The Verge's Nilay Patel, Alex Cranz, and Alex Heath discuss Apple's Vision Pro team reportedly refocusing on a cheaper headset, Meta launching a new "Wearables" organization, a new AI company startup from former OpenAI chief scientist, and a whole lot more tech news.
Category
🤖
TechTranscript
00:00:00Hello and welcome to Verchast, the flagship podcast of the ARM architecture.
00:00:08That'd be great.
00:00:09And then if anyone knows of a rival podcast, there's a flagship podcast of like RiskV.
00:00:16I would love to know who they are, so we can start a fake beef.
00:00:20But only RiskV.
00:00:21Right, no one wants an x86 podcast.
00:00:24Yeah, no one wants that. We only want the cool, weird ones.
00:00:28If anyone knows of the flagship podcast of the x86 architecture, we'll send them a gift basket.
00:00:35Someone ought to be nice to them.
00:00:37Hi, I'm your friend Eli. Alex Kranz is here.
00:00:39Hi.
00:00:40Alex Heath is here, not David Pierce. Hi, Alex.
00:00:43Hi, I'm not David. But happy to be here.
00:00:46David is off this week, but we've got Alex, which is going to be great.
00:00:49There's a lot of AI news. There's a bunch of VR news, including some scoops out of meta and the reality labs division that Alex is here to talk about and have some insight into.
00:00:59But we should start with like the news, which is a weird bit of news in that half of the news got delayed.
00:01:08Weird. Microsoft's Copilot Plus PCs are now shipping.
00:01:12They have Qualcomm Snapdragon X chips inside them.
00:01:16This is the beginning of what might be the end of the Intel dominance, the x86 dominance for Windows machines.
00:01:25But Microsoft had to delay recall, which was the flagship feature of the Copilot Plus PC because of horrible security concerns.
00:01:33Just a weird moment, Kranz, right?
00:01:35Yeah.
00:01:36A weird stutter. Even the review units didn't arrive on time because they were disabling recall.
00:01:40Yeah, like this is wild. We were expecting to get these a week ago to start kind of reviewing them before the embargo lift, and we got them the day of the embargo.
00:01:48Tom's got both the Elite and the Plus, and is checking those out.
00:01:54And a whole bunch of other people on the team all have them and are just, we are all in this one room, just furiously benchmarking constantly.
00:02:02They're having a lot of fun doing it, but we've still got a lot of work to do.
00:02:05So far, we figured out that some stuff is working, some stuff isn't. There's still a lot of weird, like, emulation bugs and stuff.
00:02:15I think Adobe Premiere was having a weird issue where it would work with the Elite processor, but not necessarily with the Plus processor.
00:02:24Some games crashing. Some of this might be, like, alleviated with more updates from everybody.
00:02:29But what was supposed to be this big, monumental moment for Microsoft and Qualcomm and everybody, instead was like, we did it.
00:02:38Yeah, here it is. We turned off half of it. Even in Best Buy, there's, like, signs that say recall coming later on the demo units.
00:02:45So if you don't know, if you were listening to the show in the previous weeks, Microsoft announced these PCs, and the flagship feature was recall.
00:02:53Which, even if it was working in a totally secure way, is weird.
00:02:58The computer watches everything you do on the computer. Like, Windows watches everything you do.
00:03:02Takes screenshots of it. Uses a local generative eye model to understand what's in those screenshots.
00:03:08And then you can just be like, there was a website I thought I liked. Do you remember what it is? And the computer will tell you.
00:03:13Or, I was working on this document. Do you remember what it was? And you can talk to the computer about what you've done with it.
00:03:17Which some people, David Pierce, very excited about. Because David would like to be friends with a computer.
00:03:24That's my understanding of his personality.
00:03:26Just friends.
00:03:28Just friends. David is happily married and has a child. Other people would like to be more than friends with a computer, as we have learned time and time again as the AI industry has dominated the headlines.
00:03:38Microsoft was storing those screenshots in basically plain text.
00:03:43So there's this open database of everything you've done on your computer on your computer.
00:03:48Huge security hole. Microsoft has had a history of security problems recently.
00:03:52Nadella sent out a memo to everyone being like, if you have to choose between shipping fast and security, I want you to pick security.
00:03:59And then they announced recall and everyone was like, this is a huge security flaw. And they just deleted it.
00:04:03So they thought they could fix it in time. They issued a blog post about fixing it in time and then they realized they couldn't and they deleted it.
00:04:09I feel like that's only half of the story here. That's just the background. That's part of the stutter step and part of why it's not as flashy to launch.
00:04:19Recall is a great feature for what I think of as the local news.
00:04:23This is a great demo for the local news and that is how you get a lot of mainstream attention on new things.
00:04:29But the real thing here on all of these Snapdragon PCs is performance and battery life where they have not been competitive with M-series chips on MacBooks for a long time and now they think they are.
00:04:41Yeah, that is absolutely the big moment here. And the other part was going to be really cool and that was going to be what got people really excited to maybe upgrade.
00:04:50But for the rest of us, it's okay, is this finally the moment where ARM on Windows works?
00:04:56Because we've seen it so many times before and every time it has fallen super short of the mark.
00:05:03And this time from the early benchmarks, battery life looking pretty good. Batteries in that M-chip range.
00:05:14We're still testing so I don't want to say which M-chip, but it's in that M-chip range.
00:05:19The speed is looking pretty good. Gaming is like possible. We've still got some work to do there to figure it out, but it's a surprise.
00:05:29It definitely doesn't seem like this is going to be a moment where just Qualcomm blows everybody away, kind of like how Apple did with the M1 chip where we were like, holy crap, Apple actually did it.
00:05:38Doesn't seem like we're going to have that moment, but it's still kind of too early to tell.
00:05:43Well, it's interesting because Apple hasn't even had the incremental moments after the M1.
00:05:47Like the M2 to M1, it's faster, but it wasn't the big step from Intel to ARM.
00:05:54M3 to M2 was faster, but it wasn't the same kind of revelatory experience.
00:05:59So Qualcomm's coming into just a very different set of expectations.
00:06:03Yeah.
00:06:04Like if you can match the M1, you've done great. That's a big step forward over Intel.
00:06:07If you can beat the M3 or the M2, it's the same incremental progress as everyone else has made.
00:06:15Yeah. I think that's kind of normal. We sometimes expect everybody to just blow each other out of the water, but Windows is hugely complicated to run.
00:06:24It is not an easy thing to run. There's so many dependencies, code for old code that it has to work through and stuff.
00:06:33And then a lot of times that stuff isn't built for ARM.
00:06:37And so they're having to do all this emulation, more so than a Mac would.
00:06:42Because a Mac, it's all built from Apple.
00:06:45Not all of it, but most of it's built by Apple for Apple, and it can kind of run and knows what to do.
00:06:51That big translation moment hasn't quite happened for Windows machines.
00:06:57They've been trying for a long time.
00:06:58They've been trying, and nobody's wanted to do it because it's like, well, nobody makes good ARM chips, so why should I build my programs for ARM?
00:07:06And now it's like, this could change that. This could change that. But it is so, so, so early to tell.
00:07:14Yeah. I get the chip story underneath this, but does Microsoft actually think that anyone is going to switch from Mac to Windows because of this AI stuff?
00:07:23Because I look at this and it's like, you basically need to tell me that I'd never have to actually understand or know how to use Windows to actually switch to Windows.
00:07:31That's how I feel about Windows.
00:07:33It needs to be so dramatically different from what I expect from a laptop as a lifelong Mac, Apple lock-in, blue bubble guy.
00:07:43And this recall thing sounds cool, but you would literally have to say AI is running your laptop for you for me to consider switching because of the software locking.
00:07:53This is a pressure dressing, right?
00:07:55This is just for them to be like, no, we can't lose any more Windows people to Chrome or Macs.
00:08:01We have to keep them here with cool features.
00:08:04It's not like, let's entice these people back. It's, please don't leave.
00:08:09Yeah. Well, no, actually, I wonder about that.
00:08:12At least that's how I feel about it.
00:08:14I've heard a lot of CEOs, Alex, I'm sure you have too, describe AI as a platform shift.
00:08:20And I'm always like, from what to where?
00:08:24Like, what platform are you talking about?
00:08:26And what they want, what they're trying to invoke is mobile, right?
00:08:30This is a platform shift on the order of mobile.
00:08:32We went from laptops to smartphones and now smartphones dominate computing.
00:08:36And what they're trying to say is this is another platform shift on that scale.
00:08:41And it's like, is it?
00:08:43Where? You still have to have a phone to use the chatbot.
00:08:48Is it that?
00:08:50And I think part of this Windows stuff for Microsoft is very much, okay, we own this entire stack.
00:08:55We have this giant investment in open AI.
00:08:57We've built Copilot. We have all these ideas about what a computer can be.
00:09:00Here's a new way to use a computer that is meaningfully different than before.
00:09:06Also, you have to pre-install Edge or whatever.
00:09:09Also, we'll push you towards our browser.
00:09:10I don't know that Apple has different ideas.
00:09:13We talked about it in the show after WWDC.
00:09:15Apple's ideas for what AI can do in its operating system are basically the same as Microsoft's ideas.
00:09:20So the idea that it's a platform shift, I think, is under some pressure.
00:09:25But I know that's how these companies think about it.
00:09:27That this will entice you to switch platforms.
00:09:29But it's just features right now.
00:09:31It's like, I still have to look at the Windows nav bar and figure that thing out, right?
00:09:34And it's like, I'm just not going to do that.
00:09:36I'm sorry, Tom Warren. I'm just not.
00:09:37So, yes, I think they think of it that way.
00:09:40But I think this recall thing shows where the real platforms are.
00:09:45And it's like building two decades of software lock-in on top of hardware, which is not AI, right?
00:09:53AI is software that you're putting on existing hardware, at least right now.
00:09:57Yeah. And I'm curious.
00:09:59People are listening. If you have an idea of where this platform shift is, I'm dying to know what people think.
00:10:03But it is what Nadella has told me.
00:10:05It's what Sundar Pichai has told me.
00:10:07Down the line, the CEOs of the biggest companies are like, this is a platform shift.
00:10:13We cannot get left behind.
00:10:15In particular with Microsoft, there's a fear that they got left behind in mobile and they can't get left behind this time.
00:10:22And so you see them kind of – I mean, they have been ahead of the curve at every single step.
00:10:27You didn't hear Elon say it's going to be trillions of dollars worth of Optimus robots?
00:10:30That's the platform shift.
00:10:32I mean, that's the true platform shift.
00:10:35It's kicking back with a brewski and letting Optimus do all your computing for you.
00:10:39It would be great if AI expressed itself not as AI integrated into the operating system,
00:10:44but you have an Optimus robot that knows how to use Windows.
00:10:47And that's actually what happens.
00:10:49I'd love that.
00:10:51That would be Alex Heath. And then Tom Warren would get the one that knows how to use Macs.
00:10:54Yeah. And Nila, you joke about how you would love to not have to use software at work.
00:10:59You just have an Optimus use your software for you.
00:11:01Yeah. I need a fully complete Optimus robot that can use Okta.
00:11:07And if you can just get me there, I'll do it.
00:11:10If Optimus can file my expenses for me.
00:11:13Oh, my God.
00:11:15Those codes, they suck.
00:11:17The CEOs of these companies will not come on Decoder where I will just ask them to look at their own software.
00:11:23We'll see. We're going to test them out.
00:11:25I will say that I have a running list of my favorite Slack rooms at the company.
00:11:30And it's the ones where people have developed their own language.
00:11:33So my number one favorite is we have a support room for our finance team where people just file tickets.
00:11:39And some people have discovered – well, some people come in there and they ask very nice questions.
00:11:44Like, hello, I need help with da, da, da.
00:11:46And some people have discovered that this room is staffed by a bot that just files tickets.
00:11:49And they just yell nouns.
00:11:51They're just like, invoice 245.
00:11:52Like, that's all.
00:11:54And the other people are very polite.
00:11:56It's the greatest room in the entire company because people realize they can just yell at a robot.
00:11:59The other one is the Verge Benchmarks room where it's just people communicating in increasingly arcane benchmarks.
00:12:05And they've developed their own language about Windows Unarmed benchmarks.
00:12:10And they're not using the normal rule.
00:12:13Like, no LLM can figure out what's going on in this room.
00:12:16Statistically, this language has never existed before.
00:12:18It's very good.
00:12:20I'm excited for it.
00:12:22We're going to have all the reviews.
00:12:24I think David is going to do a full show with Tom and Nathan, our reviews editor, just running down all this stuff on Tuesday, I believe.
00:12:31Right, Alex?
00:12:32Yeah.
00:12:33OK.
00:12:34So we'll come back around with that stuff.
00:12:36Qualcomm has made a lot of promises here.
00:12:38They bought that company Nuvia, which was a bunch of ex-Apple engineers on a chip team who were making server chips.
00:12:46And they said, no, make a consumer chip instead.
00:12:48Because Apple didn't want to use its ARM architecture to make server chips.
00:12:52And they thought there was this big opportunity there.
00:12:54But then Qualcomm bought them and said, no, make the next generation of Snapdragon.
00:12:57So they shifted them.
00:12:59I don't know if you know, there's some backstory here.
00:13:01ARM decided that Qualcomm's license didn't cover this.
00:13:04And so ARM is suing Qualcomm right now.
00:13:07And Qualcomm is suing ARM back over these chips, which is bananas.
00:13:13It's going to come to some resolution.
00:13:15There's no way this actually ends up in anything but payments.
00:13:18But these chips are very, very controversial in their way.
00:13:23Because it's a bunch of ex-Apple designers who left and started a company.
00:13:26And Qualcomm snapped them up to go compete with Apple's consumer chips.
00:13:30Can they deliver is a big question.
00:13:33I'm excited to see our reviews and see if that's true.
00:13:35Same.
00:13:36Also, this lawsuit is just the funniest lawsuit.
00:13:38It's a good lawsuit.
00:13:40If you're ARM, you're like, screw it, let's sue Qualcomm.
00:13:42I feel like Qualcomm gets sued a lot.
00:13:45They do.
00:13:46But Apple is like, no, I hate you.
00:13:48Broadcom is like, I'm suing you.
00:13:50Apple is like, I'm suing you.
00:13:52This company is under fire.
00:13:54And by the way, we talk a lot about 5G on the show.
00:13:56The level of interest in 6G is at least partially driven by people wanting to get away from Qualcomm's 5G patents.
00:14:04What if there wasn't a dominant player in wireless that we could get away from?
00:14:11And so there's already talk about 6G just to get away from Qualcomm.
00:14:14Apple bought all of Intel's modem division, and they still haven't produced a radio that can compete with Qualcomm and its patents.
00:14:21Get ready for the 6G hype cycle.
00:14:23That's all I'm saying.
00:14:25People love suing this company, and that's going to drive the 6G hype cycle.
00:14:28Some other stuff we should talk about.
00:14:30This is kind of like a gadget rapid fire round.
00:14:33Yeah.
00:14:35It's not quite a lightning round.
00:14:37We've learned some new stuff out of WWDC about various Apple products.
00:14:41Jen had a pretty good scoop.
00:14:42Apple is extending HomeKit to use the ultra wideband chip in its phones to automatically unlock doors, which is cool.
00:14:51So a combination of ultra wideband and Bluetooth low energy, it can sense when you are six feet away from your door on the outside and walking towards it and automatically unlock the door.
00:15:02Which is just sick.
00:15:04If it works.
00:15:06But if you have a smart door lock, you need a new door lock to enable this.
00:15:10Yeah, none of them work with it.
00:15:12It's a lot.
00:15:14I got two Alex's now.
00:15:16This reminds me of the digital ID driver's license thing.
00:15:21It just seems like a bad idea.
00:15:23I get that it sounds cool, like you're in a product meeting in Cupertino and it's like, oh, we should digitize this.
00:15:29But I don't really want my device automatically unlocking my house.
00:15:34Yeah.
00:15:35Yeah.
00:15:37So the argument, I felt this way about the feature where you can just point your phone at a door lock and it unlocks in express mode.
00:15:47The same way that if you use mass transit, you can just beep your phone to turn on the subway and it just pays and whatever.
00:15:52And I was like, I don't know.
00:15:54I definitely want to authenticate before.
00:15:56And I was like, no, if you steal my house key, you're in my house.
00:15:59The level of security is the same.
00:16:00And I think one good question is whether that is an acceptable level of security.
00:16:06It is, I guess, is the answer.
00:16:09If you have my phone, should you be able to get in my house?
00:16:12It's a good question everyone should ask.
00:16:14And then you can adjust that setting however you want.
00:16:16If you have my phone, you're six feet away from my house and you're walking towards it, should the house automatically unlock?
00:16:21It's a different question.
00:16:23That's a new problem.
00:16:24I would like this feature if you could set it to where if Face ID hadn't been used in, say, an hour or two hours, then it doesn't work.
00:16:33But something like that, something where the recency of me unlocking the device means that it's definitely in my hand.
00:16:39Or your phone has been stolen and another iPhone detects you shrieking because there's a void and your house locks down.
00:16:47Like the bizarre use of the Find My network.
00:16:51It's interesting because Apple has put a lot of chips and a lot of radios and a lot of things recently and hasn't really done a lot with them.
00:17:00So these ultra-wideband chips have been in a lot of Apple devices for quite a while now.
00:17:05And some of the uses are cool.
00:17:08Find My is a little bit more precise if you have the right selection of ultra-wideband devices.
00:17:14You can point your phone at a HomePod and send the music to it, which is supposedly like magic.
00:17:20It works half the time.
00:17:23I said supposedly like magic.
00:17:25Like if you were a shit magician.
00:17:27Yeah.
00:17:29If I did magic, that would be my magic trick.
00:17:32You know those people who are like, I'm floating and they're really just like standing on their tiptoes.
00:17:36That's me. That's what I'm saying.
00:17:38But if you're always doing that at sort of the wrong angle and someone's like, you're just standing on your tiptoes.
00:17:41That's about how well the HomePod thing works.
00:17:44Then they added Thread Radius.
00:17:46That was the other gen scoop from a couple weeks ago.
00:17:48Now they're adding other radios to these devices.
00:17:51You see that they're like, okay, Wi-Fi and Bluetooth can't do all the things we want to do.
00:17:58They can't do it at the power level, at the location precision level, whatever it is.
00:18:04They just can't do it.
00:18:06So we need to add more radios to this thing.
00:18:08And then you see the animations from the new Siri.
00:18:11And you're like, oh, they are super getting ready to take all the ports and buttons off these phones.
00:18:16Stop! You're willing it into existence.
00:18:19There's 45 different radios for every level of precision location and power that you can think of.
00:18:26None of them are able to do much.
00:18:28And then the buttons are going to be capacitive very soon.
00:18:31The arc of phones is that they just become panes of glass that then start to curve that we then put in front of our eyes.
00:18:39That is the next 10 years of technology.
00:18:41I want to come back to that when we talk about Meta and the Vision Pro to some extent.
00:18:47So that's the UWB chip.
00:18:49We're seeing Apple start to use it in more interesting ways, especially around the home.
00:18:53I'm very curious if they ever acknowledge these thread radios.
00:18:56We were kind of hoping they would at WWDC.
00:18:59But it's interesting that there's a bunch of thread radios in these devices.
00:19:02You could do all kinds of things with in a smart home because it used basically no power compared to everything else.
00:19:07And then the other thing we found out, which is really interesting, speaking of Apple and wireless,
00:19:12is that the next generation of CarPlay that no one has shipped yet will only be wireless, which is fascinating.
00:19:20I'm obviously obsessed with CarPlay.
00:19:22Car CEOs are always on the coder.
00:19:24I'm always asking them if they're going to use it.
00:19:26They're always like, we'll see.
00:19:28Just to reset everyone's understanding of CarPlay, because it's very confusing right now.
00:19:34There's the CarPlay you have right now, which is basically just like a second monitor for your phone.
00:19:40That's very much what's going on.
00:19:42So you plug into your car, even in wireless CarPlay, your phone sends out another video stream and receives touch input from your car,
00:19:48and you've got a second monitor for your phone.
00:19:51And if you think about it, almost everything that's happening on that screen is just phone stuff.
00:19:54It's music, it's maps, but it's all just running on your phone and showing up on this display.
00:19:58Not a lot of car stuff over there.
00:19:59Apple has likely extended this to be able to send a map to some instrument clusters,
00:20:05so they can send multiple video streams out from a phone.
00:20:08But it's still phone stuff.
00:20:10Everything's local on the phone without any understanding of the data from the car.
00:20:14The next generation of CarPlay, which they showed two years ago.
00:20:17I didn't even realize this was two years ago.
00:20:19I thought this was last year.
00:20:21Two years ago, WWC 2022, they showed that mock-up of a giant screen and a mouse screen,
00:20:26and it's all CarPlay, and there's like 45 clocks on it, because no one knows what to do with all that screen.
00:20:32The weather in 10 different cities, which you definitely need when you're driving.
00:20:35And they're like, this is the next generation of CarPlay.
00:20:37And then they showed a bunch of CarMaker logos.
00:20:40Do you remember this? They showed a bunch of CarMaker logos.
00:20:42And all those CarMakers are like, what?
00:20:44That was basically what happened.
00:20:46Porsche and Aston Martin announced last year they would ship cars this year.
00:20:51Haven't heard a word about those cars.
00:20:53No idea when they're coming out.
00:20:54And that's it.
00:20:56So Apple had sessions at this year's WWC, and I ran around talking to a bunch of people there just about CarPlay and what's going on.
00:21:04And I can't tell if this is a change in strategy, if this has always been the approach, or if this is just Apple softening its language.
00:21:17But Apple is basically saying to CarMakers, look, people have iPhones.
00:21:22They love CarPlay.
00:21:24What if you use our design toolkit to redesign all of the stuff in your car so it all looked really good together?
00:21:31And then you need an iPhone.
00:21:33Is it the context here that Apple just abandoned Project Titan?
00:21:36So if they're not doing a fully integrated vehicle, wouldn't it make sense for them to make this more of an open?
00:21:42This to me is like them going the route of Apple TV Plus and AirPlay, realizing that we are not going to make a TV.
00:21:49And so therefore we want TV Plus everywhere.
00:21:52Is that not a fair analogy here?
00:21:54I can't necessarily tell how much Titan played into it.
00:21:58This is what I'm saying.
00:22:00I can't tell if this was always the approach, and now they're just softening the language.
00:22:03The technical approach is that you have an iPhone.
00:22:07You connect to CarPlay.
00:22:09Your iPhone will go download a bunch of Apple CarMaker co-branded assets.
00:22:15In the videos they have posted, they refer to this as a co-branded experience.
00:22:18When have you ever heard Apple describe anything as a co-branded experience?
00:22:22That's bananas.
00:22:24Red.
00:22:26Sure, Project Red.
00:22:28Windows on the Mac at the very beginning?
00:22:30That wasn't co-branded.
00:22:32So there's some element of Apple designers and CarMaker designers working together to design speedometers for CarPlay cars.
00:22:39Weird.
00:22:41This is public.
00:22:43This isn't some deep dark reporting.
00:22:44This is me saying, you can go watch the videos.
00:22:46They're public.
00:22:48And they're calling it a co-branded experience where we're going to work together to express your CarMaker brand identity inside of the Apple design toolkit.
00:22:56Weird.
00:22:58It makes sense in that all the cars' UIs are terrible.
00:23:01It does.
00:23:03So the idea is you have an iPhone and you get this upgrade to your car UI so everything looks like CarPlay.
00:23:07Weird.
00:23:09How is Apple going to make any money doing this?
00:23:10For a super wealthy company like this, this is just really good branding.
00:23:17You can show your dominance in design and show these companies how you should do this.
00:23:25The specific way they're going to show their dominance in design is actually my favorite part of this whole thing.
00:23:29The only font you're allowed to use in next generation CarPlay is Apple San Francisco.
00:23:34You can change how wide it is.
00:23:36You can change how stretchy it is.
00:23:38You can italicize it.
00:23:40There's different things you can do with this font.
00:23:42It's a variable typeface.
00:23:44And it's like, yeah, but it's only one.
00:23:46Porsche and GM don't get to use different fonts.
00:23:48Comic Sans.
00:23:50That's weird.
00:23:52So there's an element of control where they're doing a very Apple thing.
00:23:56And then there's an element of complete chaos, which I think is really fascinating, which they have not talked about at all.
00:24:02Which is they will, when you have the next generation CarPlay interface up, and you're like, I want to control my massaging seats in my Aston Martin.
00:24:11You can push the button that says control the Apple San Francisco labeled button.
00:24:16And instead of showing you the next nice CarPlay thing, it'll do a thing called punching through the automaker UI.
00:24:23And it'll just show you whatever garbage Aston Martin massage seats are.
00:24:26Never mind. I take it all back.
00:24:28This is crazy.
00:24:29This is the concession to make it all.
00:24:32I can't.
00:24:33Because what car makers have been worried about is they have to build everything three times.
00:24:36They have to build their own native one because you need to have something.
00:24:39You need to operate the entire car without a phone.
00:24:41A non-negotiable if you buy a car.
00:24:43Then you need the next generation CarPlay version of everything.
00:24:47And then presumably Google will do something for Android.
00:24:50So what are you going to do?
00:24:52And they don't want to build things three times.
00:24:54So Apple's solution is like, well, build it once.
00:24:56And then for some stuff, we'll just let you show the weird stuff.
00:24:59I can't wait for Siri powered by chat GPT in my car to just open the doors while I'm driving when I request the massage seat to turn on.
00:25:09And it just hallucinates.
00:25:11It just goes in reverse.
00:25:13It just ejects you.
00:25:15And then to bring this around, Apple wants this to only work wirelessly because their goal is that CarPlay will be connected, active, and operating the interface of your car by the time you open the door.
00:25:27And if not by the time you open the door, by the time the screens light up.
00:25:30So you get in the car, you sit down, the screens light up.
00:25:32It's CarPlay.
00:25:34It should not be you sit down in your car, turn it on, and then you have some loading process.
00:25:39And so you can only accomplish that over wireless.
00:25:41So CarPlay is only going to be wireless for whatever car makers adopt the next generation of CarPlay.
00:25:46I'm hesitant to say this because I am very against this move.
00:25:50But isn't that also just a precursor to them removing all the ports?
00:25:54They're like, it's got to be wireless because we're going to…
00:25:55It's not not a precursor to them removing all the ports.
00:25:59And it's funny because to make this work, Apple also has to give up some control of what runs where.
00:26:08You don't want your speedometer running on the phone.
00:26:11So your Apple-designed speedometer, or delightful co-branded experience to be more accurate, needs to run locally on the car.
00:26:20So this is the first time I can think of that any Apple assets are going to get loaded onto somebody else's computer system in that way.
00:26:27Outside of, to your point, Heath, the Apple TV Plus app on Rokus, which is where Apple ships an app.
00:26:34And the split is kind of the same.
00:26:36It's like, OK, 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.
00:26:45The speedometer, the turn signals, right?
00:26:47And there's some differentiation in how that stuff works.
00:26:50The truly critical stuff cannot be touched.
00:26:53They call that the overlay layer.
00:26:55That's your turn signals and hazard indicators.
00:26:57The speedometer and stuff like the needles, those assets can get refreshed over time.
00:27:02So those run in a different layer called local UI.
00:27:05And then there's everything that runs on the phone.
00:27:08And then in the middle, there's like, I don't know, you want to mess with your sports settings in your fast car?
00:27:13Like, that'll just be the weird automaker UI shining through CarPlay.
00:27:17And it's like, oh, this kind of feels like a mess, right?
00:27:20It's unclear who will take this and who will use all of CarPlay, like redesign the entire car to be CarPlay, to be Apple.
00:27:30And who will say, OK, we took next generation CarPlay.
00:27:33We fund cool new speedometer.
00:27:36Every other part of the car controls will be the existing UI because we can punch through CarPlay.
00:27:40And 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 CarPlay.
00:27:49They will make an affirmative decision to be like, you know what?
00:27:52That car maker, Toyota, uses all of CarPlay.
00:27:55We're going with them over Honda, which is like, here's your weird heated seats.
00:27:59Well, the third path is Rivian or Tesla, right?
00:28:01Where it's totally controlled by the automaker.
00:28:04And like Rivian is a great example of, I think, you and I both test-driven Rivians, Neil.
00:28:09I think their interface is great.
00:28:11It's very snappy.
00:28:13Obviously, I miss some of the iPhone connection, but I could also see more automakers going that direction.
00:28:20Yeah, but they have to make good user interfaces.
00:28:23Rivian is good.
00:28:25Yeah, no, like Rivian and Tesla made good user interfaces for the most part.
00:28:30For the most part.
00:28:32Tesla's getting weirder and weirder, I would say.
00:28:35Tesla's like you turn the car on and it's like, do you want to play a game on Netflix?
00:28:42But it's a lovely experience.
00:28:44So Rivian just upgraded.
00:28:46The generation two R1 trucks just came out.
00:28:50They've changed.
00:28:52The computing architecture of those is different.
00:28:54But then they also updated the interface to be more self-shaded.
00:28:57I mean, they have Unreal Engine and they're using it.
00:29:00That said, hello to the parking lot at Apple Park when we were at WWDC.
00:29:05A lot of those people have Rivians, right?
00:29:08And there's all these rumors about Apple and Rivian doing stuff together.
00:29:11Apple Music just came to the Rivian interface.
00:29:13If Tesla's never going to break, I feel confident about that one.
00:29:19Just a prediction I'll make based on nothing.
00:29:21The Tesla UI is going to be x.com in a few years.
00:29:24That's the direction Tesla's going.
00:29:26It'll be Brock fully just like telling dirty jokes when he tries to drive the car.
00:29:30Rivian is still in the place where they might try some other stuff to claw back some market share.
00:29:37To win some new buyers.
00:29:39And there are an awful lot of Rivians.
00:29:41It's California.
00:29:43People like trucks there.
00:29:45It's something I noticed.
00:29:47All right, moving on.
00:29:49See, it's like a fast round.
00:29:51Not quite lightning.
00:29:53We're just moving.
00:29:55Kranz, tell me about this remote.
00:29:57Because you're very excited about a home remote.
00:29:58You know I love a home remote.
00:30:01I'm still sad about Logitech.
00:30:03I still have the fanciest of the Logitech remotes.
00:30:05And it had soft-touch plastic.
00:30:07Which means it's starting to rot.
00:30:09I have a TiVo remote in the corner of this house.
00:30:11It's not a good situation.
00:30:13It's so bad.
00:30:15The rot on that soft-touch is so gross.
00:30:17But there's a Kickstarter happening right now for the Haptic RS90 and the RS90X.
00:30:22And they promise all the stuff all the big universal remotes promise.
00:30:25Which is like you can control stuff via Bluetooth.
00:30:28You can control it via IR.
00:30:30You can control it via Wi-Fi.
00:30:32Usually that requires some sort of hub that you have to plug in somewhere in your house.
00:30:36This appears to require no hub.
00:30:38And also it's just got a big screen on it.
00:30:40So it's instantly the coolest remote you can have in your house.
00:30:44It's supposed to be able to control all your smart homes and stuff.
00:30:47And like I said, Logitech did this.
00:30:49I've got a control that does this.
00:30:52Doesn't have the cool screen.
00:30:53And it required a hub.
00:30:55So this is instantly cooler.
00:30:58And this also is going to come soon rather than discontinued.
00:31:01Well, it's a Kickstarter.
00:31:03Are they actually going to ship?
00:31:05The Kickstarter part of this is the really big heavy caveat.
00:31:08Because there's a whole bunch of integrations that they're promising.
00:31:11No.
00:31:13Not there yet.
00:31:15There's no Matter support.
00:31:17There's no support for some of the other smaller radios and stuff like that.
00:31:19So this is like cool.
00:31:22Neat.
00:31:24We're like two minutes away from Samsung being like Bixby AI and your TV will not do everything for you.
00:31:30I don't want it.
00:31:32We are though.
00:31:33We are.
00:31:34It's coming.
00:31:35Yeah.
00:31:36Like AI is probably, you know, the reason Logitech got out of this business was because they said nobody wants these.
00:31:40And me and four other people were like, oh.
00:31:42I want my phone to be my universal home remote.
00:31:45That's it.
00:31:46I don't want another remote.
00:31:47I want fewer remotes.
00:31:48Can we make that happen?
00:31:50Yeah, I just, that's how I use my phone.
00:31:52But like, I'm just getting like, who is this for?
00:31:56I guess is the question for me.
00:31:58It's me.
00:32:00What show are you on?
00:32:02It's me and four other people.
00:32:04It's the four people who are still sad that our Logitech remote is gone.
00:32:07Whenever I see a screen like this, I just think about how slow it will be.
00:32:11I think about touching a button and nothing happening.
00:32:13Right.
00:32:14Like, you know, I love it.
00:32:15I'm like, yeah, I want a screen like this.
00:32:16You know that thing you do with a touchscreen where you touch it and the button looks like something happened, but nothing happened?
00:32:21That's this product.
00:32:23Almost like a guarantee that that is this product.
00:32:25It is, I will say, I love that it's called the Haptic RS90.
00:32:29Great name.
00:32:30Great name for a universal remote.
00:32:32Fully good name.
00:32:34I just worry that actually people just use the apps on their smart TV.
00:32:39And what this thing is trying to solve is not.
00:32:42I mean, that's why Logitech got out of business.
00:32:44We can't control the apps on your TV.
00:32:46We can't see your TV.
00:32:48Yep.
00:32:49Yeah.
00:32:50So we'll see how this is.
00:32:52It looks like it's going to be funded.
00:32:54It looks like enough people have invested in it.
00:32:56It looks like it's going to be funded.
00:32:58This is like hot on the heels of that company Brilliant going out of business.
00:33:00Yeah.
00:33:01Which Jen covered.
00:33:03And they were like, we're going out of business.
00:33:05But like the thing we put in your wall will still work.
00:33:07So Brilliant put like a touchscreen in your wall that can control everything, which I was super hype on.
00:33:11Because the idea of having an in-wall Sonos controller screen seems sick.
00:33:16Right.
00:33:17You walk in a room, you're like, I'm going to pick a song.
00:33:19I'm not using my phone, like the whole thing.
00:33:21And then they just like went out of business.
00:33:23And it's like, that is always the worry I have with all these things.
00:33:26Yeah.
00:33:27This is going to be like almost $400 probably when it comes out.
00:33:31That's too much.
00:33:33All right.
00:33:34Speaking of, let's do one more gadget.
00:33:36And then I want to talk about what's going on in AR and VR with Heath.
00:33:38We mentioned RISC-V at the top.
00:33:40And we mentioned companies that might go out of business.
00:33:41Framework is shipping the Laptop 13, which is fascinating because they're going to have a RISC-V chip in it.
00:33:49This is the competitor arm, right?
00:33:51This is the open source competitor arm that has the performance per watt characteristics.
00:33:56I'm kind of dying to try it out.
00:33:59I also have no idea what software I could run.
00:34:01Also, I think one guy pointed out that it'll run about as fast as a Raspberry Pi 4.
00:34:07Yes.
00:34:09You got to start somewhere.
00:34:10Baby steps.
00:34:12Baby steps.
00:34:14Yeah, I have no idea what can run this, but I want to.
00:34:17This is just cool.
00:34:19This is the kind of thing I just want to have one.
00:34:21I like the actual line from the company in the PR.
00:34:28The 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 RISC-V.
00:34:38Don't use this.
00:34:40RISC-V and performance aren't yet competitive with our Intel and AMD-powered framework main boards.
00:34:45This is just to play with.
00:34:47EMMC storage.
00:34:49The nerds in our audience have been very excited at RISC-V for a very long time.
00:34:55ARM is a dominant quasi-monopolist now.
00:35:00The 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.
00:35:10If you want performance for a while, you end up on this architecture, which means companies like Qualcomm are going to dominate the market right next to Apple.
00:35:18You 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.
00:35:28Qualcomm is saying the Nuvia chips fall under its architecture license, and ARM says no.
00:35:32There'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.
00:35:42Not quite as fast as a Raspberry Pi 4, but it's the thing that might break it.
00:35:47Was it NVIDIA that tried to buy ARM a couple years ago?
00:35:51Yeah.
00:35:52It was, wasn't it?
00:35:54The finances of ARM are hilarious.
00:35:57It's a weird English company that started licensing these designs.
00:36:02SoftBank bought it, right?
00:36:04It was part of the Vision Fund.
00:36:06In the heady days of the SoftBank Vision Fund, where Masa-san was like, everything I touch will be a monopoly.
00:36:13It didn't work out.
00:36:15You might recall WeWork blew that entire business up.
00:36:17They started divesting of assets.
00:36:20They were going to sell ARM.
00:36:22NVIDIA wanted to buy it.
00:36:24Basically, everyone yelled.
00:36:26Everyone yelled.
00:36:28Apple yelled.
00:36:30All the architecture licenses were like, you cannot let our competitor buy this company.
00:36:32They said no.
00:36:34ARM is now public instead, which is a good outcome.
00:36:36Fun fact, ARM going public made up for all the losses the Vision Fund has had for the last 10 years.
00:36:41That's crazy.
00:36:42That's crazy.
00:36:44Somewhere, somewhere the WeWork people are like, god damn it.
00:36:49They're just launching a new product.
00:36:51Don't worry about it.
00:36:53Adam Neumann is like, I could have changed the world.
00:36:55It's a lifestyle.
00:36:57I watched one episode of WeCrash and I was like, I can't live this again.
00:37:01Oh, it's so bad.
00:37:03I cannot be a part of this again.
00:37:05All right, let's wrap up by talking about AR and VR.
00:37:07Keith, you've got a scoop.
00:37:09Meta just reorganized.
00:37:10Let's talk about that.
00:37:12The Vision Pro team is trying to make a cheaper headset.
00:37:14There's a lot of action here.
00:37:16What's going on?
00:37:18Yeah, maybe we'll start with Vision Pro.
00:37:20The information had a good story out confirming and putting more details on what Gurman and others have reported.
00:37:24That Apple is indeed working on a cheaper Vision Pro.
00:37:27They hope to release it by the end of next year.
00:37:30And they have shelved a future iteration of V2 of the Vision Pro as it is today.
00:37:37So another high-end, very expensive one.
00:37:38They're trying to get this cheaper one more into the ballpark of a premium, top-of-the-line iPhone price.
00:37:45So we're talking $1,200 to $1,500.
00:37:47I had heard even before the Vision Pro came out that Apple was already working on this cheaper version.
00:37:53Apparently, when Mark Zuckerberg was talking to me and others and basically saying,
00:37:58I'm going to scorch the earth with cheap headsets forever and not really care about making a business,
00:38:04which we will get to in the next story,
00:38:05Apple was like, oh, okay, we need to really work on a cheap one and get it out.
00:38:12So they've been working on it for already a couple of years.
00:38:15I do expect we'll see it next year, whether it ships next year, TBD.
00:38:20And I think, you know, I've only used the Vision Pro once in that initial demo that we all had at WWDC last year, Nilay.
00:38:27But I would think that, you know, if it was in the ballpark of $1,500,
00:38:32I would maybe pull the trigger even if the OS and the software hadn't meaningfully gotten better
00:38:39just because of, you know, the immersiveness of it.
00:38:41At $3,000, it's just like it's out of touch for most people.
00:38:46And it's still out of touch at $1,500, but it's much more of an almost impulse buy for some people at $1,500.
00:38:53So, yeah, and it will get hopefully more developers interested.
00:38:58But, yeah, Apple's plugging away on.
00:39:00And I think what the other part of this, the fact that they are canceling the more expensive one or shelving it
00:39:08just goes to show that they packed so much tech into this thing that it may not actually be a good product.
00:39:16And we kind of like have found that out already.
00:39:20Nilay, you talk about how you don't use the Vision Pro anymore.
00:39:23It's a great demo.
00:39:24Honestly, it was probably top three demos I've ever done of technology.
00:39:29And it's something that I consistently hear people do not use after the first couple of weeks.
00:39:34Would you say this was a Homer car?
00:39:36No, I mean, it's not like that ridiculous, right?
00:39:39Like it had some focus.
00:39:41They wanted it to be something.
00:39:43I think about the fact that it was a great demo.
00:39:45So much of technology right now is like great demos that fall apart as a product, right?
00:39:49And that was the Vision Pro to a T.
00:39:50And every AI demo so far has been that.
00:39:54Unless you have some very narrow use case where AI can solve a problem consistently for you.
00:39:59Every other demo is like, what?
00:40:01I don't know if I would do that.
00:40:03The Vision Pro, they have to make a big decision.
00:40:05And here is the thing that I noticed at WWDC that I've been thinking about.
00:40:09In that opening keynote video when they were like jumping off the plane.
00:40:13Yeah.
00:40:15And they were like doing the introduction.
00:40:17And everyone was wearing their flight suits.
00:40:18Do you remember this?
00:40:20Vividly.
00:40:22Ridiculous bombastic opening.
00:40:24They had Mike Rockwell on the plane.
00:40:26He's in charge of the Vision Pro.
00:40:28And he turns to the camera with the eyes on.
00:40:31And they played it for laughs.
00:40:33The eyes were a joke.
00:40:35Those eyes aren't shipping anymore.
00:40:38I can tell you that.
00:40:40This is what I'm saying.
00:40:42If you remember the initial Vision Pro stuff.
00:40:45It was like the eyes will place you in the environment.
00:40:48They were so serious about the eyes.
00:40:51So serious.
00:40:53Dead serious about the eyes.
00:40:55I asked in one of our briefings.
00:40:57The only question I asked anybody during our Vision Pro briefings.
00:41:00Have two of you ever looked at each other with the eyes?
00:41:04Have two people ever existed in the same room wearing a Vision Pro?
00:41:08And when you looked at each other with the eyes, did you laugh?
00:41:10And they were like, that's not funny.
00:41:12Because they were dead serious that this thing would bring you into the environment.
00:41:15They meant it.
00:41:16They were dead sincere in that Apple New Age sincerity way.
00:41:21When you go to Apple Park and you walk towards the theater.
00:41:23They play New Age music at you.
00:41:25And they're dead sincere about it.
00:41:27And I'm like, this makes me feel like I'm in a cult.
00:41:29There's a gap there between how sincere Apple can be and the world.
00:41:33And they were super sincere that these eyes would be the thing.
00:41:37That enabled you to wear a headset all the time.
00:41:39No, let's be real.
00:41:41They did the eyes because they wanted to be able to not call it a VR headset.
00:41:44Right.
00:41:46This next version, this cheaper version.
00:41:48They have to decide if this is a VR headset or an AR headset.
00:41:52And playing the eyes for laughs was the thing that signaled to me.
00:41:56Okay, this is going to be a VR headset.
00:41:58I will bet strongly that there will not be the eyes in any future Vision Pro product.
00:42:03And I'll tell you why.
00:42:05It's just a bunch of cost for everything, right?
00:42:07No one's taking you on that bet.
00:42:09We like our money.
00:42:11I'll take the bet.
00:42:13The eyes stay.
00:42:15It was Johnny Ive who insisted that the eyes ship.
00:42:18I don't know the exact dollar amount,
00:42:20but it added an absurd amount of cost and manufacturing complexity to the Vision Pro.
00:42:26Because they had to do that curved glass display.
00:42:29Or that curved glass in the display on the inside, right?
00:42:32It was having the front-facing OLED.
00:42:34I think it's an OLED, right?
00:42:36With the eyes.
00:42:38It's a very low-rise.
00:42:40It added an absurd amount of cost.
00:42:42No one wanted to ship it except Johnny and his people.
00:42:44They've all left.
00:42:46And they're making a joke about it in keynote videos.
00:42:50I don't think the eyes are long for this world.
00:42:53If you get rid of that and you're like,
00:42:55now you cannot be perceived by other people in the world,
00:42:58the thing is a VR headset.
00:43:00And that, to me, is concession in one way.
00:43:05They talk about this thing as AR, AR, AR, AR.
00:43:08But if you can't wear it with other people, it's a VR headset.
00:43:11And then everything they've announced since is immersive.
00:43:15New immersive videos, new experiences, new immersive games.
00:43:19And it's like, oh, this is becoming a VR headset.
00:43:21And at a $1,500 VR headset right now,
00:43:24the state-of-the-art there is a pretty good product.
00:43:27You can make something great for $1,500 in VR.
00:43:30Yeah. And I guess that maybe brings us to Meta.
00:43:33Should we go to that?
00:43:35Meta, obviously, was first to this.
00:43:37Bought Oculus a long time ago.
00:43:38They restructured Reality Labs,
00:43:41the division that does all of their hardware and Horizon,
00:43:45that cursed platform.
00:43:48This week, I had the scoop in command line.
00:43:51They are forming two new groups.
00:43:53So Reality Labs is now Metaverse and Wearables.
00:43:56So you will notice there is no VR in those two phrases, right?
00:44:00The Rift, I'm sorry, God, look how the Rift, wow.
00:44:04The Quest is now under Metaverse,
00:44:06reporting to Vishal Shah,
00:44:08who used to run ads at Instagram and is a software guy.
00:44:11So you basically have the thing that created this whole org
00:44:15that the company really bet its future on
00:44:18is a subdivision of a software org.
00:44:20Meanwhile, Wearables, I'm told,
00:44:23is leaning heavily into the Ray-Bans.
00:44:26I published this internal memo from Boz, the CTO,
00:44:29basically saying that the Ray-Bans have been a much bigger hit
00:44:33than any of us anticipated.
00:44:34Zuckerberg and others have been kind of alluding to this
00:44:37throughout the year.
00:44:39I've heard that it's somewhere in the ballpark
00:44:41of a million-ish units that they've sold,
00:44:43which is 3 to 4X, the first version of the Ray-Bans.
00:44:46And they're seeing really good retention on the product.
00:44:49So people are actually using it consistently.
00:44:51I have them.
00:44:53I had the AI turned on in them.
00:44:55I wrote in command line a few weeks ago.
00:44:58I'm not trying to pat myself on the back, but I will.
00:45:00Just saying that Meta is quietly winning the AI
00:45:04Wearable race here.
00:45:06Because the Ray-Bans are something that Meta has not done
00:45:10in a while, you could argue, maybe ever,
00:45:13but shipped a really good consumer product.
00:45:17Yeah, you're 100% right.
00:45:19It's something that normal people like,
00:45:21something that tech people like, because it's got the AI.
00:45:23And yeah, the AI still hallucinates,
00:45:25but it's very fast and it's right maybe half the time.
00:45:29So I point it at a tree in my backyard and I say,
00:45:31what is this tree and how do I water it?
00:45:33It got it right.
00:45:35So that's not, it's still a demo, right?
00:45:37It's not going to change the world.
00:45:39But in a couple of years, you can see where that's going.
00:45:41And it's like, oh, if I have reliable visual AI in Ray-Bans,
00:45:46I may use my phone less,
00:45:48which is the reason Meta is doing all this, by the way.
00:45:51So they are putting way more resources behind the Ray-Ban line.
00:45:55And they also have these full-fledged AR glasses
00:45:59that they've been working on since 2018.
00:46:01Codenamed Orion, which are insanely expensive.
00:46:05They cost thousands of dollars to make.
00:46:07I have heard from people inside Meta
00:46:09that it's a, again, incredible demo.
00:46:12They really feel like they've gotten there on the hardware.
00:46:16They're not sure what the software use cases will be yet.
00:46:19They're doing a lot of reviews of it right now.
00:46:21I expect we'll get a tease of that later this year
00:46:24at Connect, their conference.
00:46:26So they have these two lines that they're doing with glasses.
00:46:28They've got the AR glasses,
00:46:30which I leaked the whole roadmap for Meta's hardware
00:46:33a year or so ago.
00:46:35So we know that the Ray-Bans are going to be updated next year
00:46:39with a little heads-up display.
00:46:41So think more Google Glass, less full AR Magic Leap.
00:46:45And they're going to come with a wristband
00:46:48that uses neural interface technology
00:46:51from the startup MetaBot called Control Labs
00:46:54to control the display and to control some of the other inputs.
00:46:56Because you can also like play podcasts through the,
00:46:59they have speakers in the frames, right?
00:47:02That's shipping next year.
00:47:04That's going to be interesting.
00:47:06And then the AR glasses,
00:47:08the full-fledged ones that are going to be more expensive
00:47:11and also have the band will be 2027.
00:47:13So this is the path Meta is on.
00:47:15They're going full on glasses and really deprioritizing.
00:47:20They hate when I say this, but it feels like that, the quest,
00:47:23because the quest is still really struggling.
00:47:24They're not really,
00:47:26they have some early product market fit with gaming,
00:47:28but the retention is still not very good, I've heard.
00:47:31And the quest three,
00:47:33they got a lot of really good things
00:47:35about the mixed reality, right?
00:47:37But it's just, it's just not,
00:47:39I don't know why I want to put it on.
00:47:41Whereas the Ray-Bans, it's like, I get it.
00:47:43I have audio on all the time.
00:47:45I can take calls, I can take photos and videos.
00:47:47And I have AI that is starting to work more and more.
00:47:49And by the way, they look like normal glasses, right?
00:47:51So it's just obvious that this is the direction they should go.
00:47:55They seem to now be realizing that
00:47:57and they're putting all their chips there
00:47:59and laying off some people as a result.
00:48:01So that was the news out of Meta this week.
00:48:03So what's really interesting on the quest three
00:48:05is the rate of innovation on the quest three
00:48:08or improvement, I guess, in the quest three skyrocketed
00:48:11after the Vision Pro came out.
00:48:13They're like, oh, we should ship a bunch of these features.
00:48:15Like, I guess there's some competition.
00:48:17Like here's, here's travel mode.
00:48:19We took it away. It's back.
00:48:21We increased the fidelity of the pass-through, right?
00:48:23Like just stuff that they should have done.
00:48:25But they're still not, it's not winning.
00:48:27And even Apple, if they come out
00:48:29with a $1,500 version of the Vision Pro,
00:48:31they still have to make the case for VR.
00:48:33Right, which Meta has been trying for a long time.
00:48:36And now if Meta is saying, all right, it's these glasses.
00:48:39It's the thing that everyone always knew was the thing, right?
00:48:42It's augmented reality glasses
00:48:44where actual light passes through the display
00:48:46and you're looking at actual things instead of screens.
00:48:48And then we can put some information near or around
00:48:51or over those real things.
00:48:53Turns out that's the thing everybody's wanted the whole time.
00:48:57Why would Apple continue to boondoggle its way
00:49:00towards the Vision Pro even at $1,500?
00:49:02Well, because they're on this path, right?
00:49:05Apple was working on AR glasses like Orion at Meta
00:49:09and they basically shelved it semi-recently
00:49:13in the last year or two for the headset route.
00:49:15They think that if they can keep iterating on the headset
00:49:18they'll figure out glasses eventually.
00:49:20Whereas Meta is like, no,
00:49:22we are going to have glasses out in a few years.
00:49:24And Apple's fine.
00:49:26They're always late to this.
00:49:28Like we all know that.
00:49:30So they'll come in late like they usually do.
00:49:32I don't know if it'll be as late as they were with the Vision Pro,
00:49:34but they're going to be late and that's fine.
00:49:36But yeah, I think they're betting on productivity
00:49:39and being at home and coding
00:49:41and all these things with the Vision Pro.
00:49:43And Meta's like, no, we want you to lose.
00:49:45Meta's goal is to use your phone less
00:49:46because they hate being under Apple and Google's thumb.
00:49:49Whereas Apple is like, let's just extend what we do.
00:49:52And I think there's the big strategic difference.
00:49:55I will say that I have a friend from the Brooklyn years
00:49:59who in the pandemic became one of those people
00:50:01who doesn't live anywhere.
00:50:03She just lives on beaches.
00:50:05I know one of those folks, yeah.
00:50:07She's great.
00:50:09Does she work at NVIDIA or what?
00:50:11She does not work at NVIDIA.
00:50:13And she was the first normal person adopter
00:50:16of the Meta glasses with the Ray-Bans.
00:50:18All of her Instagram pictures are a beach
00:50:21taken with the Meta Ray-Bans now.
00:50:23And you can just see, oh, this thing worked.
00:50:25The thing worked.
00:50:27It looks like she's wearing Wayfarers in all the pictures,
00:50:29but it's the Meta glasses.
00:50:31And it worked.
00:50:33It's like so much.
00:50:35There's no chance that any of these people
00:50:37are wearing Vision Pros
00:50:39as they nomadically travel the world
00:50:40or to Quest 3,
00:50:42but there's 100% chance that, hey,
00:50:44your camera is now in your sunglasses
00:50:46and you're looking at cool stuff you want to share on Instagram
00:50:48is a loop that Meta can close.
00:50:50I was at a very fancy tech conference
00:50:52where I'm not even allowed to say the name earlier this week,
00:50:54but it was a lot of tech CEOs.
00:50:56And there were several of them wearing the Ray-Bans,
00:50:58like walking around.
00:51:00CEOs of companies that it would be very funny
00:51:02for me to say who they were wearing the Ray-Bans.
00:51:04And you're not supposed to record anything here,
00:51:06but they're wearing them.
00:51:08And it's like, oh, yeah,
00:51:10it's catching on.
00:51:12It's catching on with tech people,
00:51:14powerful people,
00:51:16and it's catching on with normal people,
00:51:18which is like that is a circle
00:51:20that Meta has not been able to draw
00:51:22in its entirety of existence as a hardware player.
00:51:24Yeah.
00:51:26How good did Jensen look in the glasses?
00:51:28I can't comment.
00:51:30I'm telling you,
00:51:32I think VR is interesting.
00:51:34There's stuff in VR,
00:51:36like Supernatural,
00:51:38they keep promoting as a hit.
00:51:40They claim that waving bats in VR
00:51:42is as strenuous as running.
00:51:44It feels like it.
00:51:46Sure.
00:51:47Whatever, man.
00:51:49To me, yes.
00:51:51If you can claim that fancy Peloton Beat Saber
00:51:53is the same as a swift run,
00:51:55more power to you.
00:51:57But that's the only thing they have that's sticky, right?
00:51:59And Apple doesn't have anything that's sticky yet,
00:52:01except for the monitors.
00:52:03We'll see.
00:52:05I just think Apple moving away,
00:52:07the compromise of the Vision Pro
00:52:08is they built a VR headset
00:52:10with all this passthrough and the eyes and stuff
00:52:12to make it closer to AR.
00:52:14And if what they're going to do is drop the eyes
00:52:16and go head more towards VR,
00:52:18that feels like a turn that is notable
00:52:20because the real platform,
00:52:22you want to talk about the real platform shift,
00:52:24has always been VR.
00:52:26Everybody knows this.
00:52:28Tim Cook has talked about that
00:52:30as being the big platform shift.
00:52:32And to go to,
00:52:34we just made the best VR headset,
00:52:36would be very different.
00:52:38I don't have tears here to hold me in.
00:52:40That's what's going on here.
00:52:42We got to take a break.
00:52:44We'll be right back.
00:52:46More on the VR Chats after this.
00:52:48All right, we're back.
00:52:50This time we got to talk about AI.
00:52:53This will just take two minutes.
00:52:55There's nothing going on.
00:52:57Actually, quite a lot going on AI.
00:52:59And a lot of it is just,
00:53:01as ever,
00:53:03confounding decisions by AI companies.
00:53:05Let's start with
00:53:06a confounding branding decision.
00:53:08There's a new company.
00:53:10It's called Safe Superintelligence.
00:53:12SSI.
00:53:14Good name.
00:53:16This is Ilias Hutzever,
00:53:18who, if you recall,
00:53:20was the CTO of OpenAI
00:53:22who participated in the coup against Sam Altman
00:53:24and then changed his mind.
00:53:26There's some safe superintelligence right there.
00:53:28Whether or not you have belief in that decision.
00:53:30Then they couldn't figure out
00:53:32how to bring him back to the company
00:53:34and now he's left to start doing Alex.
00:53:36Not just the CTO,
00:53:38the co-founder,
00:53:40and really the one who is driving
00:53:42the research agenda at OpenAI.
00:53:44It was his decision to lean into the transformer,
00:53:46the T in chat GPT.
00:53:48I think I've heard
00:53:50one of the largest individual shareholders,
00:53:52incredibly influential.
00:53:54He's someone that AI researchers look to
00:53:56as a demigod.
00:53:58And he has been totally quiet.
00:54:00There's been this meme of,
00:54:02where's Ilya for the last six to eight months,
00:54:04really since the November coup attempt.
00:54:06With a company that they will not
00:54:08describe anything about,
00:54:10except that they're going to invent
00:54:12super safe AGI,
00:54:14whatever the hell that means.
00:54:16His co-founder is Daniel Gross,
00:54:18who used to run AI at Apple,
00:54:20which is interesting and is now I would consider
00:54:22one of, if not the most plugged in
00:54:24venture capitalists in AI
00:54:26who's been raising a ton of money
00:54:28and deploying it into startups.
00:54:30And they are putting the call
00:54:32out for people to join.
00:54:34And I'm sure we'll be rating
00:54:36AGI in very due time.
00:54:38And the key thing here
00:54:40is Ilya saying,
00:54:42we are going to be free from,
00:54:44I forget the exact wording,
00:54:46but we're going to be free from the
00:54:48products and management day to day
00:54:50and only focus on long-term
00:54:52aligned AGI research.
00:54:54And we will not release anything
00:54:56until we hit AGI.
00:54:58So it's, this is a,
00:55:00this is a pure research lab,
00:55:02which is what Ilya has been known for
00:55:04his entire career.
00:55:05I'm going to read this paragraph to you.
00:55:07Yeah, read it.
00:55:09Because anybody who has ever worked anywhere
00:55:11is, well, I think have the same reaction
00:55:13to this paragraph.
00:55:15Our singular focus means
00:55:17no distraction by management overhead
00:55:19or product cycles.
00:55:21And our business model means
00:55:23safety, security, and progress
00:55:25are all insulated from short-term
00:55:27commercial pressures.
00:55:29I just like, what?
00:55:31Well, the business model
00:55:33is we just raised
00:55:35the amount of money.
00:55:36But that means you don't have
00:55:38a business model.
00:55:39Right.
00:55:40Yeah.
00:55:41Like that's how you get to do these
00:55:43things is like you don't,
00:55:45you don't have the,
00:55:46you don't like, you don't have the
00:55:47pressure of a business.
00:55:48Yes.
00:55:49Isn't that just all a really pointed
00:55:51jab at Sam Holtman?
00:55:53But, but yes.
00:55:55Right.
00:55:56Like I figured it was more that than
00:55:57actually, like in this case,
00:55:58I was assuming that he was going
00:56:00less like I'm here to make money
00:56:02because he's made it very clear he
00:56:03doesn't care about money and more
00:56:04like I'm here to like say I
00:56:06don't like what Sam Holtman did.
00:56:08And so I want to make a company
00:56:09where we're going to do cool stuff.
00:56:11All I'm saying is I watched Max and
00:56:12two of her six-year-old friends from
00:56:14kindergarten try to play lemonade
00:56:16stand and they are distracted by
00:56:18management overhead and product
00:56:19cycles.
00:56:20And their business model means they
00:56:22often compromise on safety,
00:56:24security, and progress.
00:56:25Like there's just like no way
00:56:28you can put any group of people
00:56:30together and not have management
00:56:31overhead.
00:56:32It is not possible, especially if
00:56:34you're trying to hire
00:56:36the most highly paid
00:56:38sought after talent that exists
00:56:40in technology today.
00:56:41And that's what it's like.
00:56:42I don't I read this paragraph.
00:56:44I'm like, this doesn't you can't do
00:56:45this.
00:56:47The VC math here is simple.
00:56:49We will give you whatever you want.
00:56:51And when you invent AGI, it will
00:56:52change everything.
00:56:53It will be the new fire and
00:56:55we'll figure it out.
00:56:56That's I guarantee you that's the
00:56:58extent of the diligence.
00:57:00I was going to ask, like, where is
00:57:01the money coming from to recruit all
00:57:03of these very highly paid?
00:57:05Well, they won't say where they've
00:57:06raised money from.
00:57:07So if you know, let me know.
00:57:09But if you're listening to this, I
00:57:10want to find out.
00:57:11Daniel is there.
00:57:12They're all super wealthy
00:57:14and they will have no trouble raising
00:57:16whatever they want, because when you
00:57:17co-founded OpenAI,
00:57:19you can raise whatever you want.
00:57:20And then in particular, participated
00:57:23in the notable management overhead
00:57:25of doing a huge coup against
00:57:27the CEO.
00:57:28Yes.
00:57:29A failed coup.
00:57:30Yeah.
00:57:31It's like your track record of not
00:57:32having management overhead is very
00:57:34bad.
00:57:35Also, like a question I have is what
00:57:37is safe AGI?
00:57:38I keep hearing this intelligence.
00:57:40It's not AGI or super intelligence,
00:57:42which is actually now the new word
00:57:43for what comes after AGI, apparently
00:57:45is the thing I keep hearing.
00:57:47And so we keep inventing new terms
00:57:49because they realize, like the terms
00:57:50they had, no one understands or they
00:57:52don't actually even understand what
00:57:53they are.
00:57:54What is safe?
00:57:55What does this mean?
00:57:56Like Ilya's whole thing at OpenAI was
00:57:58alignment, right?
00:57:59He set up this alignment group that
00:58:00was going to get 20 percent of
00:58:02OpenAI's compute budget.
00:58:03Apparently, they never did.
00:58:05That led to a lot of infighting.
00:58:06And I think, you know, helped
00:58:08him try to kick Sam out.
00:58:11But I always hear this and I'm like,
00:58:13aligning for who and
00:58:15under what values are we aligning?
00:58:17And they don't seem to have an
00:58:18answer.
00:58:19And that's been like a consistent
00:58:20problem with AI, right, is
00:58:22that they're always like, yeah, this
00:58:23is going to be for everyone.
00:58:25But we only trained it with these
00:58:27certain data sets.
00:58:28We only trained it using like the
00:58:30pre-established knowledge of the
00:58:32certain people creating it.
00:58:33So it's actually fairly insular and
00:58:35doesn't work for everyone.
00:58:37And there are enormous edge cases
00:58:39every time we deploy it.
00:58:40And, you know, as Neela and I
00:58:42uncovered when we were reporting on
00:58:43the coup attempt, like this gets
00:58:45kind of portrayed in the media as
00:58:47this, you know, acceleration versus
00:58:49deceleration, almost like religious
00:58:52faction war that's happening.
00:58:54And I do think there's part of that.
00:58:55I do think Ilya is almost like
00:58:57religious and is like fervor here.
00:58:59But a lot of this is just like
00:59:01boring company stuff, business
00:59:03stuff.
00:59:04It's like you wanted budget.
00:59:06You didn't get it.
00:59:07You got layered.
00:59:08Therefore, you don't want to work
00:59:09there anymore.
00:59:10Open AI just hired a CPO.
00:59:12He used to run product at
00:59:13Instagram and a CFO who was the
00:59:15CEO of Nextdoor, a public company.
00:59:17It's very clear where open AI is
00:59:19going. They're going to IPO.
00:59:21They want to build a massive
00:59:22business. They're going to get rid
00:59:23of being a nonprofit.
00:59:24It's very clear.
00:59:25I mean, this week made some
00:59:26noises about not being a
00:59:27nonprofit.
00:59:28Right. And so if that's the
00:59:29direction they're going, you're
00:59:31Ilya. It's like you just want to
00:59:32do research.
00:59:33Right. And that's where a lot of
00:59:34these researchers still are is
00:59:35they just want to do cutting edge
00:59:37frontier model development.
00:59:39And this is him saying, come
00:59:41here, we'll do it.
00:59:42And we should also note this
00:59:44is now the third offshoot of
00:59:46open AI that was has been
00:59:48created to create safer AI
00:59:51from open AI whose mission is
00:59:52also to create safe AI.
00:59:54It's not a great look.
00:59:56Yeah. So you have a topic.
00:59:58I would I would actually put XAI
00:59:59in this camp.
01:00:00Elon was the co-founder of Open
01:00:02AI.
01:00:03Now you have SSI, which is also
01:00:05I feel like it's a medical term or
01:00:07something.
01:00:09I got this once.
01:00:10It was it was you take a pill for
01:00:12it. It's fine.
01:00:14It's a neo-zempic SSI.
01:00:16Yeah.
01:00:18Look, I just it's great that
01:00:20everybody thinks they can
01:00:21accomplish these things.
01:00:22I actually think it's
01:00:23unclear at all
01:00:25whether transformers and
01:00:27LLMs can ever
01:00:29do the things that people want
01:00:31them to do.
01:00:32And in fact, over time, it has been
01:00:34more and more clear that perhaps
01:00:35they cannot.
01:00:36And so what I'm dying to know is
01:00:38if SSI or any of the other
01:00:40companies that think they can build
01:00:41AGI, if they have another idea.
01:00:44Because if the only idea is we're
01:00:45going to make a slightly better LLM,
01:00:48I actually don't.
01:00:49I'm not you're not going to make
01:00:50you're not worried about whether
01:00:51it will be safe because it's not
01:00:52going to be super intelligent.
01:00:53I'm just.
01:00:54Why are people trying to build an
01:00:55AGI?
01:00:56Like these people are nerds.
01:00:58They they watch.
01:00:59That's why you just answered your
01:01:00own question.
01:01:01No, but they watch the movies.
01:01:02They play the games like this man
01:01:04is setting out to be the
01:01:06the audio clips you hear when you're
01:01:08walking through the ruins a thousand
01:01:10years later in a video game of
01:01:12the guy being like, I don't know
01:01:13what's happening. Why is my A.I.
01:01:14going and killing the world?
01:01:16Like that's what he's setting
01:01:17himself up for.
01:01:18If he was actually successful
01:01:20in his grand ambitions.
01:01:22And I'm just like.
01:01:23Why?
01:01:24Because the industry thinks that
01:01:25we they have invented a new form
01:01:27of intelligence and
01:01:29that hasn't happened before.
01:01:31And so it's I think the natural
01:01:32progression of this is that it's
01:01:33all smarter than us and
01:01:35replaces us.
01:01:36And it seems inevitable
01:01:38to the people building this stuff.
01:01:39What we see is like
01:01:41Chachapiti can't get a math
01:01:42equation right.
01:01:43Right.
01:01:44That's like what we see what
01:01:45they're saying.
01:01:46We've got to keep in mind like the
01:01:47gate, this stuff like GPT four
01:01:49was in existence for like over a
01:01:51year before open A.I.
01:01:52actually deployed it in the
01:01:53world.
01:01:54So they're seeing things we're
01:01:55not. I'm not going to like give
01:01:56them too much of the credit
01:01:57there.
01:01:58But I do think there's
01:02:00you know, Yann LeCun, who's the
01:02:01chief A.I.
01:02:02researcher at Meta, has been
01:02:03saying what you just said,
01:02:05Eli, that LLMs are not the path
01:02:07to AGI.
01:02:08So his group is actually working
01:02:09on what comes after the
01:02:10transformer.
01:02:11What is the architecture we need?
01:02:13Because these are parrots.
01:02:15Right.
01:02:16These LLMs are our fancy
01:02:17prediction engines.
01:02:18Right.
01:02:19So you don't really unlock
01:02:20intelligence that way.
01:02:21Yeah.
01:02:22No, I mean, yesterday I
01:02:23assigned our futures team to
01:02:24write languages, not
01:02:25intelligence.
01:02:26Like, I feel like we just need
01:02:28to say some very explicit
01:02:30declarative sentences that are
01:02:31true.
01:02:32Like dogs are very intelligent
01:02:34and cannot generate a
01:02:35PowerPoint.
01:02:36Like, you know, I mean, like
01:02:37we're getting confused about
01:02:39what intelligence is at a very
01:02:40high level.
01:02:42I'm I've been lately thinking a
01:02:44lot about that.
01:02:45The Steve Jobs slide of
01:02:46technology and liberal arts
01:02:48and it's like, oh, you guys
01:02:49forgot about this whole
01:02:50street.
01:02:51This whole street's missing
01:02:52from this conversation right
01:02:53now.
01:02:55And there's just a piece of
01:02:56this puzzle where you can bet
01:02:58on this stuff to get better at
01:02:59an infinite rate.
01:03:00But like most curves level
01:03:02off and it is unclear to me
01:03:04what actually makes things
01:03:05intelligent.
01:03:06That said, well, you know,
01:03:07good luck.
01:03:08Good luck building your
01:03:09company with no management
01:03:10overhead.
01:03:11Come on, decoder.
01:03:12If you pull that off.
01:03:13Boy, do I have a podcast for
01:03:14you.
01:03:15Right.
01:03:16I spent an hour talking about
01:03:17that every week with people.
01:03:19But like it turns out it
01:03:20always looks like three six
01:03:22year old girls have
01:03:23management overhead.
01:03:24I promise you maybe more than
01:03:25any company in America.
01:03:30Speaking of confounding
01:03:31decisions by AI companies.
01:03:34Perplexity started out so hot,
01:03:36man.
01:03:37Mm hmm.
01:03:38They were the ones.
01:03:39Right.
01:03:40You can pick any model you can
01:03:41ask these questions.
01:03:42It was better searching Google
01:03:43and it turns out, boy, have
01:03:44they been awful shady along
01:03:45the way.
01:03:46What's going on here?
01:03:47Well, yeah.
01:03:48So perplexity, for those who
01:03:49don't know, is trying to
01:03:51unseat Google with an AI
01:03:53search engine.
01:03:54It's actually a good product
01:03:56most of the time.
01:03:57I would say that I like their
01:03:58AI answers more than Google's
01:04:01on average.
01:04:02And now we kind of know why
01:04:03it's because they're stealing
01:04:04a bunch of content.
01:04:06And yeah, so Wired did some
01:04:09good reporting and found out
01:04:11that they were going around
01:04:13their crawler blocker.
01:04:15I've heard rumors that
01:04:16perplexity has been doing this
01:04:18at an even grander scale with
01:04:20even much larger corpuses of
01:04:22data.
01:04:24And this is the open AI
01:04:26playbook on steroids a little
01:04:28bit, which is just run fast and
01:04:30beg for forgiveness.
01:04:31And that's what they're doing.
01:04:34They're basically going around
01:04:35people who have said, no, you
01:04:37are not allowed to summarize
01:04:38this and doing it anyway.
01:04:40And the way they're doing it
01:04:41is both shady and hilarious.
01:04:44So they publish some IP ranges
01:04:47of their crawlers, and they
01:04:48publish the name of their
01:04:49crawler.
01:04:50So poorrobots.txt, a text file
01:04:54that should not bear the weight
01:04:56of all intellectual property law
01:04:57on the internet, but which
01:04:58currently does.
01:04:59You can set it to disallow
01:05:00these crawlers.
01:05:01You can set it to disallow
01:05:02GPT bot, which is open AIs.
01:05:05There's Googlebot Plus, I think.
01:05:06So there's Googlebot, which is
01:05:07the regular search index, and
01:05:09there's the other one, which is
01:05:10Google's AI crawler.
01:05:12I feel like I need to disclose
01:05:13here.
01:05:14We used to disclose GPT bot at
01:05:15Vox Media, and then our company
01:05:17signed a deal, and now we don't.
01:05:20There you go.
01:05:21There's your disclosure.
01:05:23And you could disallow
01:05:24perplexity.
01:05:25And on top of that, they did
01:05:26this thing where, like, here
01:05:27are our IP ranges.
01:05:28Like, if you don't trust
01:05:29robots, you can disallow these
01:05:30IPs.
01:05:31And then what Wired found, and
01:05:32then the folks at MacStories,
01:05:33which is a great indie Apple
01:05:34site, they found that their
01:05:37content was showing up anyway
01:05:39in perplexity, and then the
01:05:41investigations basically
01:05:42revealed these IP ranges were
01:05:43fake, and the crawler was fake,
01:05:45and they were setting up
01:05:46headless Windows 10 machines to
01:05:49browse the web.
01:05:50Yes.
01:05:51Like, you know those pictures of
01:05:52like the streaming fraud farms
01:05:54where hot dogs are just swiping
01:05:56Android phones?
01:05:58There's your VC money.
01:05:59I love it.
01:06:00Just swiping away on a Windows
01:06:0210 browser.
01:06:03Just imagine if those Windows
01:06:04machines had recall.
01:06:08It's just, like, crazy.
01:06:11And the CEO of Perplexity has
01:06:12talked a lot of shit about
01:06:13Google.
01:06:14Like, why are these results so
01:06:15bad?
01:06:16And, like, Google, for better or
01:06:18worse, is trying to play by the
01:06:20rules.
01:06:21It's still pissing everybody
01:06:22off, but they're attempting to
01:06:23play by the rules.
01:06:24They have to.
01:06:25I mean, Google is under consent
01:06:27decrees.
01:06:28They're a massively regulated
01:06:29mega-cap tech company, so they
01:06:31can't do this kind of stuff.
01:06:32Yeah, and here's a startup being
01:06:34like, what if we lied?
01:06:36Yeah, what if we just didn't do
01:06:37any of that?
01:06:38And to be clear, I keep making
01:06:39this comparison.
01:06:40You called it the opening
01:06:41iPlayBook.
01:06:42This was the Google PlayBook 20
01:06:43years ago.
01:06:44Yep.
01:06:45Right?
01:06:46Google just allowed all of
01:06:47Viacom's content to be on
01:06:48YouTube.
01:06:49Google indexed every book in the
01:06:50world without permission.
01:06:51Google did image search, which
01:06:52it got sued over.
01:06:53And it won all those lawsuits
01:06:55because those products were new.
01:06:57Mm-hmm.
01:06:58So they would show up in court
01:06:59and be like, judge YouTube.
01:07:00And the judge would be like,
01:07:01YouTube's sick.
01:07:02And Viacom would be like, our
01:07:03marketing staffers are using
01:07:04this anyway because it's sick.
01:07:08Google image search, so useful.
01:07:09It's concocted.
01:07:10Nothing like this has ever
01:07:11existed before.
01:07:12Let's concoct some reasoning to
01:07:13allow Google to take all this
01:07:14stuff for free because the
01:07:16theory is that you'll go buy the
01:07:19image from Getty once you've
01:07:20found it on Google.
01:07:21It's unclear what that theory
01:07:22really came out to in the end.
01:07:23I never did that.
01:07:24Yeah.
01:07:25Alex is not like, I opened a
01:07:27Getty library the next day, a
01:07:29Getty account the next day.
01:07:31But those were the theories
01:07:32back then.
01:07:33Google Books, that might have
01:07:34worked, right?
01:07:35You can search all the books
01:07:36and maybe you'll actually buy
01:07:37the book in the end.
01:07:38Did it work out totally?
01:07:40I don't know.
01:07:41But those were the theories
01:07:42back then because the products
01:07:43were so new, the experiences
01:07:44were so new.
01:07:45Perplexity is just like, yeah,
01:07:46we stole a bunch of stuff and
01:07:48what we've delivered is an
01:07:49incremental improvement on
01:07:50Google search.
01:07:51And I don't think that's going
01:07:52to win this time.
01:07:55Maybe.
01:07:56I am shocked at how little the
01:07:58people inside these AI companies
01:08:00and their investors, again, I
01:08:02was just at a very fancy tech
01:08:03conference with a lot of these
01:08:04people, how little they
01:08:06actually care about this
01:08:07problem, how it is really just
01:08:08like this is a bunch of speed
01:08:10bumps that will go over to
01:08:12getting where we're going to go.
01:08:14And maybe it's because of the
01:08:15fact that Google is Google and
01:08:17they did all this 20 years ago
01:08:18and they're a trillion,
01:08:19multi-trillion dollar company.
01:08:21I think it's proof in history
01:08:23shows that, yeah.
01:08:24I was going to say, I think
01:08:25it's back to what Nealai said
01:08:26about liberal arts education.
01:08:28They just don't have any
01:08:29historical context.
01:08:30They're like, yeah, this is
01:08:31going to work.
01:08:32They didn't take any history
01:08:33classes.
01:08:34I don't need to know about
01:08:35Google.
01:08:36This is fine.
01:08:37No, these are systems people.
01:08:38Yeah.
01:08:39They don't think in terms of
01:08:40the values that we get mad
01:08:42about when it's like, hey, I
01:08:43made a thing.
01:08:45I charged people for it and you
01:08:46took it.
01:08:47That's not OK.
01:08:49With intellectual property,
01:08:51with writing and stuff like
01:08:52that, for some reason, it's
01:08:53just not a thing that a lot of
01:08:55tech people that I know, it
01:08:57doesn't register for them.
01:08:59I don't know why.
01:09:00Right.
01:09:01Because they didn't take
01:09:02liberal arts.
01:09:04They're like, what is an
01:09:05English class?
01:09:06I want to spend more time.
01:09:07I don't have I don't have a
01:09:08view of this.
01:09:09I haven't locked this into my
01:09:10brain yet.
01:09:11So people have thoughts.
01:09:12Send them to me because I need
01:09:13the help.
01:09:14But I can feel this weird
01:09:15shift that I've been calling
01:09:17the copyright apocalypse
01:09:19because some people care about
01:09:20it a lot now, more than ever
01:09:23before.
01:09:24This is mine.
01:09:25And you took it between
01:09:26influencers on social media
01:09:28platforms, rages out of
01:09:30control now.
01:09:32Right.
01:09:33I'm a Taylor Swift fan.
01:09:34And those three notes of a
01:09:35song sound like three notes
01:09:37and Olivia Rodrigo fan.
01:09:38We have to do fandom war now
01:09:40until the music industry
01:09:42doles out writing credits is a
01:09:44real thing that happens at the
01:09:47craziest level.
01:09:48This is a real thing that
01:09:49happens.
01:09:50And that is copyright
01:09:51maximalism by really young
01:09:53people.
01:09:54And then over here, you have
01:09:56nobody owns anything.
01:09:57It's just something that I
01:09:58should get it for free.
01:09:59And those things are just in
01:10:00tension in a way that it
01:10:01feels unresolved and
01:10:03potentially unresolvable.
01:10:05Like you can't, you can't
01:10:07have a bunch of YouTubers
01:10:08being like, I made this
01:10:09YouTube video first.
01:10:10And anybody who even makes a
01:10:11video that looks like it is
01:10:13stealing from me right next
01:10:15to everything in the
01:10:16internet should be free.
01:10:17Maybe that's it.
01:10:18Maybe there's not a lot of
01:10:19love for media companies, as
01:10:20we all know.
01:10:21And the fact that it's
01:10:22scraping wired or the verge
01:10:24or whatever is like,
01:10:25whatever, they're fine.
01:10:26But when it's like your
01:10:27favorite influencer, your
01:10:28favorite creator, and they're
01:10:29on their feed going, hey,
01:10:30like this model, don't use
01:10:32it because they trained on
01:10:33my stuff.
01:10:34I do think you're right.
01:10:35I think people feel a lot
01:10:36more viscerally about that
01:10:38than they do.
01:10:39Oh, like companies, right?
01:10:40I think that's when it
01:10:41becomes the individual.
01:10:42And again, what Google was
01:10:44up against, what Napster was
01:10:45up against was the big bad.
01:10:47Right?
01:10:48Like the only face of the
01:10:49anti-Napster movement was
01:10:50Lars Ulrich from Metallica.
01:10:52And it did not work for him.
01:10:55But really, the face was
01:10:56like Hollywood in the music
01:10:58industry, and no one ever
01:11:00feels any sympathy towards
01:11:01Hollywood in the music
01:11:02industry.
01:11:03Like Hollywood can be like,
01:11:04you wouldn't steal a cigarette
01:11:05or whatever they would say.
01:11:06You wouldn't, what was it?
01:11:07You wouldn't download a-
01:11:08You wouldn't download a car.
01:11:10Right.
01:11:11So Hollywood can put up the
01:11:12banners that say, you
01:11:13wouldn't download a car.
01:11:14And everyone's like, screw
01:11:15you, I'm super going to
01:11:16download a movie.
01:11:17And if I could download a
01:11:18car, I would.
01:11:19That'd be sick.
01:11:20Yeah.
01:11:21It rules.
01:11:22I've had this 3D printer for
01:11:23years not making cars.
01:11:25But they were the big bads.
01:11:27Influencers are like people.
01:11:29Influencers are like people.
01:11:31You know what I mean.
01:11:32Influencers are actual people
01:11:33that you have parasocial
01:11:34relationships with.
01:11:35Yeah.
01:11:36And there's something very
01:11:37emotional there that is not
01:11:38resolved.
01:11:39And it's going to come to
01:11:40your head, because the next
01:11:41story we have on our list
01:11:42here is a bunch of social
01:11:45networks that want to
01:11:46introduce AI in various ways.
01:11:48Alex, you played with one.
01:11:49And before we get to that, I
01:11:50just want to call out TikTok
01:11:52is now letting people make
01:11:55AI avatars of creators.
01:11:57Yeah.
01:11:58Right?
01:11:59And saying this is your ad
01:12:00creative.
01:12:01And this is going to come for
01:12:02every social platform.
01:12:03Yeah, that's right.
01:12:04Like in a massive way.
01:12:05Like if you are a creator and
01:12:06you feel the burnout and you
01:12:07can make 50 more ads during
01:12:09your brand deal using it, you
01:12:11might do it.
01:12:12Yeah.
01:12:13No, this is happening.
01:12:14So, yeah, I wrote about this
01:12:15new app, which was, you know,
01:12:16I get a lot of test flights to
01:12:17try things.
01:12:18And this is the first test
01:12:19flight I've gotten in a while
01:12:20that made me feel something
01:12:21very strongly, where I was
01:12:22like, oh, when I opened it.
01:12:25And it is like Instagram meets
01:12:28character AI, where basically
01:12:31it's a bunch of AIs in a feed
01:12:33commenting and posting together.
01:12:35And there's humans, but they
01:12:36don't need to be there.
01:12:37And like, it's actually in some
01:12:39cases, not most, but in some
01:12:40cases, hard to tell who is
01:12:42human and who is not.
01:12:44And they just raised a bunch of
01:12:45money.
01:12:46It's an ex-Snap people.
01:12:48And the reason I wrote about it
01:12:49is because Meta is about to do
01:12:50the exact same thing.
01:12:52Zuckerberg told me so last
01:12:54September.
01:12:55And I think later this year,
01:12:56we're going to see them start to
01:12:57roll this out, where basically
01:12:58anyone can create their own AIs,
01:13:00send them into Instagram, send
01:13:02them into Facebook.
01:13:04And same thing with businesses.
01:13:05And the idea is that if you're
01:13:07an influencer and you feel
01:13:09overwhelmed by your DMs, maybe
01:13:10an AI handles that for you, or
01:13:12maybe an AI actually posts on
01:13:14your behalf.
01:13:15Zuckerberg is super excited
01:13:16about this.
01:13:17I know because he's talked to me
01:13:18about it.
01:13:19It's very freaky to imagine what
01:13:21this actually does to our social
01:13:23networks.
01:13:24It's coming way faster than we
01:13:25think, though.
01:13:26Well, it's interesting because
01:13:27in South Korea, they've been
01:13:29doing a little bit of something
01:13:30like this, right?
01:13:31With K-pop stars, where you can
01:13:33engage with K-pop stars on sites
01:13:35owned by the K-pop stars
01:13:36companies.
01:13:37And it's not always that.
01:13:39Sometimes it's them.
01:13:40It's not always them.
01:13:41Usually it's a bot.
01:13:43It's a person pretending to be
01:13:44them.
01:13:45And so that parasocial, how do
01:13:47we automate that parasocial
01:13:48stuff, has been happening in
01:13:49Korea for a long time.
01:13:50But I think the difference here
01:13:51is that was gated to, okay, you
01:13:53have to go download this app for
01:13:55this publisher and use it,
01:13:57whereas this would be on
01:13:59Instagram?
01:14:00This would be like you open
01:14:01Instagram and it's like, oh, is
01:14:02that account in AI?
01:14:03They may have free arms.
01:14:05Okay, maybe then.
01:14:06Yeah.
01:14:07Well, that's coming.
01:14:08So then there's the other part,
01:14:09which is you send agents out on
01:14:11your behalf as an influencer or
01:14:12someone where you're like,
01:14:14someone wants to pay money to
01:14:16subscribe.
01:14:17And one of the things I offer is
01:14:18like DMs with me.
01:14:20Yeah.
01:14:21And then my agent will just like
01:14:22do that.
01:14:23Yeah.
01:14:24And I won't do it.
01:14:25And that's infinitely scalable,
01:14:26which if you are an individual
01:14:27creator and all of these
01:14:28platforms are organized towards
01:14:30tearing down collectives and
01:14:32enhancing individual creators,
01:14:34if you're an individual creator,
01:14:35your ability to scale is limited
01:14:37by your emotional capacity to do
01:14:40this work and your minutes in the
01:14:42day.
01:14:43Yeah.
01:14:44And so, of course, you're going
01:14:45to let the robot talk to your
01:14:46fans.
01:14:47It is the biggest no-brainer in
01:14:48the history of the world.
01:14:49So, of course, you're going to
01:14:50say, OK, you want to brand X
01:14:52wants to do an integration with
01:14:53me, but they wanted 50 different
01:14:55versions as creative.
01:14:56I will make one and let the AI
01:14:58deep fake me into five more.
01:15:00And that is already happening.
01:15:02Like TikTok is enabling the
01:15:05feature.
01:15:06It's happening now.
01:15:07Yeah.
01:15:08Well, Alex, to your question
01:15:09about why, like I was talking to
01:15:10Vu Tran, the CEO of Butterfly,
01:15:11this new app about this, the why
01:15:13because I showed this app to my
01:15:15wife and she was like, just like
01:15:16it was an immediate just like
01:15:19I'm like, I think you're going to
01:15:21meet a lot of people like that
01:15:22who are like, why does this need
01:15:23to exist?
01:15:24Like, why do we need to, like,
01:15:25fill our time with just empty
01:15:28social media?
01:15:29And his point was like, yes, it
01:15:30feels kind of broken right now.
01:15:32And I wrote about this.
01:15:33It feels like when the hosts in
01:15:34Westworld break a little bit,
01:15:35it's like not there.
01:15:37But he was like, you know, I got
01:15:39into tech by like making friends
01:15:41on forums, talking to people that
01:15:43I never met.
01:15:44I only knew their username.
01:15:45He's like, we will get the
01:15:46models quickly to that quality.
01:15:48And as someone who also came up
01:15:50and got into tech in a similar
01:15:51way, like through like making
01:15:52friends on forums, like what
01:15:54happens when like I never met
01:15:56those people, right?
01:15:57I don't know a lot of their
01:15:58names ever.
01:15:59So if they were AI, would it
01:16:02have made a difference?
01:16:03I mean, we had a whole like
01:16:04philosophical conversation about
01:16:05this, but he was like, you don't
01:16:06actually want to talk to people.
01:16:08There's traits of being human
01:16:09that you like and AI will
01:16:11quickly copy all those traits.
01:16:13And I'm like, mind blown.
01:16:16And also no, no, thank you.
01:16:18But is it inevitable?
01:16:20That's the question.
01:16:21It ignores like in the 80s and
01:16:2490s, there were the 900 numbers
01:16:26and you could like call like the
01:16:28Crypt Keeper was one like Corey
01:16:30Haim and stuff like that.
01:16:31And Corey Haim was not answering
01:16:33the phone to talk to a little
01:16:3512 year old girl who had a crush
01:16:36on him, right?
01:16:37Like that was a voiceover
01:16:38pretending to be him.
01:16:39And quickly...
01:16:40Wait, Alex, did we just learn
01:16:41something very important about
01:16:42what you spent your money on
01:16:43when you were a child?
01:16:44It was not me.
01:16:45Because if I had called a 900
01:16:47number, I would have not had a
01:16:50good time.
01:16:51My dad super caught me calling
01:16:53a 900 number once and he was
01:16:55like, you're never using the
01:16:56phone again.
01:16:57But that was like part of why,
01:16:59I mean, it was one that was the
01:17:00cost of 900 numbers, but that
01:17:01was the other part of it where
01:17:02it was like 900 numbers were a
01:17:04little sketchy as a kid.
01:17:05You weren't really supposed to
01:17:06call them.
01:17:07And it was because like, yeah,
01:17:08it was a bunch of like, not con
01:17:10artists, but basically licensed
01:17:11con artists pretending to be
01:17:13other people.
01:17:14And it's like, that's gross.
01:17:17You hear this from tech people
01:17:18all the time.
01:17:19They're like, this is just the
01:17:20same as this other small problem
01:17:21that we had in the past.
01:17:22But now you're doing it at
01:17:23scale.
01:17:24Yeah, like now you're doing it
01:17:25at scale and that's...
01:17:26Like what if we took a small
01:17:27problem and made it global?
01:17:28It's like, I think maybe
01:17:30there's something different
01:17:31here.
01:17:32This thing, by the way, that
01:17:33you just mentioned, it's come up
01:17:34several times now.
01:17:35It is this unshakable belief
01:17:38that these models are going to
01:17:39get better.
01:17:40Yeah.
01:17:41And like on the rate of like
01:17:42Moore's law, we'll get there.
01:17:43Yeah.
01:17:44And I don't, that metric
01:17:45doesn't exist.
01:17:46Like there isn't a Moore's law
01:17:47of LLM models yet.
01:17:48You know, we've only really
01:17:49experienced 3.5 to 4 in the
01:17:51public.
01:17:52Mm-hmm.
01:17:53And I understand that they've
01:17:55all got stuff they're looking
01:17:56at that we can't see.
01:17:57Sure.
01:17:58Someone's got to give me the
01:17:59number.
01:18:00Like give me the rate of change
01:18:01before I believe that this model
01:18:03will be able to simulate
01:18:04whatever happened to me on
01:18:05forums when I was a child.
01:18:07Like, you know, like you
01:18:08wouldn't last a minute in the
01:18:09asylum when they raised me.
01:18:10It's like, I don't think so.
01:18:12Well, so yeah, there's no
01:18:13Moore's law.
01:18:14There's
01:18:18I mean, there there's no
01:18:19Moore's law.
01:18:20There's scaling law.
01:18:21And the idea is that as you
01:18:22continue to put more data and
01:18:24compute into these models,
01:18:25they get exponentially better.
01:18:26That continues to hold true.
01:18:28If you compare the output, the
01:18:30creative output of like GPT-2
01:18:32or GPT-3 to 4, it's a massive
01:18:35difference.
01:18:36But they're also running into
01:18:38that line is not linear.
01:18:39That's what I'm saying.
01:18:40I'm saying everyone who works
01:18:41and I believes that it is still
01:18:42linear.
01:18:43Yeah, but it's not because
01:18:44we've just seen multiple stories
01:18:45about how they're all desperate
01:18:47for data to the point that
01:18:48they're like, what if we make
01:18:49AI machines to just create more
01:18:51content for us?
01:18:52That can't go wrong.
01:18:53Oh, the thing I'm hearing is
01:18:54it's going to be synthetic data
01:18:55for training.
01:18:56So literally, you're going to
01:18:57have the AIs get good enough
01:18:58that they start talking and
01:18:59then the AIs train on the AIs.
01:19:01That's where we're going.
01:19:02All right.
01:19:03We got to end this here.
01:19:04I can't.
01:19:05I can't stay here.
01:19:06I'm just saying make real
01:19:07friends.
01:19:08Go to the forums while you can.
01:19:10All right.
01:19:12Get in there while the getting
01:19:13is good.
01:19:14Get past the username.
01:19:15Get their actual name.
01:19:16Because they may be an AI to
01:19:17hear.
01:19:18Request a driver's license
01:19:19immediately.
01:19:21We're taking a break.
01:19:22We'll be right back with the
01:19:23lightning round.
01:19:28All right, we're back.
01:19:29It's the lightning round.
01:19:30I will say the first segment of
01:19:31the show was much more like
01:19:32lightning round and this is
01:19:33just going to be different.
01:19:35It's fine.
01:19:36This lightning round remains
01:19:37unsponsored, but many options
01:19:38exist.
01:19:40For you, the potential
01:19:41advertiser.
01:19:42But you can't do AI ads.
01:19:44Yeah.
01:19:45You have to let me do it.
01:19:48You can only do AI ads if
01:19:49they're trained on Nila's
01:19:50voice.
01:19:51What's the most dangerous
01:19:52advertising you can buy?
01:19:53It's letting me do the ads.
01:19:54All right.
01:19:55Kranz, let's start with you.
01:19:57NVIDIA is super rich.
01:20:00Can I say, so the story, right,
01:20:01is NVIDIA overtook Microsoft
01:20:03and Apple as the world's most
01:20:04valuable company.
01:20:05Microsoft and Apple have been
01:20:06going back and forth on
01:20:07market cap for a while.
01:20:09NVIDIA is not bigger than both
01:20:10of them.
01:20:11The crazy thing about this
01:20:12whole story to me is, one,
01:20:14NVIDIA has been around for a
01:20:15million years.
01:20:17Video game fans exist.
01:20:20Right?
01:20:21Like people put NVIDIA cards
01:20:22in their PCs all over the
01:20:23place.
01:20:24It has been on this run for a
01:20:26long time, right?
01:20:28This is years now of NVIDIA
01:20:29stock skyrocketing.
01:20:31And the way CNN covered it, the
01:20:32headline in CNN was, this
01:20:33company whose name you can't
01:20:34pronounce is now the world's
01:20:35most richest company.
01:20:37It's like, what?
01:20:38It's actually pretty easy to
01:20:39pronounce.
01:20:40I feel like.
01:20:41Yeah.
01:20:42One, it's not that hard.
01:20:43It's easier than SSI.
01:20:44An editor and a writer
01:20:45together.
01:20:46And they're both like, I don't
01:20:47know what this is.
01:20:48We got this.
01:20:49It's just like at some point, I
01:20:50understand we run a tech
01:20:51publication.
01:20:52We're like swimming in the
01:20:53water.
01:20:54But like, they've been headed
01:20:55here for at least two years.
01:20:57Yeah.
01:20:58We had to hear about them like
01:20:59cryptocurrency, right?
01:21:01NVIDIA had this big, all these
01:21:02big stock jumps because of
01:21:03cryptocurrency, because
01:21:04everybody was buying the GPUs
01:21:05because that was the best way
01:21:06to trade.
01:21:07Oh, my God.
01:21:08This was completely out of my
01:21:09brain.
01:21:10This was gone.
01:21:11This information had been
01:21:12deleted.
01:21:13Yeah.
01:21:14Because it was 2020.
01:21:15Just like crypto.
01:21:16We don't think about 2020.
01:21:17And so that was happening to
01:21:18NVIDIA.
01:21:19And so it was like steadily
01:21:20ticking up.
01:21:21And it was like, oh, boy, AMD
01:21:22and Intel might be able to beat
01:21:23it.
01:21:24And then it was like, we're
01:21:25going to take all the crypto
01:21:26stuff out of our GPUs.
01:21:28And then everybody was like,
01:21:29hey, AI's here.
01:21:31Yeah.
01:21:32Yeah.
01:21:33And others have been.
01:21:34Lost their minds.
01:21:35The effect of this inside
01:21:36NVIDIA is fascinating.
01:21:38I've heard and there's been some
01:21:39reporting and I want to keep
01:21:40doing more reporting on it.
01:21:41So if you have stories, if
01:21:42you're listening to this and you
01:21:43have gotten fabulously wealthy
01:21:44at NVIDIA, hit me up.
01:21:45We can talk off the record.
01:21:46They have a retirement problem
01:21:48inside NVIDIA.
01:21:49There's a there was a post I saw
01:21:51where it said, if you join as a
01:21:53mid-level software engineer five
01:21:54years ago and your total comp was
01:21:56seven hundred thousand, which is
01:21:57a lot. Right.
01:21:58But in fact, that's actually
01:21:59pretty middle of the road for a
01:22:00good engineer.
01:22:01Your compensation yearly now is
01:22:03ten million dollars.
01:22:05So rules.
01:22:07They actually have people just
01:22:09retiring like en masse.
01:22:11And so like the question
01:22:13for NVIDIA is like, how what do
01:22:15you do to motivate people when
01:22:17people have gotten so fabulously
01:22:19wealthy so fast?
01:22:20And it's obvious there's going to
01:22:21be a correction. It's obvious
01:22:22NVIDIA is not going to say the
01:22:23richest, the most valuable
01:22:24company in the world. Right.
01:22:25So how does that collapse
01:22:27happen? Is it a slow collapse?
01:22:29Is it a fast one?
01:22:30And what happens to the
01:22:31employee base is fascinating.
01:22:32Well, you get rid of management
01:22:33overhead and product cycles.
01:22:34There you go.
01:22:35That's obvious what you do.
01:22:36Merge with SSI.
01:22:39It's going to last a while.
01:22:40Right. Because they are the only
01:22:42ones making the really good
01:22:44A.I.
01:22:45processors for server levels.
01:22:47Right. Well, they have software
01:22:48lock into with.
01:22:49Yeah, like Google.
01:22:50Google. Google's their main
01:22:51primary competitor.
01:22:52Right.
01:22:53Yeah.
01:22:54Yeah.
01:22:55Yeah.
01:22:56Arm AMD.
01:22:57I would say AMD is like an arms
01:22:58dealer is probably the biggest
01:23:00one.
01:23:01Google doesn't sell TPUs really
01:23:02as much.
01:23:03But NVIDIA has got a big
01:23:04head start.
01:23:05This is one of the things that
01:23:06has kept Intel propped up for
01:23:07years was because Intel had this
01:23:09huge head start in the server
01:23:11space.
01:23:12Ninety nine percent of the
01:23:13server space.
01:23:14So, you know, they can bomb a
01:23:15couple of years
01:23:17against Qualcomm and be fine
01:23:19because they got the servers.
01:23:20And I think NVIDIA is kind of
01:23:22feels like not quite as cushy a
01:23:24position, but similarly
01:23:26cushy.
01:23:28Yeah.
01:23:29Yeah.
01:23:30I mean, they have CUDA.
01:23:31Like what they have is a
01:23:32software layer that makes their
01:23:34GPUs more valuable.
01:23:35Yeah.
01:23:36Which AMD does not.
01:23:37But if you run Azure, I mean,
01:23:39we've had all these people in
01:23:40the show.
01:23:41Now let's talk to them all.
01:23:42If you run Azure or AWS or
01:23:43Google Cloud or whatever, you do
01:23:44not want
01:23:46to be beholden to CUDA and
01:23:48NVIDIA.
01:23:49Yeah.
01:23:50You want all kinds of things to
01:23:51happen.
01:23:52That's why Apple and NVIDIA
01:23:53don't look at each other
01:23:54anymore.
01:23:55It is very funny that Apple and
01:23:56NVIDIA hate each other.
01:23:57It is.
01:23:58It used to be just like a
01:23:59little funny because Apple
01:24:00would insist that bad graphics
01:24:02cards were good graphics cards.
01:24:04And that was just funny, like
01:24:06in its way.
01:24:07Now it's deeply funny.
01:24:09Yeah.
01:24:10Extremely good.
01:24:11Because all that open AI
01:24:12integration is happening in
01:24:13H100s that Apple does not want
01:24:14to talk about.
01:24:15Yeah.
01:24:16And doesn't want to pay for.
01:24:17And that's the thing.
01:24:18Market forces will ensure that
01:24:19NVIDIA does not stay the top
01:24:21GPU supplier in the world
01:24:22because they are too
01:24:23expensive.
01:24:24And no one wants Jensen being
01:24:26the king of GPUs.
01:24:27Everyone wants to have
01:24:28options.
01:24:29And so NVIDIA's challenge to
01:24:30justify this valuation and to
01:24:31keep growing is to build lock-in.
01:24:33It would make a lot of sense for
01:24:35them to start competing
01:24:36directly with Google Cloud and
01:24:37Azure to become a one-stop shop
01:24:39because otherwise people will
01:24:42outchip them.
01:24:43That's just going to happen.
01:24:44It may take time.
01:24:45And we've been seeing that for
01:24:46years with them.
01:24:47On the gaming side, they got
01:24:49really, really good at the
01:24:50lock-in.
01:24:51They got really good at being
01:24:52like, you can't use these AMD
01:24:53processors because we went out
01:24:54to every single game developer
01:24:56and made sure they were using
01:24:57our tools.
01:24:58So you had to use, like, we get
01:25:00the better stuff.
01:25:01So it's going to be really
01:25:04interesting over the next
01:25:05couple of years as they just
01:25:06like absolutely duke it out
01:25:08with some of these other
01:25:09companies.
01:25:10And I think a lot of people are
01:25:11going to be surprised at how
01:25:12vicious they get.
01:25:13Oh, yeah.
01:25:14But it's like one step down
01:25:15from that, right, Alex?
01:25:16Yeah.
01:25:17Heath was saying these things
01:25:18are free.
01:25:19Like, Apple's not paying for
01:25:20it.
01:25:21If there's no business model one
01:25:22step down from we paid all of
01:25:23the money in the world for H100s,
01:25:25the price of the H100 will just
01:25:26fall.
01:25:27Yeah.
01:25:28If the demand is there right
01:25:29now because everyone thinks
01:25:30they're going to monopolize
01:25:31something, what it is remains
01:25:33to be seen.
01:25:35But that's why the price is
01:25:37high.
01:25:38We'll see.
01:25:39All right, I'm going to go
01:25:40next because I want Heath to
01:25:41end the lightning round with
01:25:42this.
01:25:43We just continue to track tech
01:25:46litigation.
01:25:47The United States Department of
01:25:49Justice has sued Adobe for
01:25:51deceiving subscription pricing
01:25:53in Creative Cloud.
01:25:55And it actually is deceiving.
01:25:57It is deceiving in one specific
01:25:58way, which is when you sign up
01:26:00for Creative Cloud, you can pick
01:26:02annually build, annually build
01:26:04monthly, or monthly.
01:26:06Annually build monthly is the
01:26:07default choice.
01:26:09So that means you pay for a
01:26:10year, but they charge you per
01:26:11month, and you've signed a one
01:26:13year contract.
01:26:14And if you try to cancel early,
01:26:15they just charge you the rest of
01:26:16the year.
01:26:17Or you can sign up for monthly
01:26:18where you can just cancel after
01:26:19every month.
01:26:20Because it's the default,
01:26:21because it's the cheapest,
01:26:22people pick it, and they think
01:26:23it's a monthly subscription.
01:26:24There's no fine print.
01:26:25It doesn't actually disclose
01:26:26that this will happen.
01:26:27They try to cancel to get stuck
01:26:29with the rest of the bill.
01:26:31This is bad.
01:26:32This is straightforwardly bad.
01:26:34For a company with basically a
01:26:36monopoly market share in these
01:26:38tools, it is also just beyond
01:26:41the pit.
01:26:42Like, this is piddly shit.
01:26:43What are you doing?
01:26:44Like, people are going to buy
01:26:45Photoshop, man.
01:26:47That's a local gym move on
01:26:49their part.
01:26:50Yeah.
01:26:51I've been thinking about this a
01:26:52lot with Adobe.
01:26:53I've basically insisted that
01:26:55Verge over cover Adobe over the
01:26:56past few years because they're
01:26:58such a central player in
01:26:59everything that's going on.
01:27:01Like, influencer media depends
01:27:04on the existence of Adobe.
01:27:06That's weird, right?
01:27:07Premier runs YouTube.
01:27:09Like, YouTube never built
01:27:10creative tools, which is one of
01:27:12the weirdest things about
01:27:13YouTube.
01:27:14They had that one app on the
01:27:15phone once, but they've never
01:27:17built a video editor.
01:27:18The success of TikTok is they
01:27:20made a video editor that's
01:27:21really good, and they made all
01:27:22those templates, and they made
01:27:23it easy to do all this stuff.
01:27:24YouTube never did that.
01:27:26They were like, upload some
01:27:27videos, and then people built
01:27:28the entire creator's ecosystem
01:27:30on the back of Premier, largely,
01:27:32and a little bit of Final Cut
01:27:33because there's so many
01:27:34creators who use Macs, but
01:27:35mostly Premier.
01:27:37And Google never built that
01:27:38stuff.
01:27:39So you just, like, look at this
01:27:40industry.
01:27:41Like, Adobe's the center of
01:27:42everything.
01:27:43They're the center of all these
01:27:44AI creativity battles, right,
01:27:46because they have Adobe stock,
01:27:47and people use Photoshop and
01:27:48General Fill.
01:27:49And the more you cover Adobe,
01:27:50the more you realize there's a
01:27:52massive difference in how the
01:27:54company perceives itself and
01:27:55what it does and the tools it
01:27:56makes and how everyone else
01:27:58perceives Adobe.
01:27:59It's a yawning gap.
01:28:01I've said this before.
01:28:02We had Shantanu Ryan on
01:28:03Decoder.
01:28:04He never talks to anyone
01:28:05really.
01:28:06He shows up on CNBC after
01:28:07earnings reports and never
01:28:08gives wide-ranging interviews.
01:28:09I was so excited for that
01:28:10interview.
01:28:11This guy never talks.
01:28:12We asked him all the
01:28:13questions.
01:28:14Some of the questions were
01:28:15pretty harsh.
01:28:16The audience was mad at me just
01:28:17saying, like, why would you
01:28:18platform Adobe?
01:28:19They're like, why would you
01:28:20platform Adobe, which is an
01:28:21incredible question, because
01:28:22it's Adobe.
01:28:23I don't think I've seen a
01:28:24company burn that this much
01:28:26goodwill ever, right?
01:28:28And then on top of it, they had
01:28:30their Terms of Service debacle
01:28:33where we've covered Terms of
01:28:35Service changes like this so
01:28:37many times, like, I could
01:28:38barely even pay attention to it,
01:28:40where they change their Terms of
01:28:41Service to let them do content
01:28:43moderation on their cloud
01:28:44services.
01:28:46So if you run a cloud service,
01:28:48people will put bad things on
01:28:50your cloud service.
01:28:51That is just as true a fact of
01:28:53nature as there is.
01:28:54Like, horrific shit will end up
01:28:56on your cloud storage system if
01:28:57you have a scaled cloud storage
01:28:59system.
01:29:00And what you don't want to do is
01:29:01give a bunch of contractors
01:29:02PTSD, which is a real thing that
01:29:04has happened that we have
01:29:05covered by making them look
01:29:06through stuff on cloud services
01:29:08or look through reports.
01:29:09So you build automated systems to
01:29:10go through your cloud services
01:29:11and find the child abuse
01:29:13material.
01:29:14That's what you do.
01:29:15So Adobe changes its terms to
01:29:17let people know it's going to be
01:29:18used on Creative Cloud.
01:29:19People assume that they're going
01:29:20to train on AI, and they freak
01:29:22out at Adobe, and that Adobe has
01:29:24so little goodwill that even as
01:29:26they tried to explain this,
01:29:27people are like, we don't
01:29:28believe you.
01:29:29And so now they've updated their
01:29:30Terms of Service again.
01:29:31And you just connect that
01:29:32directly to, oh, you're getting
01:29:34sued by the government because
01:29:36annual paid monthly was not
01:29:38clear to people, and you charge
01:29:40them the rest of the money when
01:29:41they cancel.
01:29:43It's crazy to me that this
01:29:45company is so important but does
01:29:47not realize that it needs to
01:29:48maintain the goodwill of the
01:29:50people that depend on it.
01:29:51Well, you don't have to when
01:29:52you're a monopoly.
01:29:53You can just be a monopoly.
01:29:56I've met all their executives.
01:29:58They're not dumb.
01:30:00It's baffling to me.
01:30:03Yeah, but they're, you know,
01:30:04they can do what they want.
01:30:05They're monopoly.
01:30:06What are you going to do?
01:30:07Go to DaVinci?
01:30:10Maybe.
01:30:11Maybe.
01:30:12A lot of people, yeah.
01:30:15Them not acquiring Figma, I
01:30:16think, will go down.
01:30:17We're going to come back on the
01:30:18failed Figma deal, one, because
01:30:21Figma will turn into a different
01:30:22company, and we'll see if they
01:30:23can go compete with Adobe.
01:30:24But that was driven as much by
01:30:27some of this burned goodwill as
01:30:28anything, and I think we're
01:30:29going to see the creative tools
01:30:31industry bifurcate on that
01:30:32moment over the course of
01:30:33history.
01:30:34Because if they had acquired
01:30:35Figma, a lot of things would
01:30:37have been a lot different, I
01:30:38think.
01:30:39All right, last lightning round
01:30:40one.
01:30:41We've got to end here.
01:30:42Heath, this is like the best
01:30:43story in history.
01:30:44Okay, yeah.
01:30:45The San Francisco Standard has a
01:30:46fun story where techies are
01:30:48increasingly paying up to $1,200
01:30:50an hour for fashion consultants
01:30:52because of CEOs like Mark
01:30:54Zuckerberg deciding to wear
01:30:55chains.
01:30:57And I just encourage everyone to
01:30:59read this story.
01:31:00There is a side-by-side photo of
01:31:02a guy who's gone through one of
01:31:03these fashion consultants.
01:31:05I would argue he looked better
01:31:07before.
01:31:08I'm not going to judge, you
01:31:09know, whatever.
01:31:11Or at least the same.
01:31:12It's just like, as my
01:31:13grandfather would say, you know,
01:31:14money doesn't buy sense.
01:31:16And reading this, techies, just
01:31:20be who you are.
01:31:21You don't need to try to look
01:31:22cool, and you don't need to pay
01:31:24$1,200 an hour to try to look
01:31:26cool.
01:31:27The internet is there.
01:31:29You could just research on your
01:31:30own.
01:31:31This side-by-side photo is
01:31:32brutal.
01:31:33It's brutal.
01:31:34It's not brutal because he looks
01:31:35better or worse.
01:31:36He looks the same.
01:31:37Yeah, the same.
01:31:39Both of them are kind of just
01:31:40baggy outfits.
01:31:41Yeah.
01:31:42And then, like, it's perfect
01:31:43because at the end, like, the
01:31:44story, they're, like,
01:31:45interviewing a founder who's
01:31:46done this.
01:31:47And then he's like, I'm making
01:31:48an AI app that will do this for
01:31:49people for $20 a month.
01:31:51And it's like, the circle just
01:31:53completes itself.
01:31:55That's amazing.
01:31:56You know, Amazon tried to do
01:31:57that with their weird camera.
01:31:58Yeah.
01:31:59Super didn't work.
01:32:00Yeah.
01:32:01I would say, if you're paying
01:32:02someone this much money to take
01:32:03you to Bloomingdale's, you can
01:32:04do better than this photo.
01:32:05That's what I'm saying.
01:32:06Like, it's...
01:32:07Oh, I don't think he went to
01:32:08Bloomingdale's.
01:32:09No, it says.
01:32:10That's what it says.
01:32:11Yeah.
01:32:12He's in the dressing room at
01:32:13Bloomy's.
01:32:14Oh, my gosh.
01:32:15Get out.
01:32:17Leave the store, sir.
01:32:19No, that's a bad look if that's
01:32:20what you're pulling out of
01:32:21Bloomingdale's.
01:32:22That's like, you get that at
01:32:23Dillard's.
01:32:25That's brutal.
01:32:26All right.
01:32:27We got to end it here before we
01:32:28do more shaming.
01:32:29Sorry.
01:32:30Sorry for your fashion
01:32:31shaming.
01:32:32Dillard's is fine.
01:32:33You can look cool no matter
01:32:34where you shop.
01:32:35I love a Dillard's.
01:32:36Everyone is constantly asking
01:32:37me where the jackets come from
01:32:38and all of them were pandemic
01:32:39purchases at either Home Depot
01:32:41or Walmart.
01:32:42I'm just telling you, that's
01:32:43like, I had money to spend and
01:32:44there are only two stores to
01:32:45spend it at.
01:32:48You can look cool no matter
01:32:49where you shop.
01:32:50Dillard's is fine.
01:32:51Just know who you are.
01:32:52That's a quote here.
01:32:53That's it.
01:32:54Just know who you are.
01:32:55Just be that person as loudly
01:32:56as you can.
01:32:57All right, that's it.
01:32:58And also make an agent of
01:32:59yourself and have it flirt with
01:33:00other people on Instagram for
01:33:01money.
01:33:02More gold chains.
01:33:03More gold chains.
01:33:07All right.
01:33:08We'll be back on Tuesday with
01:33:09more of the Windows ARM laptop
01:33:11benchmark scores.
01:33:12That's a big deal.
01:33:13I'm excited for those reviews
01:33:14to come out.
01:33:15And then we got all kinds of
01:33:16stuff.
01:33:17It's a summer.
01:33:18It's like wild time in tech.
01:33:19And then we're going to get rid
01:33:20of all of our management
01:33:21overhead and product cycles.
01:33:22Just pure intelligence.
01:33:23Can't wait.
01:33:24VBI.
01:33:25Thanks to Heath for filling
01:33:26in.
01:33:27Good luck to David doing
01:33:28whatever he's doing.
01:33:29That's it.
01:33:30That's for a chest.
01:33:32And that's it for the
01:33:33Vergecast this week.
01:33:34Hey, we'd love to hear from
01:33:35you.
01:33:36Give us a call at
01:33:37866-VERGE-11.
01:33:38The Vergecast is a
01:33:39production of the Verge
01:33:40and Vox Media Podcast
01:33:41Network.
01:33:42Our show is produced by
01:33:43Andrew Marino and
01:33:44Liam James.
01:33:45That's it.
01:33:46We'll see you next week.