a few years. And Apple's fine. They're always late to this. Like we all know that. So they'll come in late like they usually do.

50:32I don't know if it'll be as late as they were with the Vision Pro, but they're going to be late and that's fine. But yeah, I think they're betting on productivity and being at home and coding and all these things with the Vision Pro. And Meta's like, no, we want you to loot. Meta's goal is to use your phone less because they hate being under Apple and Google's thumb, whereas Apple is like let's just extend what we do and I think there's the big strategic difference I will say that I have a friend from the Brooklyn years who in the pandemic became one of those people who doesn't live anywhere she just lives on beaches I know one of those folks, yeah she's great does she work in Nvidia or what?

51:12she does not work in Nvidia and she was the first normal person adopter of the meta glasses, the Ray-Ban like all of her instagram pictures are a beach taken right every bands now then you can you can just see like oh this thing worked like the thing worked it looks like she's wearing wayfarers and all the pictures but there it's the meta glasses and like it worked it's like so much there's no chance that any of these people are wearing visual pros as they like nomadically travel the world or the quest three but there's a hundred percent chance that hey your camera is now in your sunglasses and you're looking at cool stuff you want to share it Instagram is a loop that Matt can close.

51:51I was at a very fancy tech conference where I'm not even allowed to say the name earlier this week, but it was a lot of tech CEOs. And there were several of them wearing the Ray-Bans, like walking around CEOs of companies that it would be very funny for me to say who they were wearing the Ray-Bans. And you're not supposed to record anything here, but like they're, they're wearing them. And it's like, Oh, like, yeah, this is like catching on. It's catching on with tech people, powerful people. And it's catching on with normal people, which is like that is a circle that meta has not been able to draw in its entirety of existence as a hardware player.

52:25How good did Jensen look in the glasses? I can't comment. I'm telling you, I think VR is interesting. There's stuff in VR, like Supernatural, they keep promoting as a hit. And Supernatural just commissioned some study where they claim that waving bats in VR is as Srenyos is running. It feels like it. Sure. Sure. Whatever, man. To me, yes. Agreed. If you can claim that fancy Peloton Beat Saber is the same as a swift run, more power to you. But that's the only thing they have that's sticky, right? And Apple doesn't have anything that's sticky yet except for the monitors. We'll see. I just think Apple moving away, the compromise of the Vision Pro was they built a VR headset with all this pass-through in the eyes and stuff to make it closer to AR.

53:15And if what they're going to do is drop the eyes and go head more towards VR, that feels like a turn that is notable because the real platform, you want to talk about the real platform shift, has always been VR. Everybody knows this. Tim Cook has talked about that as being the big platform shift. And to go to, we just made the best VR headset, would be very different. All right, we've been going, we're way too long. This is out of control. Does this show ever go long? I didn't think that ever happened. I don't have tears here to hold me in. That's what's going on here. We got to take a break. We'll be right back.

53:45More on the Roach S after this. Support for this show comes from LinkedIn. When you're a small business owner, your business is on your mind 24-7. So when you're hiring, you need a partner that works just as hard as you do. That hiring partner is LinkedIn Jobs. When you clock out, LinkedIn clocks in. LinkedIn makes it easy to post your job for free, share it with your network, and get qualified candidates that you can manage all in one place. LinkedIn's new features can help you write job descriptions and then quickly get your job in front of the right people with deep candidate insights. Either post your job for free or pay to promote.

54:24Promoted jobs get three times more qualified applicants. At the end of the day, the most important thing to your small business is the quality of candidates. And with LinkedIn, you can feel confident that you're getting the best. Based on LinkedIn data, 72 % of SMBs using LinkedIn say that it helps them find high-quality candidates. Find out why more than two and a half million small businesses use LinkedIn for hiring today. Find your next great hire on LinkedIn. Post your job for free at linkedin.com slash track. That's linkedin.com slash track to post your job for free. Terms and conditions apply.

55:00Oh, hey, welcome to gift wrapping. Whoa, Zoe Saldana. Hey, can you wrap these, please? Wow, iPhone 17s. You splurged. At T-Mobile, you can get four iPhone 17s on them. It's the perfect gift for everyone. I'm the worst. I only got my mom a robe. Well, it's better than socks. So I have to trade in my old phone, right? No, at T-Mobile, there's no trade-ins needed when you switch. Keep your old phone or give it as a gift. Incredible. In fact, wrap up my old phone, too, for my Aunt Rosa. Forget that. Aunt Liz will be jealous. Sounds like my family drama. Oh, I got it. I'll give it to my abuela. I'll take reindeer paper with, hey, where are you going?

55:35To T-Mobile. Thanks, Zoe. The holidays are better at T-Mobile. Get four iPhone 17s on us. No trade in needed when you switch. Plus four lines for just$25 a line. With 24 monthly bill credits and four eligible portings on essentials. For well-qualified customers, auto pay plus taxes, fees, and$35 device connection chart. Credits end-in balance due if you pay off earlier. Cancel. Contact us. Finance agreement. 256 gigabytes. $830 required. Visit T-Mobile.com. Support for this show comes from Zapier. When people talk about AI, they often talk about the future. How will AI affect the economy? What will AI's potential be in the future?

56:09will AI be able to make popular art? But the more important question is, what can AI actually do for your business today? If you're looking for concrete implementation of AI for your business, you can try Zapier. Zapier says they can help you and your team actually reap the benefits of AI, not just talk about it. With Zapier's AI orchestration platform, you can bring the power of AI to your work. Connect top AI models like ChatGPT and Cloud to the apps your team already uses. so you can add AI exactly where you need it. Whether that's automatically enriching leads, coaching your sales rep, resolving IT tickets, or more, you can do it all with Zapier.

56:48Zapier is for everyone, tech expert or not. No IT bottlenecks, no complexity, no AI hype, just results. Teams have already automated over 300 million AI tasks using Zapier. Join the 3.4 million companies already automating with Zapier and transform how you work with Zapier and AI. Get started for free by visiting zapier.com slash verge. That's Z-A-P-I-E-R dot com slash verge.

57:25all right we're back this time we got to talk about AI this will just take two minutes there's nothing going on. Nothing. Actually, quite a lot going on AI. And a lot of it is just, as ever, confounding decisions by AI companies. Let's start with a confounding branding decision. There's a new company. It's called Safe Super Intelligence, SSI. Good name. This is Ilya Suhzever, who, if you recall, was the CTO of OpenAI who participated in the coup against sam altman and then changed his mind there's some safe super intelligence right there whether or not you have belief in that decision uh then they couldn't figure out how to bring him back to the company and now he's left to start doing alex what is going on here yeah i mean ilia not just the cto the co-founder and really the one who is driving the research agenda at open ai it was his decision to lean into the transformer the t and chat tpt so um And I think I've heard one of the largest individual shareholders, like incredibly influential.

58:27He's someone that AI researchers look to as like a demigod, right? And he has been totally quiet. There's been this meme of like, where's Ilya for the last, you know, six to eight months, really, since the November coup attempt. And yeah, he's back with a company that they will not describe anything about except that they're going to invent super safe AGI, whatever the hell that means. His co-founder is Daniel Gross, who used to run AI at Apple, which is interesting. And is now I would consider one of, if not the most plugged in venture capitalists in AI who's been raising a ton of money and deploying it into startups.

59:04And they are putting the call out for people to join. And I'm sure we'll be rating Ilya's former org at OpenAI in very due time. And the key thing here is Ilya saying we are going to be free from, I forget the exact wording, but we're going to be free from the product and management day to day and only focus on long term aligned AGI research. And we will not release anything until we hit AGI. So this is a pure research lab, which is what Ilya has been known for his entire career. I'm going to read this paragraph to you. Yeah, read it. Because anybody who has ever worked anywhere will, I think, have the same reaction to this paragraph.

59:47Our singular focus means no distraction by management, overhead, or product cycles. And our business model means safety, security, and progress are all insulated from short-term commercial pressures. Sorry? What? Well, the business model is we just raised an ungodly amount of money and we're going to keep raising ungodly amounts of money. But that means you don't have a business model. Right. Yeah. Like that's how you get to do these things is like you don't have the pressure of a business. Yes. Isn't that just all a really pointed jab at Sam Holtman? But yes. Right? It obviously is. It was more that than actually, like in this case, I was assuming that he was going less like I'm here to make money because he's made it very clear he doesn't care about money.

1:00:38And more like I'm here to like say I don't like what Sam Altman did. And so I want to make a company where we're going to do cool stuff. All I'm saying is I watched Max and two of her six-year-old friends from kindergarten try to play lemonade stand. And they are distracted by management overhead and product cycles. And their business model means they often compromise on safety, security, and progress. It's like there's just like no way you can put any group of people together and not have management overhead. It is not possible, especially if you're trying to hire the most highly paid, sought-after talent that exists in technology today.

1:01:16And that's what it's like. I read this paragraph. I'm like this doesn't – you can't do this. The VC math here is simple. We will give you whatever you want. And when you invent AGI, it will change everything. It will be the new fire and we'll figure it out. That's I guarantee you that's the extent of the diligence. I was going to ask, like, where is the money coming from to recruit all of these very highly paid? Well, they won't say where they've raised money from. So if you know, let me know. But if you're listening to this, I want to find out. Daniel is they're all super wealthy and they will have no trouble raising whatever they want, because when you co-founded OpenAI, you can raise whatever you want.

1:01:56And then in particular, participated in the notable management overhead of doing a huge coup against the CEO. Yes. A failed coup. Yeah. It's like your track record of not having management overhead is very bad. Also, like a question I have is what is safe AGI? I keep hearing this. No, it's super intelligence. It's not AGI. Or super intelligence, which is actually now the new word for what comes after AGI apparently is the thing I keep hearing. And so we keep inventing new terms because they realize like the terms they had, no one understands or they don't actually even understand what they are.

1:02:29What is safe AGI? What does this mean? Like Ilya's whole thing at OpenAI was alignment, right? He set up this alignment group that was going to get 20 % of OpenAI's compute budget. Apparently they never did. That led to a lot of infighting and I think, you know, helped him try to kick Sam out. but I always hear this and I'm like aligning for who and under what values are we aligning and they don't seem to have an answer and that's been like a consistent problem with AI right is is that they're always like yeah this is going to be for everyone but we only trained it with these certain data sets we only trained it using like the pre-established knowledge of the certain people creating it so it's actually fairly insular and doesn't work for everyone and there are enormous edge cases every time we deploy it.

1:03:16And, you know, as Neela and I uncovered when we were reporting on the coup attempt, like this gets kind of portrayed in the media as this, you know, acceleration versus deceleration, almost like religious faction war that's happening. And I do think there's part of that. I do think Ilya is almost like religious and is like fervor here. But a lot of this is just like boring company stuff, business stuff. It's like You wanted budget. You didn't get it. You got layered. Therefore, you don't want to work there anymore. OpenAI just hired a CPO who used to run product at Instagram and a CFO who was the CEO of Nextdoor, a public company.

1:03:52It's very clear where OpenAI is going. They're going to IPO. They want to build a massive business. They're going to get rid of being a nonprofit. It's very clear. And so this week made some noises about not being a nonprofit. Right. And so if that's the direction they're going, you're Ilya. It's like you just want to do research. Right. And that's where a lot of these researchers still are is they just want to do cutting edge frontier model development. And this is him saying, come here, we'll do it. and uh we should also note this is now the third offshoot of open ai that was has been created to create safer ai uh from open ai whose mission is also to create safe ai it's not a great look yeah so you have anthropic i would i would actually put xai in this camp elon was the co-founder of open ai uh now you have ssi which is also i feel like it's a medical term or something I got this once.

1:04:46It was, you take a pill for it. It's fine. It's a Neo-Zempic SSI. Yeah. Look, I just, it's great that everybody thinks they can accomplish these things. I actually think it's unclear at all whether Transformers and LLMs can ever do the things that people want them to do. And in fact, over time, it has been more and more clear that perhaps they cannot. lot yeah and so what i'm dying to know is if ssi or any of the other companies that think they can build agi if they have another idea yeah because if the only idea is we're going to make a slightly better llm i actually don't i there's i'm not you're not going to make you're not worried about whether it'll be safe because it's not going to be super intelligent i'm just why are people trying to build an agi like these people are nerds they they they watch that's why you just answered your on question.

1:05:36No, but they watch the movies, they play the games. This man is setting out to be the audio clips you hear when you're walking through the ruins a thousand years later in a video game of the guy being like, I don't know what's happening. Why is my AI going and killing the world? That's what he's setting himself up for if he was actually successful in his grand ambitions. And I'm just like, why? Because the industry thinks that they have invented a new form of intelligence. And that hasn't happened before. And so it's, I think the natural progression of this is that it's all smarter than us and replaces us.

1:06:11And it seems inevitable to the people building this stuff. What we see is like chat GPT can't get a math equation, right? Right. That's like what we see, what they're seeing. We got to keep in mind, like they gate this stuff, like GPT four was in existence for like over a year before open AI actually deployed it in the world. So they're seeing things we're not, I'm not going to like give them too much of the credit there, but I do think there's, uh, you know, Jan Lacun, who's the chief AI researcher at Meta has been saying what you just said, and he lied that LLMs are not the path to AGI. So his group is actually working on what comes after the transformer.

1:06:46What is the architecture we need? Um, because these are parrots, right? These LLMs are, are fancy prediction engines, right? And you, You don't really unlock intelligence that way. Yeah. No, I mean, yesterday I assigned our futures team to write languages, not intelligence. Like, I feel like we just need to say some very explicit declarative sentences that are true. Like, dogs are very intelligent and cannot generate a PowerPoint. Like, you know what I mean? Like, we're getting confused about what intelligence is at a very high level. I've been lately thinking a lot about the Steve Jobs slide of technology and liberal arts and it's like oh you guys forgot about this whole street, this whole street's missing from this conversation right now and there's just a piece of this puzzle where you can bet on this stuff to get better at an infinite rate but like most curves level off and it is unclear to me what actually makes things intelligent.

1:07:41That said good luck, good luck building your company with no management overhead die. Come on, Decoder. If you pull that off, boy, do I have a podcast for you. I spend an hour talking about management overhead every week with people, but it turns out three six-year-old girls have management overhead. I promise you. Maybe more than any company in America.

1:08:05Speaking of confounding decisions by AI companies, perplexity started out so hot, man. They were the ones. You can pick any model. You can ask these questions. It was better searching Google. And it turns out, boy, have they been awful shady along the way. What is going on here? Well, yeah. So perplexity, for those who don't know, is trying to unseat Google with an AI search engine. It's actually a good product most of the time. I would say that I like their AI answers more than Google's on average. And now we kind of know why. It's because they're stealing a bunch of content. And yeah, so Wired did some good reporting and found out that they were going around their crawler blocker.

1:08:50I've heard rumors that Perplexity has been doing this at an even grander scale with even much larger corpuses of data. And this is the open AI playbook on steroids a little bit, which is just run fast and beg for forgiveness. And that's what they're doing. They're basically going around people who have said, no, you are not allowed to summarize this and doing it anyway. And the way they're doing it is both shady and hilarious. So they publish some IP ranges of their crawlers and they publish the name of their crawler. So poor robots.txt, a text file that should not bear the weight of all intellectual property law on the internet but which currently does you can set it to disallow these crawlers you can set it to disallow gpt bot which is open ai there's google bot plus i think so there's google bot which is the regular search index and there's the other one which is google's ai crawler i feel like i need to disclose here we used to disclose gpt bot at vox media and then our company signed a deal and now we don't there you go there's your disclosure um and you could just you could disallow perplexity and on top of that they did this thing where like here are ip ranges like if you don't trust robots you can disallow these ips and then what wired found and then the folks at mac stories which is a great indie apple site um they found that their content was showing up anyway in perplexity and then the investigations basically revealed these ip ranges were fake and the crawler was fake.

1:10:20And they were setting up headless Windows 10 machines to browse the web. Yes. Like, you know those pictures of like the streaming fraud farms where hot dogs are just swiping Android phones? There's your VC money. I love it. Just swiping away on a Windows 10 browser. Just imagine if those Windows machines had recall. It's just like crazy. And the CEO of Perplexity has talked a lot of shit about Google. Like, why are these results so bad? And like Google, for better or worse, is trying to play by the rules. It's still pissing everybody off, but they're attempting to play by the rules. They have to.

1:11:00I mean, Google is under consent decrees. They're a massively regulated mega cap tech company. So they can't do this kind of stuff. Yeah. And here's a startup being like, what if we lied? We just didn't do any of that. And to be clear, I keep making this comparison. You called it the opening iPlaybook. This was the Google Playbook 20 years ago. Google just allowed all of Viacom's content to be on YouTube. Google indexed every book in the world without permission. Google did image search, which we got sued over. And it won all those lawsuits because those products were new. So they would show up in court and be like, judge YouTube.

1:11:35And the judge would be like, YouTube's sick. And Viacom would be like, our marketing staffers are using this anyway because it's sick. Google Image Search, so useful. Let's concoct in – nothing like this has ever existed before. Let's concoct some reasoning to allow Google to take all this stuff for free because the theory is that you'll go buy the image from Getty once you found it on Google. It's unclear what that theory really came out to in the end. I never did that. Yeah. Alex is not like, I opened a Getty library the next day, a Getty account the next day. But those are the theories back then.

1:12:08Google Books, that might have worked, right? You can search all the books. Maybe you'll actually buy the book in the end. Did it work out? Totally. I don't know. But those are the theories back then because the products are so new. The experiences are so new. Perplexity is just like, yeah, we stole a bunch of stuff. And what we've delivered is an incremental improvement on Google search. And I don't think that's going to win this time. maybe i am shocked at how little the people inside these ai companies and their investors again i was just at a very fancy tech conference with a lot of these people how little they actually care about this problem how it is really just like this is a this is a bunch of speed bumps that will go over uh to getting where we're gonna go um and maybe it's because of the the fact that google is google and they did all this 20 years ago and they're a trillion, multi-trillion dollar company.

1:12:56I think it's back to what... Yeah. I was going to say, I think it's back to what Neelai said about liberal arts education. They just don't have any historical context. So they're like, yeah, this is going to work. They didn't take any history classes. I don't need to know about Google. This is fine. No, these are systems people. Yeah. They don't think in terms of the values that we get mad about when it's like, hey, I made a thing. I charged people for it and you took it. That's not okay. with intellectual property, with writing and stuff like that, for some reason, it's just not a thing that a lot of tech people that I know, it doesn't register for them.

1:13:34I don't know why. Because they didn't take liberal arts. They're like, what is an English class? I want to spend more time. I don't have a view of this. I haven't locked this into my brain yet. So people have thoughts. Send them to me because I need the help. But I can feel this weird shift that I've been calling the copyright apocalypse. Because some people care about it a lot now, more than ever before. The this is mine and you took it between influencers on social media platforms rages out of control now. Right. I'm a Taylor Swift fan and those three notes of a song sound like three notes in Olivia Rodrigo fan.

1:14:14We have to do fandom war now until the music industry doles out writing credits is a real thing that happens. at the craziest level. This is a real thing that happens. And that is copyright maximalism by really young people. And then over here, you have nobody owns anything. It's just something that I should get it for free. And those things are just in tension in a way that it feels unresolved and potentially unresolvable. Like you can't, you can't have a bunch of YouTubers being like, I made this YouTube video first and anybody who even makes a video that looks like it is stealing from me right next to everything in the internet should be free.

1:14:52Maybe that's it. Maybe there's not a lot of love for media companies, as we all know. And the fact that it's scraping wired or the verge or whatever is like, whatever, they're fine. But when it's like your favorite influencer, your favorite creator, and they're on their feed going, hey, like this model, don't use it because they trained on my stuff. I do think you're right. I think people feel a lot more viscerally about that than they do, oh, like companies, right? I think it's when it becomes the individual. And again, what Google was up against, what Napster was up against was the big bad, right?

1:15:24Like the only face of the anti-Napster movement was Lars Ulrich from Metallica. And it did not work for him. But really the face was like Hollywood in the music industry. And no one ever feels any sympathy towards Hollywood in the music industry. Like Hollywood can be like, you wouldn't steal a cigarette, whatever they would say. You wouldn't, what was it? You wouldn't download a - You wouldn't download a car. Right. So Hollywood can put up the banners that say you wouldn't download a car and everyone's like, screw you. I'm super going to download a movie. If I could download a car, I would. That'd be sick.

1:15:55Yeah. That rules. I've had this 3D printer for years not making cars. But they were the big bats. Influencers are like people. Influencers are like people. You know what I mean. Influencers are actual people that you have parasocial relationships with. And like there's something very emotional there that is not resolved. and it's going to come to a head because, you know, the next story we have on our list here is a bunch of social networks that want to introduce AI in various ways. Alex, you played with one. And before we get to that, I just want to call out TikTok is now letting people make AI avatars of creators.

1:16:32Yeah. Right. And saying, this is your ad creative. And this is going to come for every social platform. Yeah, that's right. Like in a massive way. Like if you are a creator and you feel the burnout and you can make 50 more ads during your brand deal using it, you might do it. Yeah, no, this is happening. So yeah, I wrote about this new app, which was, you know, I get a lot of test flights to try things. And this is the first test flight I've gotten in a while that made me feel something very strongly where I was like, Oh, when I opened it, um, and it's, it is like Instagram, it meets character AI where basically it's a bunch of AIs in a feed commenting and posting together.

1:17:10And there's humans, but they don't need to be there. And it's actually, in some cases, not most, but in some cases, hard to tell who is human and who is not. And they just raised a bunch of money. It's an ex-snap people. And the reason I wrote about it is because Meta is about to do the exact same thing. Zuckerberg told me so last September. And I think later this year, we're going to see them start to roll this out, where basically anyone can create their own AIs, send them into Instagram, send them into Facebook. And same thing with businesses. And the idea is that if you're an influencer and you feel overwhelmed by your DMs, maybe an AI handles that for you, or maybe an AI actually posts on your behalf.

1:17:50Zuckerberg is super excited about this. I know because he's talked to me about it. It's very freaky to imagine what this actually does to our social networks. It's coming way faster than we think, though. Well, it's interesting because like in South Korea, they've been doing a little bit of something like this, right? With K-pop stars, where you can engage with K-pop stars on sites owned by the K-pop stars companies. And it's not always that sometimes it's them. It's not always that usually it's a bot, it's a person pretending to be them. And so like that parasocial, how do we automate that parasocial stuff has been happening in Korea for a long time.

1:18:25But I think the difference here is that was gated to, okay, you have to go download this app for this publisher and use it. Whereas this would be on meta Instagram. This would be like you open Instagram and it's like, oh, is that account in AI? They may have three arms. Okay. Maybe then. Yeah. Well, that's coming. So then there's the other part, which is you send agents out on your behalf as an influencer or someone where you're like, someone wants to pay money to subscribe. And one of the things I offer is like DMs with me. Yeah. And then my my agent will just like do that. Yeah. And I won't do it.

1:19:00And that's infinitely scalable, which if you are an individual creator and all these platforms are organized towards tearing down collectives and enhancing individual creators, if you're an individual creator, your ability to scale is limited by your emotional capacity to do this work and your minutes in the day. Yeah. And so, of course, you're going to let the robot talk to your fans. It is the biggest no-brainer in the history of the world. Of course you're going to say, okay, you brand X wants to do an integration with me, but they want 50 different versions as creative. I will make one and let the AI deepfake me into five more.

1:19:35And that is already happening. TikTok is enabling the feature. It's happening now. Yeah. Well, Alex, to your question about why. I was talking to Voo Tran, the CEO of Butterflies, this new app, about this, the why. because I showed this app to my wife and she was like, just like, it was an immediate, just like, I'm like, I think you're going to meet a lot of people like that who are like, why does this need to exist? Like, why do we need to like fill our time with just empty AI social media? And his point was like, yes, it feels kind of broken right now. And I wrote about this. It feels like when the hosts in Westworld break a little bit, it's like not there.

1:20:12But he was like, you know, I got into tech by like making friends on forums, talking to people that I never met. I only knew their username. He's like, we will get the models quickly to that quality. And as someone who also came up and got into tech in a similar way, through making friends on forums, what happens when... I never met those people, right? I don't know a lot of their names ever. So if they were AI, would it have made a difference? I mean, we had a whole philosophical conversation about this, but he was like, you don't actually want to talk to people, there's traits of being human that you like, and AI will quickly copy all those traits.

1:20:49And I'm like, mind blown. And also no, no, thank you. But is it inevitable? That's the question. It ignores like, in the 80s and 90s, there were the 900 numbers and you could like, call like the Crypt Keeper was one like Corey Haim and stuff like that. And Corey Haim was not answering the phone to talk to a little 12 year old girl who had a crush on him right like that was a voice actor pretending to be him and quickly wait so we just learned something very important about what you spent your money on when you were a child it was not me because if i had called a 900 number i would have not had a good time my dad super caught me calling a 900 number once he was like you're never using the phone again but that was like part of why i mean it was one that was the cost of 900 numbers but that was the other part of it where it was like 900 numbers were a little sketchy as a kid you weren't really supposed to call them and it was because like yeah it was a bunch of like not con artists but basically licensed con artists pretending to be other people and it's like that's that's gross the the you hear this from tech people all the time you're like this is just the same as this other small problem that we had in the past like but now you're doing it at scale yeah like now you're doing it at scale and that's like what if we took a small problem and made it global it's like i think maybe there's something different here this thing by the way that you just mentioned it's come up several times now it is this unshakable belief that these models are going to get better yeah and like on the rate of like moore's law we'll get there and i don't that metric doesn't exist like there isn't a moore's law of llm models yet you know we've only really experienced 3.5 to 4 in the public but i and i understand that they've all got stuff they're looking at that we can't see sure someone's got to give me the number like give me the rate of change before i believe that this model will be able to simulate whatever happened to me on forums when i was a child like you know like you wouldn't i guess the minute in the asylum when they raised me it's like i don't think so well so yeah there's no moore's law there's

1:22:53i mean there there's no moore's law there's scaling law and the idea is that as you continue you to put more data and compute into these models, they get exponentially better. That continues to hold true. If you compare the output, the creative output of like GPT-2 or GPT-3 to 4, it's a massive difference. But they're also running into how to... But you're saying that line is not linear. That's what I'm saying. I'm saying everyone who works in AI believes that it is still linear. But it's not because we've just seen multiple stories about how they're all desperate for data to the point that they're like, what if we make AI machines to just create more content for us?

1:23:27that can't go wrong. Oh, the thing I'm hearing is it's going to be synthetic data for training. So literally you're going to have the AIs get good enough that they start talking and then the AIs train on the AIs. That's where we're going. I can't stay here. I'm just saying make real friends. Go to the forums while you can. All right. Get in there while the getting's good. Get past the username. Get their actual name. Because they may be an AI to hear. Request a driver's license immediately. We're taking a break. We'll be right back with the lightning round.

1:24:25LinkedIn's new features can help you write job descriptions and then quickly get your job in front of the right people with deep candidate insights. Either post your job for free or pay to promote. Promoted jobs get three times more qualified applicants. At the end of the day, the most important thing to your small business is the quality of candidates. And with LinkedIn, you can feel confident that you're getting the best. Based on LinkedIn data, 72 % of SMBs using LinkedIn say that it helps them find high-quality candidates. Find out why more than two and a half million small businesses use LinkedIn for hiring today.

1:24:59Find your next great hire on LinkedIn. Post your job for free at linkedin.com slash track. That's linkedin.com slash track to post your job for free. Terms and conditions apply. Support for this show comes from Darktrace. Defenders in cybersecurity are always there when we need them, keeping the systems that run our daily lives safe from bad actors. At Darktrace, they think defenders everywhere deserve a bigger spotlight for keeping the lights on. They deserve blockbuster movies about breaches that never happened. And they should get a parade every time they block a novel threat. But most of all, they deserve AI cybersecurity that can stop novel threats before they become breaches.

1:25:42Across email, clouds, networks, and more. That's why Darktrace is going beyond for cyber defenders. such as IT decision makers, CISOs, and cybersecurity professionals to give them the power to see across their entire attack surface and stop the threats other platforms miss. Darktrace is a cybersecurity defenders deserve and the one they need to defend beyond. The world needs defenders. Defenders need Darktrace. Visit darktrace.com slash defenders for more information. Support for this show comes from Figma. Figma is the collaborative design platform where teams come together to turn ideas into the world's best digital apps, websites, and experiences.

1:26:26And now, there's a faster way to bring those ideas to life. Introducing Figma Make, an AI-powered prompt-to-code tool that helps you turn ideas into something real, fast. Figma Make is a different kind of vibe-coding tool. With just a few prompts, you can turn any idea into a code-backed prototype or app that anyone can explore, iterate on, and share. And it's all connected to Figma. So you can start from scratch or from your design file. Figma Make is made for designers, product managers, and builders who want to move fast and explore more ideas. Try it now and see how quickly you can make your ideas real with Figma Make.

1:27:04Just go to Figma.com slash Vergecast. That's F-I-G-M-A dot com slash Vergecast

1:27:17All right, we're back. It's the lightning round. I will say the first segment of the show was much more like lightning round, and this is just going to be different. It's fine. This lightning round remains unsponsored, but many options exist. For you, the potential advertiser, but you can't do AI ads. Yeah. You have to let me do it. You can only do AI ads if they're trained on Eli's voice. What's the most dangerous advertising you can buy? It's letting me do the ads. All right. Kranz, let's start with you. NVIDIA is super rich. Can I say, so the story, right, is NVIDIA has overtook Microsoft and Apple as the world's most valuable company.

1:27:55Microsoft and Apple have been going back and forth on market cap for a while. NVIDIA is not bigger than both of them. The crazy thing about this whole story to me is, one, NVIDIA has been around for a million years. video game fans exist right yep like people put nvidia cards in their pcs all over the place it has been on this run for a long time right yeah like this is years now of nvidia stock skyrocketing and the way cnn covered it the headline in cnn was this company whose name you can't pronounce is now the world's most richest company it's like yeah it's actually pretty easy to pronounce i feel like yeah it's not one it's not that hard that's easier than ssi an editor and a writer together and they're both like i don't know what this is we got this it's just like at some point i understand we run a tech publication we're like swimming the water but like they've been headed here for at least two years yeah yeah we had to hear about them like cryptocurrency right nvidia had this big all these big stock jumps because of cryptocurrency because everybody was buying the gpus because that was the best way to train oh my god this was completely out of my brain i this was gone this information you wiped it yeah because it It was 2020.

1:29:02Just like crypto. We don't think about 2020. And so that was happening to NVIDIA. And so it was like steadily ticking up. And it was like, oh boy, AMD and Intel might be able to beat it. And then it was like, we're going to take all the crypto stuff out of our GPUs. And then everybody was like, hey, AI is here. Yeah. And others have been... Lost their minds. The effect of this inside NVIDIA is fascinating. I have heard, and there's been some reporting, and I want to keep doing more reporting on it. So if you have stories, if you're listening to this and you have gotten fabulously wealthy at NVIDIA, hit me up.

1:29:36We can talk off the record. They have a retirement problem inside NVIDIA. There was a post I saw where it said if you joined as a mid-level software engineer five years ago and your total comp was$700 ,000, which is a lot, right? But in tech, that's actually pretty middle of the road for a good engineer. Your compensation yearly now is$10 million. That rules. they actually have people just retiring like in mass. And so like the question for NVIDIA is like, how, what do you do to motivate people when people have gotten so fabulously wealthy so fast? And it's obvious there's going to be a correction.

1:30:13It's obvious NVIDIA is not going to stay the richest, the most valuable company in the world. Right. So how does that collapse happen? Is it a slow collapse? Is it a fast one? And what happens to the employee base is fascinating. Well, you get rid of management overhead and product cycles. There you go. That's obvious what you do. Merge with SSI. It's going to last a while, right? Because they are the only ones making the really good AI processors for server levels, right? Well, they have software lock-in, too, with Google. Yeah, like Google's their main primary competitor, right? Well, yeah, ARM, AMD.

1:30:49I would say AMD is like an arms dealer is probably the biggest one. Google doesn't sell TPUs really as much. But NVIDIA's got a big head start. This is one of the things that has kept Intel propped up for years was because Intel had this huge head start in the server space, 99 % of the server space. So they can bomb a couple of years against Qualcomm and be fine because they got the servers. And I think NVIDIA's kind of feels like not quite as cushy a position, but similarly cushy. Yeah. I mean, they have CUDA. Like what they have is a software layer that makes their GPUs more valuable. Yeah. Which AMD does not.

1:31:27But if you run Azure, I mean, we've had all these people on the show and Alex talks to them all. If you run Azure or AWS or Google Cloud or whatever, you do not want to be beholden to CUDA and NVIDIA. Nope. You want all kinds of things to happen. That's why Apple and NVIDIA don't look at each other anymore. It is very funny that Apple and NVIDIA hate each other. It used to be just like a little funny because Apple would insist that bad graphics cards were good graphics cards. And that was just funny, like in its way. Now it's deeply funny. It's extremely good. Because all that open AI integration is happening in H100s that Apple does not want to talk about.

1:32:04Yeah, and doesn't want to pay for it. And that's the thing. Market forces will ensure that NVIDIA does not stay the top GPU supplier in the world because they are too expensive. and no one wants Jensen being the king of GPUs, right? Everyone wants to have options. And so NVIDIA's challenge to justify this valuation and to keep growing is to build lock-in. And it would make a lot of sense for them to start competing directly with Google Cloud and Azure to become a one-stop shop because otherwise people will out-chip them. That's just going to happen. It may take time. And we've been seeing that for years with them.

1:32:37But on the gaming side, they got really, really good at the lock-in. They got really good at being like, You can't use these AMD processors because we went out to every single game developer and made sure they were using our tools. So you had to use – we get the better stuff. So it's going to be really interesting over the next couple of years as they just absolutely duke it out with some of these other companies. And I think a lot of people are going to be surprised at how vicious they get. Oh, yeah. But it's like one step down from that. Right, Alex? Yeah. Heath was saying these things are free.

1:33:08Like Apple's not paying for it. But if there's no business model one step down from we paid all of the money in the world for H100s, the price of the H100 will just fall. Like if the demand is there right now because everyone thinks they're going to monopolize something, what it is remains to be seen. But that's why the price is high. We'll see. All right. I'm going to go next because I want Heath to end the lightning round with this. We just continue to track tech litigation. the United States Department of Justice has sued Adobe for deceiving subscription pricing in Creative Cloud. And it actually is deceiving.

1:33:47It is deceiving in one specific way, which is when you sign up for Creative Cloud, you can pick annually billed, annually billed monthly, or monthly. Annually billed monthly is the default choice. So that means you pay for a year, but they charge you per month, and you've signed a one-year contract. And if you try to cancel early, they just charge you the rest of the year or you can sign up for monthly where you can just cancel after every month because it's the default because it's the cheapest uh people pick it and they think it's a monthly subscription there's no fine print it doesn't actually disclose that this will happen they try to cancel to get stuck with the rest of the bill this is bad this is straightforwardly bad for a company with basically a monopoly market share in these tools it is also just beyond the Like, this is piddly shit.

1:34:33What are you doing? Like, people are going to buy Photoshop, man. That's a local gym move on their part. I really think about, I've been thinking about this a lot with Adobe. I've basically insisted The Verge over cover Adobe over the past few years because they're such a central player in everything that's going on. Like, influencer media depends on the existence of Adobe. That's weird, right? Premiere runs YouTube. Like, YouTube never built creative tools. which is one of the weirdest things about YouTube. They had that one app on the phone once, but they've never built a video editor. The success of TikTok is they made a video editor that's really good.

1:35:12They made all those templates and they made it easy to do all this stuff. YouTube never did that. They're like, upload some videos and then people built the entire creator's ecosystem on the back of Premiere, largely, and a little bit of Final Cut because there's so many creators who use Macs, but mostly Premiere. And Google never built that stuff. So you just like look at this industry. Adobe is at the center of everything. They're at the center of all these AI creativity battles, right? Because they have Adobe stock and people use Photoshop and Generofil. And the more you cover Adobe, the more you realize there's a massive difference in how the company perceives itself and what it does and the tools it makes and how everyone else perceives Adobe.

1:35:49It's a yawning gap. I've said this before. We had Shantanu Ryan on Decoder. He never talks to anyone really. He shows up on CNBC after earnings reports and never gives wide-ranging interviews. I was so excited for that interview. This guy never talks. We asked him all the questions. Some of those questions were pretty harsh. The audience was mad at me just for paying attention to Adobe. They're like, why would you platform Adobe? Which is an incredible question because it's Adobe. I don't think I've seen a company burn that this much goodwill ever. Right? And then on top of it, they had their terms of service debacle where we've covered terms of service changes like this so many times.

1:36:28Like I could barely even pay attention to it. where they change their terms of service to let them do content moderation on their cloud services. So if you run a cloud service, people will put bad things on your cloud service. That is just as true a fact of nature as there is. Like horrific shit will end up on your cloud storage system if you have a scaled cloud storage system. And what you don't want to do is give a bunch of contractors PTSD, which is a real thing that has happened that we have covered by making them look through stuff on cloud services or look through reports. So you build automated systems to go through your cloud services and find the child abuse material.

1:37:03Like, that's what you do. So Adobe changes its terms to let people know it's going to do this in Creative Cloud. People assume that they're going to train on AI, and they freak out at Adobe. And Adobe has so little goodwill that even as they tried to explain this, people are like, we don't believe you. And so now they've updated their terms of service again. And you just connect that directly to, oh, you're getting sued by the government because annual paid monthly was not clear to people. And you charge them the rest of the money when they cancel. It's just like – it's crazy to me that this company is so important but does not realize that it needs to maintain the goodwill of the people that depend on it.

1:37:41Well, you don't have to when you're a monopoly. You can just be a monopoly. But it doesn't make you – I've met all their executives. They're not dumb. Yeah. It's like baffling to me. Yeah, but they're – they can do what they want. They're monopoly. What are you going to do? Go to DaVinci? Maybe. Yeah. Maybe. A lot of people, yeah. Them not acquiring Figma, I think, will go down. We're going to come back on the failed Figma deal. One, because Figma will turn into a different company, and we'll see if they can go compete with Adobe. But that was driven as much by some of this burned goodwill as anything.

1:38:19And I think we're going to see the creative tools industry bifurcate on that moment over the course of history. because if they had acquired Figma, a lot of things would have been a lot different, I think. All right, last lightning round one. We got to end here. Heath, this is like the best story in history. Okay, yeah. The San Francisco Standard has a fun story where techies are increasingly paying up to$1 ,200 an hour for fashion consultants because of CEOs like Mark Zuckerberg deciding to wear chains. And I just encourage everyone to read this story. There is a side-by-side photo of a guy who's gone through one of these fashion consultants, I would argue he looked better before.

1:38:58I'm not going to judge, whatever. Or at least the same. It's just like, as my grandfather would say, money doesn't buy cents. And reading this, Techie's like, just be who you are. You don't need to try to look cool. And you don't need to pay$1 ,200 an hour to try to look cool. The internet is there. You could just research on your own. This side-by-side photo is brutal. It's brutal. It's not brutal because he looks better or worse. He looks the same. Yeah, the same. Both of them are kind of just baggy outfits. Yeah. And then it's perfect because at the end, the story, they're interviewing a founder who's done this.

1:39:36And then he's like, I'm making an AI app that will do this for people for$20 a month. And it's like, the circle just completes itself. That's amazing. You know, Amazon tried to do that with their weird camera. Yeah. Super didn't work. Yeah. I would say if you're paying someone this much money to take you to Bloomingdale's, you can do better than this photo. That's what I'm saying. Oh, I don't think he went to Bloomingdale's. No, it does. That's a question. He definitely doesn't. Yeah. He's in the dressing room at Bloomy's. Oh, my gosh. Get out. Leave the store, sir. No, that's a bad look if that's what you're pulling out of Bloomingdale's.

1:40:12That's like, you get that at Dillard's. That's brutal. All right. We got to end it here before we do more streaming. Sorry for your fashion shaming. Dillard's is fine. Shop where – you can look cool no matter where you shop. I love a Dillard's. Everyone is constantly asking me where the jackets come from, and all of them were pandemic purchases at either Home Depot or Walmart. I'm just telling you that's like – I had money to spend, and there are only two stores to spend it at. It's like you can look cool no matter where you shop. Dillard's is fine. Just know who you are. That's a quote. That's it.

1:40:43Just know who you are. Just be that person as loudly as you can. All right. That's it. And also make an agent of yourself and have it flirt with other people on Instagram for money. That's the future. More gold chains. More gold chains. All right. We'll be back on Tuesday with more of the Windows ARM laptop benchmark scores. That's a big deal. I'm excited for those reviews to come out. And then we got all kinds of stuff. It's a summer. It's like wild time in tech. And then we're going to get rid of all of our management overhead and product cycles. Just pure intelligence. VVI. Thanks to Heath for filling in.

1:41:17And good luck to David doing whatever he's doing. That's it. That's Rich Cast Rock and Roll.

1:41:24And that's it for the Verge Cast this week. Hey, we'd love to hear from you. Give us a call at 866-VERGE-11. The Verge Cast is a production of the Verge and Vox Media Podcast Network. Our show is produced by Andrew Marino and Liam James. That's it. We'll see you next week.

1:42:05Liberty Mutual Savings vary. Underwritten by Liberty Mutual Insurance Company and affiliates. Excludes Massachusetts.