OpenAI and Jony Ive's AI super-gadget
01:53:50 Share

Hi, folks. Logan here. And Tulsi. We're from the Google DeepMind team. We're releasing Gemini 2.5 Pro, our most intelligent model. It's now available for you to test out in Google AI Studio. It's been awesome to see what everyone's been building. From creating mini games from a single prompt to debugging 50,000 lines of code, we're excited to see what else you create. Try it out today on Google AI Studio by going to aistudio.google.com and let us know what you built. ♪

Hello, and welcome to the Roadcast, the flagship podcast of Alex Heath and I trying to talk to Sergey Brin and being deflected at every opportunity. It was rough. It was bad. I'm your friend, Alex Heath is here. Hey, what's going on? David Pierce is here. Hello. So Alex and I were at I.O., at Google I.O. It's big news of the week. Google announced a bunch of stuff. We're going to talk about it. But there was this moment.

where Sergey was in the AI sandbox, like the demo area. He was there last year too. And he like talked to reporters for a long time, just riffing about AI. So he was there at the end this year. And I kept describing it as like the sword in the stone. Like all these reporters kept on trying to talk to him and just like failing. Including me.

including both of us. This makes me think somebody must have given him a speech not to talk to reporters. Cause the other thing that Sergey Brin likes to do is just say some truly wild stuff that is not helpful. If you're Google, uh,

But he likes to talk to people. So I assume somebody gave him a certain speech not to talk to reporters. No, I don't think it was that actually. He was working on a demo. Like he was in the demo. He was like playing with the stuff. Right. And he it was flow the new video tool that Google released that runs on VO3. It's like it's basically it's like iMovie for AI video. Right.

And he was trying to get it, I think, to make himself. Like he was prompting it to make himself. I missed that. That's what I heard like secondhand. But he was, I mean, I watched him using it and say out loud, like this isn't working the way it should. And then someone else, you know, then all the Googlers were like, this isn't working. And he looked, he was very locked in on what I was doing. And then someone else told me that,

It's not supposed to make Sergey Brin, so he was trying to make it make Sergey Brin. That's amazing. So he was totally focused on this thing not working, which is why he wasn't talking to anyone. And then he went over to do the glasses demo. Is that the AI executive of Googling yourself every once in a while? You train the new model and then see how successfully it can replicate you. And that's your model of how big a deal you are. Yeah, probably. I mean, how often do you think Sundar Googles himself?

I like that must be crazy making because you are the one person who can actually just change the search rankings and be like, like Sundar Pichai is super cool.com is going to be the first result. Everything is an episode of Silicon Valley. Every single thing. I mean, it was just very funny because everyone saw him. He was just like in the middle of the room and I had just failed. I walked over and I was like, hey, man. And he was like, whatever. It's like left. And then like three other reporters had that experience. And Alex, you know.

Star reporter Alex Eve like super confidently saunters over there. And I was like, he's going to try to pull that sword out the stone, man. He just like didn't happen. The eye roll that I got from a Google communications executive as I approached Sergey was really good. Yeah. It was good. But later he crashed another keynote. Yeah. Like Demis Hassabis, the CEO of Google DeepMind was doing like a keynote interview type deal, fireside chat interview.

And so he just like came on stage with him and then he did in fact say just bonkers things.

Yeah. Including like, I just get to tell this guy what to do and we may be in a stack of simulations. Yeah, there's a lot. His answer to do we live in a simulation was if you think we live in one simulation and the people who made that simulation also live in a simulation and you have to deal with that. Yeah. Nila, you've been going to IO for longer than me. Did this feel like a return to early IO for you? Like, I heard a few people be like, especially the Sergey talk, but...

It just felt like 10 years ago. Like, I don't know. Do you see what I'm saying there? I do. It was different. Google is just more corporate now in that way. We'll come to it. I think it was very confident, and I think that's what people were picking up on. Google's feeling itself again in a real way that we should talk about. Yeah. They're like, we can do stuff, and they haven't felt that way. Like-

For like the past several years, like here's what we've done. We've added one feature to Android and Samsung is going to use it. And that has been IO. And this was very different. There was a lot of confidence. We should go after that. We're going to come to that in a second. I think we have to start.

with johnny ive and sam altman trying to upstage io by announcing a company called io like literally the next day before we get into all the news we're gonna talk about that we're gonna talk about like i said google io there's lightning round brendan has brendan uh two pieces of housekeeping one uh we have a survey about the vert cast what you want from us what you want from our podcast that's voxmedia.com slash survey go take the survey someone will make a chart and then we'll uh

We'll listen to them. I think we're obligated to say that we'll listen to them. I think you know the truth. Here's what I promise is that if you take the survey, I will give a readout on this podcast of Nilay yelling at whoever presents us the chart at the end of this process. So do the survey. No one can tell me what to do. That's what we sell here.

And then second, speaking of no one can tell us what to do, one of the ways that we protect ourselves from being told what to do is by having a direct financial relationship with you, the reader slash listener. And Verge subscriptions are on sale for Memorial Day because we're like, why not have a sale? They're 40% off, which means they're $35 a year. That's theverge.com slash subscribe. And

And what you were buying, just to be very clear. Yes, a bunch of great reporting. David getting increasingly redder on the YouTube channel every week. That's something that we provide for that money. And then, you know, our ethics. What we fundamentally sell here is our ethics policy. You just can't tell us what to do. We don't do brand integrations, sponsorships. I don't thank a phone case maker in the middle of the review videos because no one can tell us what to do. And we get to do that because you pay us the money directly and then take a survey and then we don't listen to the survey. It's a weird feedback loop.

But it works. I think it's been working for us. $35 is good too. That's like a salad in Mountain View. It really is. That's one Erewhon smoothie in LA terms. It's good. We're new to this particular game. This is our first ever sale. I've

I literally feel like I'm trying to sell you a Cadillac right now. But we are learning all these moves and it's like kind of fun. It's like new problems to solve. So let us know how we're doing. And by let us know how we're doing, I mean spend $35 a year on our product. Thank you.

Is that how you sell anything? Doing great. Neal is on three hours of sleep, guys. I just want to say that going into this show. My flight was delayed coming home from IO and so it effectively turned into a red eye and I'm just super loopy. I'm drinking coffee during the afternoon Verge cast for the first time in years. So we're going to get through it. It's going to be great. Okay, let's start with Johnny Ive.

In his arms. I mean, that's what they announced. They announced Johnny Ives walk, right? Sam Altman, Johnny, I've released the video. I think they got scooped a little bit, Alex. Like I think the journal and Bloomberg kind of had it.

And then they put out this video to announce it. That's my sense of it. They had some embargoed stuff. Johnny and Sam Altman did a couple interviews. They did one with Bloomberg, one with The Times. And then in a very good spot, people saw Johnny actually filming this video that he did with Sam on May 2nd, like early May. That's what it was.

this video was done almost a month ago. So there's this announcement. The basics of the announcement, as near as I can understand it, is that Johnny Ives started a new company that is staffed by a bunch of his star former Apple designers. That company is called IO. Very funny because this company lasted for two minutes.

And it has been sold to OpenAI for $6.5 billion in OpenAI stock, which is also a very complicated idea. And then Johnny Ive himself is not going to go work at OpenAI. He's going to stay at his design agency, Love From, which will now have no other clients except for OpenAI. At some point, yeah. The only tick that's more complicated than that is that OpenAI's investment firm, I believe, was associated with I.O.,

But OpenAI will tell you that OpenAI's investment firm is not part of OpenAI. Yeah, fun fact. A thing I found out this week is that the OpenAI startup fund is not part of OpenAI. None of this makes any sense. It's all pretend. Like, the thing to remember at the beginning of this is all the money is pretend.

Like all the money that Sam Altman can raise is always pretend. All of the different machinations of how the money moves around is pretend. Whether OpenAI is or is not a for-profit company is pretend. This new company they created is pretend. Like it's all made up. And the products that they've announced, the most pretend of all. Yes.

So they released a video. So they announced this deal. They're going to work together. They're going to make a new generation of hardware products. I think they're coming. The first are coming next year. You got to say like announced at the end of next year and shipping in 2027. But what they really shipped was this video. Yes. And I think the reason my adult, my sleep deprived brain is like the video was the people. Lots of people saw them making this video in early May.

And I think everybody understood something was up. And the fundamental thing they announced in this video is that Johnny Ive walks around like he owns the place. Yeah. You just have to watch it. It is literally one of the most confident walks in history. It's very good.

And it's this very, you know, grandiose thing. And it looks like they're just walking through San Francisco, you know, with a camera following them. And, you know, it's just the normal city street on a given day. And then you see people shooting when they were actually filming and people were like filming the set. And it's all staged, like everyone walking across the intersection with Johnny with his arms swinging like crazy.

It wasn't made by AI, but it also wasn't. Yeah, it does have a lot of vibes. Lots of people on various social platforms are like counting the wine bottles behind them. I saw one person fully crash out being like, the cups are moving from shot to shot. And it's like, yeah, it's because it's a lot of different shots. They edited together. It's fine, you guys. Anyway, the video is those two sitting in a restaurant having little cups of espresso together, basically saying that they love each other. Mm-hmm.

And that there's a new class of computing devices to be made with AI. And then Johnny is going to do that in some way with OpenAI. That is the sum of...

the announcement in the video. Also, San Francisco is awesome. There's a lot of love for San Francisco and just generally like, what if we change the world? You know, like it's, it's just this big, like, you know, idea of two guys changing the world. I mean, who doesn't love that? I think the only like specific thing I can think of that comes out of it is the,

there is one device that already exists in some form. They have made a thing together. And there's been reporting that these two have been working together on hardware for a long time now. So something exists. I don't know how many of them there are. I don't know how finished it is, but like,

it is clear that there is one product in actual honest to God development, but their plan is to build a whole family of devices. Yeah. So Altman actually says in the video that he just took the prototype home. So this was an early May. He just got to take the first device home. I think we should talk about the reporting on the device and how it may or may not work. I just want to really quick through really quickly run through how this whole thing came about. Cause I think that's probably surprised a lot of people when they saw this. So,

I've talked to OpenAI and LoveFrom now, and the rundown is basically that two years ago, Sam and Johnny were introduced. About a year ago, actually almost exactly a year ago, they decided to start a new division inside LoveFrom to work on AI devices together.

And that division includes a bunch of legendary early Apple design leaders that worked under Ive, including Evans Hankey, who ran the industrial design group at Apple after he left. And then she left in 2023 and went to Love From. And they were doing this for several months. And then employees at OpenAI were starting to come in and work at Love From together with them. And then towards the end of last year,

Love From was thinking about raising a lot more money for IO, this division of the company doing hardware development.

And that was when Altman was like, what if we just buy you guys fully? So they had already invested in this love from subsidiary through the open AI startup fund that is not owned by open AI. Um, and they just bought the rest of it. And so now a team of about 55 people who are a lot of early Apple design people are going over to open AI and a new hardware division that's rolling up to directly to Sam and

And Johnny is staying at love from with his design team, but will now also oversee all of design at open AI, not just hardware. And Johnny also works with Airbnb. Um, he helps Brian Chesky on high level design stuff. He works with Ferrari. He's actually designing the first ever electric Ferrari. Um, and,

And I think he wants to keep doing other things. But I think once their projects that they're doing, you know, end, they're going to be almost fully devoted to open AI. But he's also staying independent. So pretty unusual. Because if you're Johnny, the idea that you will ever have another boss has to just not even. Yeah. It's just like not a square on the bingo card that exists. Yeah. But also like, Neil, your point about the thing that they made is this video is the thing, right? Like this is this is the cachet.

that OpenAI literally couldn't buy for any amount of money except to give it all to Johnny Ive. There is like, this company is desperate to convince people that it can make good, interesting, cool products and that it is not going to get lapped by big tech companies as all of the like science behind this stuff gets closer and closer. And like, it's,

The way to win here is to convince everybody that you're Apple and that you're going to do the hardware software services thing. And boy, is there not a better way to do that than to put Johnny Ive and his confident walk in your video. So like, I frankly would be shocked if at the end of all of this, Johnny Ive is like day-to-day involved in making products with OpenAI. But it doesn't matter. This, this, today it was like,

like the announcement was the thing. And that's, that's all you need from Johnny. Yeah. Literally just need his voice in their keynote. Right. For, you know, being like plastic is inevitable and like, it's fine. He'll be, he'll be fine. But I, before we get into the device, cause there's some reporting about what it might be. We have a lot of guesses or what it might be. The idea here is that Sam Holtman is Steve Jobs, right? Yep. That it's very clear. Like even in that video, he's like, I'm everything in my life has led to this moment. Altman, uh,

makes reference to the iPhone and the MacBook Air. And he's like, we're going to build the next version of this. And then, you know, they just spend a lot of time looking deeply into each other's eyes and talking about working together in California. And you're like, oh, I know exactly what they want people to take from this. Right. It's just very clear. They want to be Apple and they hired Apple's legendary design team. And then you're going to position Altman as the next Steve Jobs. And Altman reveres Steve Jobs. He's best friends with Brian Chesky, who also reveres Steve Jobs. Right.

Chesky was just on decoder. He talks about Apple as though he is like the world's foremost PhD student of Apple. It's very funny. Also, that episode doesn't come out. It'll come out next week by the time you're listening to this. I asked Chesky very directly about IVE and Allman, and all he said was, I'm very proud to have introduced them. Good dodge. I congratulated him on that dodge. But...

There's some big foundational things to point out about wanting to be the next Steve Jobs with AI as the foundational technology for the next device. And I think the most important one is that the AI systems don't work well yet. You just have to acknowledge it, right? I've and Jobs together...

The first thing they made was the iMac, right? That's the first big hit product that those two combined to make. And the iMac was an, right? It was just an all-in-one CRT. It was a piece of hardware. Like the big innovation was that it was blue. And the thing that made it a hit was that the web existed.

Right? That was the whole thing of the iMac was there's a new application. There's a new killer app for desktop computers. And we're going to sell you this blue one with a handle. And it's easier to set up than the Windows PC. And that rejuvenated Apple. And

The technology bet was that people wanted to get on the internet and that worked. That was just there. It was, it was happening. The next one was the iPod and I, it's easy to forget now, but the technology bet there was John Rubenstein went and found a tiny Toshiba hard drive and Phil Schiller invented the click wheel and they were able to package that together into the iPod. But the core technology was Toshiba made this little hard drive and they don't know what to do with it.

And you can like go read Steven Levy's book. It's called the perfect thing. That one's called, but it's literally like, what are we gonna do with this hard drive? And the music player, the iPhone was like, we invented multi-touch and that is a thing. We're going to make a whole phone out of this. And you just keep going on and on. There's these like big core technologies that I was able to package into new kinds of products with a visionary who was like, I know what kinds of technologies are,

will lead to products. Like the iPhone, for example, they were doing the iPad first and Steve Jobs was like, no, make a phone. Then we'll come back into the iPod, right? Like that's the dynamic between the two characters here. AI isn't that product yet. Like this is like I made an iPod and sometimes the hard drive just lies to you, right? Like that would be weird. I invented the iPhone, but the screen sometimes tries to bang Kevin Roos.

That is the level of capability that the core technology they're trying to build these products has. And I find myself just like utterly full of skepticism that OpenAI as it's presently configured, having lost a bunch of its research people and its product people.

Can can turn the corner. I don't know. Like maybe I'm just like very sleepy, you know, and more cranky than usual. But I'm just like all of the hit products had a core technology innovation that definitely worked. Right. And people were like really skeptical of some of them. People are really, really skeptical of typing on a multi-touch keyboard.

And the thing they were able to do was design their way through that and make it good. I mean, I think that's a lot of credit to give to the first generation of all of those technologies. Like the first iPhone was not good. It was like fascinating and cool, but it was like it was bad at a lot of things. But multi-touch worked. Multi-touch worked. I think so. Okay, here's what's wild to me. And I think this is going to be a theme of today's episode is technology.

I think your assessment of how good AI is, is correct. And I think most people don't care. I think most people think AI is very good. Yeah. And I, I increasingly don't know what to do with that. That like people, the, the like approval rating of chat GPT is actually very, very, very, very high. Yeah. Whether or not it's any good or lies to you or not is, is,

but also in a weird way sort of beside the point at this point. Like people are so enamored with this stuff that they're willing to look past all that stuff. And so maybe it doesn't matter. And I'm like, this is where I'm so torn on all of this that like,

the big question about all of these AI devices, right. It's like, a, do people want another device that they have to buy? Um, I think the answer to that is yes. Like everybody always wants another device to buy. Like that's fine. Point and shoot cameras have made a comeback, David. I think we can say yes. Smartwatches. When there was a reason for smartwatches, people got on board with wireless headphones that are really expensive. Like people will keep buying new gadgets if there's a reason for them. The question is, is there an

AI specific thing to do. And I think to me, that is both a like, is the tech any good question? And is there a UI for this that makes sense question? The UI for this that makes sense, I think is still a wildly open question. But like, is the tech good enough for people to want to use it all day every day?

I think the answer is yes. Whether it should be or not is a different question, but I think it is. The UI is voice. And I think back to Nilay's point, like there was a key tech bet of how people wanted to use it with each of the things that Johnny did with jobs. I think here people do want to talk to ChatGPT. They want to do advanced voice mode and they don't want to have to configure the action button on the iPhone and wait three seconds for it to load up and connect to the internet to actually work. I think if you had a...

Always on listening, ambient, aware thing that you can have a conversation with. I think people would do that. I mean, my wife told me yesterday, she was like, I just said thank you back to ChatGPT. And it made me feel very strange, like unprompted, you know? Like people are building relationships

I mean, we were hearing this at IO afterwards. People are having real, almost, relationships with these AIs. And yeah, if you introduce something that you can just talk to all day long, it's the Her concept, which Sam Altman is obsessed with, the Spike Jonze film. And what I've heard about this product and the form factor is a little mushy is basically

Basically, the idea is that bike demo we saw at IO this week where a guy is trying to fix his bike and he's talking to the AI throughout it and it knows what he's looking at and it's guiding him through it. And if it does work, if it's not hallucinating, that is just fundamentally a more human way to use technology. It's just like talking to it and it's helping you. Yeah.

Yeah. Does it lie? Does it get things wrong? Yes. Does it matter and will it be good enough? Maybe. Yeah. It's very funny how many tech demos at every stage of technology involve trying to fix something.

Like my first ever Microsoft HoloLens demo, they had me change a spark plug in an ATV. And like, this is the future of changing these spark plugs. And like, we're just doing it again. And lots of people have changed lots of spark plugs without like data centers in overdrive. Yeah, there's only like three ideas, right? It's like every new technology is for playing Assassin's Creed, fixing your car more successfully and buying elaborate vacation plans.

Like that's it. That's all we're doing with technology. 100% in Spanish. Yeah. Yeah. Those are the three. Which one is it? You just tell me which one it is. Right. I get all that. My skepticism on AI is not, do people like it? I've really come around to there is a widening gulf between how people talk about how much they hate AI online and then people using it in real life and how much they use it, how much they like it.

You can sense it. There is a turn. Sundar at I.O. on stage said we're entering a new phase of the platform shift. And what he meant was the products are coming out now. Right. We're not just talking about model scores. Like, here's a bunch of new products. Right.

I think Chattopaddy is that product for OpenAI, but it does... Chattopaddy, as it is currently configured, will never return on the massive amount of investment that OpenAI has taken. Correct. They need to be Apple. They need the iPhone. They need the app economy of the iPhone. They need to be the richest company in the world. But literally, that is the only target that makes sense for their level of investment. That's a big problem. And I'm just... Again, I'm just coming back to...

Does the core technology let them do the thing they want to do? Or is it people just like it so much that they will skip over the fact that it's brittle? And what I've has traditionally been successful at is designing to the limitation, right? He makes, he makes the limitation, the most successful part of the product. It's a little hard drive. The whole thing looks like a little hard drive. And we're going to talk about how many songs it has. Remember unapologetically plastic, like these are his moves, right?

Right. Like he says, I, I made it feel inevitable. And the thing that's inevitable is the limitation of the technology is the product. And you don't feel the limitation because it's designed to make that the, the thing that makes the product good. And you might think that I'm just like a sleepy man who's rambling, but think about the first iMac and how it was translucent. So you could see that CRT display.

That was the limitation of the product. They had to build it around this giant vacuum display, and he made it the centerpiece of the product so that over the lifetime of the product, they made that plastic clearer and clearer. That was the investment in the materials, was to do the appropriate EMF shielding in the plastic itself to be transparent instead of just translucent. And so the CRT became the center of this. But that's IVE.

Like at the highest level, that is Johnny Ive. We're going to make the antenna the band of the iPhone 4 so you can't, if you hold it wrong, the signal drops. That's Johnny Ive stuff, right? How do you design to the limitation of AI as it currently exists? I think the question of how much credit for that kind of stuff Ive gets is going to be one of the pieces of this that is really interesting. Because if you look at the Johnny Ive era at Apple after Steve Jobs...

It's a lot less of what you just described. Yep. And it is a lot more of Johnny Ive pursuing things that are aesthetically lovely for the sake of making things that are aesthetically lovely. And that's, that's fine. That's a perfectly good approach to doing a lot of things. That's also how you get the butterfly keyboard and it's how you get

like laptops that literally just stopped working for people for a gen, like they just made a series of bad decisions about laptops in order to make pretty laptops. It's how you get a mouse that you have to charge. That's just the thing that happened. And I think about even like the, the Apple watch, which was a true, like Johnny I've led project.

Johnny, I've decided to build a watch and then worked backwards towards the Apple Watch. And they landed in the wrong place. They landed in a place that I think is really fascinating, but it took like two pivots from the first idea before Apple had a hit product. And so I think this question of like,

Johnny's capability to make beautiful things is like beyond question. Well, we'll see. He's doing the electric Ferrari. We'll see how good that looks. I mean, nobody's perfect every time. Right. But I think like in terms of, I would bet, I would bet a lot of money on whatever they make looking and feeling great. Yeah. I think that's the bet that has to be the bet. I mean, look, so Sam Altman had this meeting inside open AI that was reported on by the wall street journal and,

And he told them that this $6.5 billion acquisition has the potential to add $1 trillion in value to OpenAI, which is just like, you know, making up numbers. You can just say things. But, like, this is where he started to talk a little bit about those limitations, Neal, that you're talking about. He said the device isn't going to be a pair of glasses and that it's something to wear, although it sounds like they're not going to want to call it a wearable. And then...

He's saying we're not going to ship 100 million devices literally on day one, duh, but he does think they'll get there faster than any company ever. And Almond is just – he's great at hype, right? Yeah. Obviously, he's hyping this up to employees who are like, why did we just give Johnny Ive $6 billion and he's not joining us? Yeah.

but, um, he did. So this is from the meeting. He said that the product will be capable of being fully aware of a user surroundings and life will be unobtrusive, able to rest in one's pocket or on one's desk and will be a third core device. A person would have after a Mac book pro and an iPhone. So here we go. This is like a better humane pin. I mean, not for nothing. The humane pin, uh,

was designed by two former Apple designers who regretted their work on the iPhone. Many of whom were called legendary when they started this company. Imran Chowdhury, as legendary as it comes, coming out of Apple. Altman gave them millions of dollars in funding. He was the single biggest investor in the company. Like, yeah.

It's just we're doing Humane again, but now it's Johnny Ive. I did say that Humane Pin was a poor product, which tracks with your review, David. Yeah, and the truth. That's right. Good call by Johnny. I would say that I called that watching the first demo video of the Humane Pin. And maybe I've lost some friends forever because of that. I've also called the Rabbit a poor product, which I think Jesse, Lou, and Rabbit loved because Johnny Ive knows who they are.

And they released a long statement being like, we love that the pie is growing. And I, good luck guys. Let me read this. Jesse sent us a statement. It's actually amazing. He says, it's an honor to get mentioned by Johnny Ive and Sam Altman. However, we don't like to be put side by side with Humane, a company that stopped trying, got acquired and shut down. That's very good. That's brutal. You can call us shit, but don't compare us to Humane. That's very good. That's very good, Jesse. It's really good. Yeah. Yeah.

You can tell that he wrote that, not the robot. You can tell when it's like, that's human. Don't compare me to humane. But those are the products. And there's a million of them. David, you read a bunch of note-taking products. I'm going to have a microphone that's recording all the time. Last time I saw Joanna Stern, she was wearing a wristband that records everything.

And the lights were backwards. It was red when it was off. It's called the B. V-Song reviewed it for us. It's a delightful and terrible product. Right. I mean, there's just like a million of these devices that's like, what if we just feed input into an AI and then it does something? Yeah. Summarize or talk to you or whatever. And that's what I mean by the limitation. The limitation of those products is very, very obvious.

Right? The AIs can't, they can't take actions for you. Right? The agents don't work yet. Maybe they will. We'll talk about that with IO. Like there's a lot of activity around how to make the agents work and go do stuff for you when you ask them to do it. The AI, right? They don't actually know a bunch of stuff. The search features don't work as well as they should. Like there's just a bunch of stuff where the limitation of the product is so obvious because the interface is so good. I'm just going to talk to this thing and it's going to talk to me. And then everything beyond that is like,

house of cards. And I just, yeah. Okay. Maybe we can make a more beautiful humane pin and we won't call it a pin. Maybe we can make earbuds that talk to you like her, which Altman really has been obsessed with for a long time. What's it going to do? Yeah. That makes it valuable. Yeah. I do. I do come back to that Google bike demo. I think the idea is just always on AI that can be your girlfriend, boyfriend, or your assistant, um, on the device itself. Ming-Chi Kuo, who's a pretty reputable analyst. He, he covers Apple, um,

He's saying that mass production for this is expected to start in 2027. The current prototype is slightly larger than the AI pin with a form factor as compact and elegant as an iPod shuffle. And then one of the intended use cases is wearing the device around the neck. It will have cameras and microphones for environment detection, no display. It's expected to connect to smartphones and PCs.

So, yeah, it's one of these just always on listening devices, but made from brushed aluminum. Yeah. And Eli, to your point, this is everybody's idea, right? Like this is this is the thing everyone is is pushing towards. I have I have four of them on my desk right now that are like a little thing with a microphone in it that you wear on a lanyard that goes around with you. And I think in a lot of ways, that is actually probably the correct definition.

for this kind of stuff right now. If the goal is just like lots and lots of AI input, especially if this thing is going to have a camera, which I think it has to, to come even remotely close to what

Sam and Johnny are talking about here. Sure. Like a little thing on a lanyard makes a lot of sense. I will say I went back this morning and watched the intro to the original iPod shuffle, which was from 2005. And Steve Jobs like shows off the thing, does a great reveal, the whole thing. And then they have this giant set of promotional photos of people wearing it around a lanyard. Like Apple really wanted to make this the thing that you would wear it. They had a white lanyard. It was like the cap for the USB port was also the cap for the lanyard.

it looks so stupid. And just looking back, I'm like, I don't think I ever once saw a person in the wild wearing an iPod shuffle around their neck. The shuffle became like a pocket device. Uh, and the problem with like the humane pin, maybe not above all else, but one real problem with the humane pin was that it sat right here. It was a thing you had to look at all the time. And that was looking at you all the time. And, uh,

This is very similar to that. I think you have to do it if you want to have a camera, but that's just a high bar to clear for any of these kinds of products. Yeah. Also, anything that size does not have a battery or a chip in it that can do any of the things that they want it to do. I will say, I think there are a lot of ways to do this, and one of them is just to offload a ton of that onto your phone and just do all the work over Bluetooth and Bluetooth.

local Wi-Fi and not worry so much about it, which is like, that's what Meta's glasses do. Humane didn't do it and it was a huge mistake. And I suspect anyone who is doing this going forward will learn that lesson of like, this thing actually isn't ready to be a fully self-contained device. But I think if you just take it as like a pure input system and do all of the work on your phone,

I can sort of see how you'd get a battery that could last. So you're selling 100 million Johnny Ive designed microphones. And cameras. That require an iPhone to use. You know what it makes me think of? Do you remember the Google Clips camera? Yeah. It's going to be that with a microphone. I mean, there's a couple of players in there who have a vested interest in not letting Johnny Ive supplant their products. One of them is Apple. They're stumbling around lately, but...

I think after this video in which Johnny, I've referred to his own products as legacy devices that are 10 years old and should be replaced are not going to show up and be like, yes, you can have preferred Bluetooth access to it. You're always on chat. Like good luck. No watchmaker in history has gotten that access to iOS. And maybe, you know, maybe governments around the world will shatter Apple into a thousand pieces and let this happen. But you've now made a $1 trillion bet on that. So good luck.

Well, the deep irony of all this is that is the exact thing that OpenAI and all these other companies are desperate to get around. They're like, we have to build the next platform so that not everything is, you know, intermediated by your phone and these two operating system makers. And yet, they all are going to have to be completely intermediated by your phone, unnoticed.

Otherwise, the hardware is going to be bad. Right. And then the other one is Android, which is more open and would allow many, many more of these things to happen. But Google is, as we'll talk about in the next segment, a pretty ferocious competitor to open AI when it comes to AI products. And then your market in the United States is people with Android phones. Good luck, Johnny. Like, there's just a problem there, like, to solve. Like, either all the compute is in the device, which is what Meta is pushing towards.

And I mean, we've heard Zuck. I think Zuck has said it to you, Alex. Like he's furious that Apple won't give him access to the Bluetooth stack on the iPhone. Yeah. And the glasses are worse on the iPhone than on Android. I don't think Altman's going to get that deal. Again, I'm just like, what are the limitations? Johnny has great success and maybe he did a better job of it when Steve Jobs was there than after. But his great success is designing to the limitations, making the limitations feel like they're not there because they're so apparent.

In having covered Apple for all of that time, right? In like gadget blog days where we would obsess, we will write thousands of words about screen radius and an LCD screens. Like we did it. And we, I, we are immersed in the culture that that developed, but it's now the design culture everywhere. Part of the problem for I've like in a big way is every product company has a design culture. That is his design culture, right? Like why is that a problem?

Cause he has to do something new. He has to stand out. Like his moves are, you know, um, uh, when I was like learning to play the guitar, my guitar teacher was trying to teach me how to play guns and roses and Van Halen. And I was like, this stuff is whatever. It's a garbage. It's like, it's so cliche. And he was like, that's cause they invented it. And now everything is this. That's Johnny. Johnny Ives, Eddie Van Halen. But you listen to like Van Halen. Now you're like, this is good. This is some campy stuff. And at the time it was like a revolution. It's because it's everything.

Yeah. Like it's the foundation for everything. And like, he's got, he's got to reinvent some moves here. And I just, I'm just, I'm focused on, there are some real limitations to these products and making the limitations, the strengths, maybe he'll do it. I'm not a great designer, right? He's the legendary designer, but what's, we've seen a lot of shots.

Yeah, I am going to remain cautiously optimistic about this just because of the fact that I do think a lot of people would talk to ChatGPT in a device like this if it worked. If it was a good-looking version of the Humane Pen that actually worked, that didn't take five minutes to load, that didn't also blow up after five minutes because the battery doesn't work. If all of that is fixed, which, yes, is a huge if,

And you get access to the leading ChatGPT models and it syncs with your ChatGPT account that, by the way, already knows everything about you because you are saying thank you to it even when you don't need to. That's interesting. And one of the things that Altman said in this meeting that really stuck out to me was he said, we both got excited about the idea that if you subscribe to ChatGPT, we should just mail you new computers.

Sure. And he actually said something else at a VC conference I caught a couple weeks ago where he said the goal for open AI is to now be the core AI subscription for your life. So my prediction, and this is an informed prediction, is that ChachyP's subscribers will just get these devices.

or they'll get them heavily subsidized. And he's going to build a subscription model around not just software, but hardware, which everyone has tried. Apple has tried it. Didn't really work. I mean, they have the installment program, but they wanted to bundle everything more. And that program never got off the ground. We'll see if they can make it work. Do you know why that program never got off the ground? Because the cell carriers wouldn't let them get it off the ground. Ah, there you go. The cell carriers want that recurring revenue. Yeah.

And that's another thing is like, do they need cell carriers for this? Right. How does that change things? TBD? Probably. Yeah. A different set of gatekeepers who are even worse to deal with than Apple. Yeah. Question of how ambitious this thing sets out to be. I think it's going to be just as interesting as anything else because like,

Alex, to your point, there are a lot of pieces of the puzzle here. And OpenAI and Johnny Ive together control more of those pieces than anybody else has, right? Like in terms of being able to actually make the whole stack work together, that's a lot of the stack. But it's not all of the stack. And a lot of the stack is either out of their control or will require like honest to God like physics breakthroughs that

no one has made yet. If there's no display, the physics breakthroughs are probably not as bad as we think. I mean, they probably have disassembled Humane and figured out exactly everything they did wrong and are going to learn from it. And Sam saw that as like a, I'm going to invest in this thing to see how it doesn't work. And I'm going to do it the right way with Johnny. We'll see. I mean, Humane was not

full of dumb people. No. Like the cautionary tale of humane is it's like, like one of the things Conan O'Brien always says is it takes a lot of really smart people working really hard to make a bad movie. Yeah. And like the humane pin is, is the perfect version of that to me. It's like a lot of people doing their best who are very smart and very capable and have really long track records who made a horrible product. Yeah. And there's like, that's such an easy way to go.

Yeah, we should not beat up on Humane too much more. I just am saying like- Disagree. No, I think you should read the Jesse statement again, actually. But like if by the end of next year, there's an option in ChatGPT to get this, you know, Johnny Ive Pokedex mailed to your house for an extra $20 a month in your pro subscription, and it will automatically hook into everything ChatGPT already knows about you.

And it's an always on listening device for talking to chat GPT. Do I think people will want that? Yeah. I think a lot of people would want that. I,

I totally agree. We'll see. If you're Sam Altman, you've looked at how many people happily ponied up $200 a month just to use your cool stuff. And I think the subscription is whatever they want it to be for this kind of stuff. I think it is where they're headed. All right, we should wrap this up. I will say that I can already hear the people furiously writing us tweets and emails about the iPhone upgrade program, which currently exists. What we are talking about

Just to be 100% clear is that Apple wanted to bundle getting a new iPhone every year into Apple One, and they were not able to do that. Also, subscription hardware is a bad idea, and everyone should feel bad about doing it. Do you remember when Logitech said a forever mouse and we laughed at them? This is a forever mouse. Sam Altman's Forever Mouse sounds like an incredible movie. I just swear I would go watch whatever that is.

That's a stage spectacular that I would absolutely go watch. I don't know what it is. It feels like one of those places in London that's next to the Wax Museum. Sam Altman's Forever Mouse. Ask ChatGPT to render that and email that to us instead of your angry emails about the iPhone. There we go. I beg of you. All right, we've got to take a break. And we're going to talk about the other side of I.O., which is actually Google I.O. See what I did there? We'll be right back. This podcast is supported by Google.

Hi, I'm Dave, one of the product leads on Google Gemini. We just launched Gemini Canvas. It's my new go-to for real-time collaboration with Gemini. Write docs, edit code, get feedback, iterate, all in one new interactive space from a blank slate to a built-out prototype. My favorite part? Ask Gemini to leave feedback and suggestions just like you would with a teammate. Check it out for free at gemini.google.com. ♪

All right, we're back. We got to talk about Google I.O., which if not for Johnny Ive and Sam Altman literally trying to upstage Google I.O. with their own company called I.O.

would have easily been the only thing we talked about in the first cast today because so much happened there. I do feel a little bad. We allowed OpenAI and Sam Altman to get away with this by putting it first in the show. They did it again. That was you. By the way, I want to be 100% clear. That was David's decision. Yeah, I did that and I stand by it and I am deeply annoyed that I'm giving Johnny and Sam what they want.

Uh, it was, it was interesting timing because Google had a great IO, like by all accounts. I think they announced a bunch of stuff that people were excited about. The demos are all really good. They announced a bunch of stuff that is shipping soon. Like we got to try a bunch of this stuff, including the glasses, which are, you know, they're just more prototype AR glasses, but they were there and we, people got to try them. There's a video with V wearing them. Uh, and they Googled just everyone.

from Sundar on down, just full of confidence. David, run us through the big news and let's get into it. Oh boy. Okay. So I think...

The overarching story of Google I.O. is Google is absolutely convinced it's winning at AI. And especially I talked to a bunch of the folks over there ahead of I.O., including Demis Hassabis, who runs DeepMind and all their AI stuff. And they are like really feeling themselves with Gemini 2.5 and are convinced that not only are they ahead, but they might be.

ahead in a way that is hard for others to quickly catch up with, which I think is fascinating. And so now Google is like, okay, we have what we perceive to be the best model. We have vast distribution everywhere on the internet, and we are going to put AI in front of you in every single imaginable way, all the time, everywhere. So there were a bunch of things. There were new models, new imagined VO models to make images and video. Uh,

which were like not super impressive. I don't know how they felt in the crowd at shoreline, but like watching some of those on YouTube, it was like, this is like not as cool as you think it is. You're talking about the video stuff. Yeah. Oh no. I disagree. Some of the video stuff was cool, but the part where they demoed the, like the, the woman, the filmmaker who made a video about herself being born and it was called Ancestra. And I was like, this is a horror movie. Do you know that you made a horror movie?

It's Darren Aronofsky. Yeah. I agree that what they showed on stage was not as compelling, but the stuff I've seen on like X that people have been making with this is mind blowing. It's like full on reality is over stuff. Yeah. No, some of it is very good. So anyway, so there are a lot of like new things coming to existing products, sort of improvements inside of Gmail. But I would say the

biggest things we heard a lot about were AI mode, which is the new tab inside of search that is basically just a full AI experience next to traditional search in a way that is messy and complicated and we should talk about. There's also Gemini is now being baked into Chrome, which is fascinating in the context of everything else going on with Google, which may or may not have to sell Chrome at some point in the very near future. And

The Gemini app continues to get better. So Google has set it up so that there's Project Astra, which is it's like, they called it a concept car of an AI assistant. And that's where they try all their like truly wild stuff. And it was Astra where they're showing off all of the like

wild demos where it's like watching you do stuff as you live your life and then can speak up. But as stuff gets better from there, it all graduates into the Gemini app, which got a bunch of new stuff. There's new features for Gemini live. There's now like some of the live stuff is also in the search app. It's very confusing. You were saying this, like it makes sense and it does not. No, it does. None of it makes any sense. There's one part. There's one little part that makes sense, which is Google has figured out a vocabulary that,

For what things are shipping mainstream products and what stuff is weird Google ideas. Yes. So there are projects of which there are too many and all of their names are confusing because it's Google. The projects are like, here are the demos. Here's this thing. Project Mariner is the tech demo of we're going to let our agent go click around the website.

And then the verb, which David just used, is that stuff graduates to the actual products like search. And then there's like weird hybrids where search has the new AI mode tab, which is not a project, but an actual thing that rolled out to everybody in the United States. But eventually things in AI mode will graduate to the main search experience. So they've just introduced. I'm not saying it's great. I'm saying they've introduced a verb to make sense of.

of like how Google does stuff, which is instead of shipping products, they now just graduate. Right. Oh my God. Like you graduate from preschool. It's very simple guys. Project Astra is the Google lens to Gemini's AI mode. Jesus. That's true. I hate that. Yeah. It's a lot, but, but that little, once you like understand that Google needed a way to make people stop asking when the thing would become the real thing,

And they came up with the word graduate. Like some of it clicks into place. And then everyone at Google would be like, yeah, it's still pretty bad though, right? The important thing though, is that there are essentially two places that really matter to Google. One is the Gemini app and one is search. And so every other thing it's working on is at some point,

moving towards one of those two places. And the Gemini app is designed to be your assistant, right? Like that's the thing. It's trying to be what ChatGPT is for lots of people. It's competing with Claude. Like it is the all-in-one operating system for your life thing that Google is trying to do. Search is search. And AI mode and all this stuff is a big bet in a slightly different direction that it can be

what search has been for 25 years, but like much better. And I think Gemini and search are actually, they overlap in sort of how you can use them, but they are different things in Google's mind. But those are the two places that all of this is designed to end up. Gemini is the big agentic assistant

That goes and does things for you. That's the demo, right? Alex has been talking about the bike demo. If you go watch the video of that demo in particular, it's the guy's got his phone open. He's looking at his bike. He says, pull up the instruction manual. Gemini goes and does a search, gets the PDF of the bike's instruction manual. And he says, scroll to the page about whatever breaks. And it finds that by scan. Like you have to believe that Huffy has produced an OCR capable bike.

PDF, but like whatever, there's a lot of steps in there, but it scans the manual, finds a thing. You watch it literally scroll the phone in the background. It calls the bike shop and has a whole interaction on the phone with the bike shop. That's agent stuff, like big hairy agent stuff. And then search is like the vision for AI mode is that every search result page you get is a custom built application.

All the way down to like, we can build charts for you. We might actually build custom code apps. Like you ask a question about stats in the NFL and we will build a data visualization app for you on the fly and show that to you as a search result, which is a big, huge idea, but not an agent. Right. And at some point they're going to overlap. Gemini can do that. This is what I don't understand. I mean, the reason these products are separate is because Google makes all of its money from search.

And they are shipping their org chart quite literally, which is like Demis has his fiefdom of DeepMind and now the app and he's brought product into his org. So now he controls the app fully. He doesn't control search. That's a whole other org that is 10 times as large that is the most profitable software business in the history of the world. And they cannot disrupt that to the degree that they're willing to do wild stuff in Gemini.

And should these be one product? Like, does OpenAI say, no, we have a separate app for you to search with ChatGPT? No, they don't. It's just one big product. And it's actually, it's way simpler to understand. And Google's struggle, I think, was that from this IO is that, yes, they have the best models. They suck at product. They suck at making it work holistically and making it simple for people to understand.

And that's still something they're working through. That's Google. Google loves to ship its work. And it was even on like a parent on stage at IO where like different executives got to announce the same product in different ways. Yeah. Because everybody needed a bite. You know, like this stuff happens. But from the product perspective, what I actually saw was all of their big bets were

That they've taken over time and all of their data about people is now resulting in products that look like they might work, which again has been my criticism of AI for a long time. I don't know if they're actually going to work, but they have Gmail. Yeah. They just have Gmail. And so a lot of their like personal context boils down to, well, we have, we, we were just going to read your email, right? Like we know when your flights are because it's in your email. So we don't have to do anything else. Like,

What if we were just better at reading your email to you is like an incredible product for you to solve. And now Google has all this like research and the models to do it. And then you can just productize that. And I think the search piece of it. Yeah, I do think there's two different charts and that's all their money. So they can't screw with it too much. But there's a world in which you would make that decision anyway.

Because you don't want every search to be like agentic. Sure. Right. You do just want some information presented back to you without the emotional package of your agent reading your email to you. And I, I don't know where that line is. And I think it's actually useful for Google to,

To know, like on some timeline they converge, but not know when or how, and just see how it plays out. I also think if you're Google, there is very little evidence to suggest that you need to overhaul search soon. Uh,

Like, everybody is like, oh, you have to cannibalize your own product or somebody else to do it for you. Google just continues to destroy everybody at search. And everybody's like, oh, you know, Zoomers use ChatGPT and Google is just like, show us in the market share, friendos. So, they are playing games with their metrics. Yeah. Pretty flatly. Because, you know, Eddie Q was...

at the search remedies trial and he said look you don't need to do this search is already under threat searches in ios safari have gone down for the first time in 22 years and google's we talked about this google stock price dropped like that day because everyone is waiting yeah but it didn't drop right uh so sundar is on decoder you'll hear it on monday and so i asked about that directly

And we'll just play the clip. Eddie Q on the stand in the trial the other day said, search in Safari for the last month dropped for the first time in 22 years. That's a big stat. Obviously, your stock price was affected by it. There was a statement. Is that trend bearing out that the standard Google search is dropping from devices and different kinds of searches are increasing? No, look, we've been very clear. We are seeing overall query growth in search. You know, it looks...

But have you actually seen the drop in Safari? Look, we have a comprehensive view of how we look at data across the board. There can be a lot of noise in search data. But everything we see tells us we are seeing query growth, including across Apple's devices and platforms. And specifically, I think we quantified the query growth from AI overviews. And what's healthy is that the query growth

It's continuing to grow over time. So first of all, children, if your parents ask you how you did on your report card, you should just say, I have a comprehensive view of the data. That's good. That's really good. Incredible answer. That whole interview coming on Monday was a good one. But Alex, you weren't laughing because he keeps saying query growth.

Yeah. And I don't, from what I have gathered from our conversations over the past couple of days, you're, you're very skeptical of this. Well, so two things can be true. So overall queries for Google can be growing, but the growth rate can be declining. So yes, do I think that queries in Google products or across everywhere that search touches have just like literally gone negative like year over year? No. Like in aggregate, is it probably growing? Yes. Is it growing as fast? Google will not say.

And also, like, are the amount of searches done per person growing? Or are fewer people doing more searches? Right? There's a lot that they are not saying. And they're just trying to flat out say, like, broadly, query growth is growing. Because the moment you drill down into what's actually happening, maybe you learn, oh, wait, like, maybe in North America, in a key cohort for advertisers, which is like, you know, 20 to 35, a significant chunk of people are doing increasingly commercial intense searches on ChatGPT.

And that's like if you actually try to drill down into how search works, which they don't want to do, they want to just talk about it at a high level, you maybe learn stuff like that. And we saw that reaction in the stock price because I think investors are so on edge. And you can see it in the way Google just how the stock is traded. It's heavily discounted relative to its peers because there is this fear, not only of will the company be broken up, but is...

ChatGPT and these other AI products actually eating into search in a meaningful way. And so far, Google's answer is our types of queries are expanding. People are using search in deeper, new ways, which, yes, that's true. That's how these AI products work. You want to thank it and have a conversation with it. That doesn't mean that overall query volume is growing like it did. I'm just not sure there's any evidence to suggest either way, right? Like, all we know is that

people like ChatGPT, right? I mean, no, there is. I mean, there's the fact that like 500 million people use ChatGPT every week and people also use all these other AI chatbots. There was like a bunch of things that did not exist three years ago that now exist that directly...

do what Google used to do. And like, we all see it and I bet a lot of our listeners feel it like in their daily lives. You think all 500 million of those people every week are firing up ChatGPT to do web search? I think I'm getting answers from ChatGPT that I would have normally used Google for. Whether it's like searching the web or not, it could be like...

do a math equation for me it could be like what is a time zone conversion like there's a ton of things you use google for and yes chat gpt doesn't do the commercial shopping stuff especially yet though i think they will and that's when it really gets potentially scary for google

But yeah, do I and my family members and my friends, are we using chat GPT like we used to use search? Absolutely. And I think a lot of people are. There's also other search replacements. Yeah. Perplexity. Last year we talked about TikTok a bunch, right? Like people just search in TikTok and TikTok has search surfaces. Whether or not they're useful, like I think TikTok search has like dramatically declined in quality since all these other products at the scene. Yeah.

I think the point is like to connect this to there's Gemini and their search. They need search to remain competitive and vital. Yeah. And that's why you don't make your big bet on Gemini because then you are head up against chat GPT and perplexity and whatever else. Right. But you're saying, oh, actually, this thing is just really good. Like you don't need to open this other app. Like Google search is now just incredible. Yeah.

And when assistance happen, we're going to have one ready for you. The real fear for Google here, I think, is less about how sort of meaningful a competitor ChatGPT is right now and how good a brand ChatGPT is. Like the thing that no one has ever been able to touch with Google is like a sheer awareness. And ChatGPT is like running at it really fast in terms of like when I need to do something, where do I go? The answer for 20%.

20 years has been Google. And like the Omnibox in Chrome was the most important surface on the internet for two decades. And ChatGPT is pushing at that about as fast and hard as you could possibly imagine. And like ChatGPT is a terrible business and has many limitations as a product. But like, again, it is hitting that like

mainstream Kleenexization, like kind of at record speed. And I think if I'm Google, that's the thing I'm afraid of. And that's the reason, Neil, to your point, that's why you don't bet on Gemini, because your brand is Google. And so what you want everybody to think of is you want to think of Google as the thing that is like ChatGPT, not Gemini. Gemini is like off here doing other stuff. But I really think like the confidence we saw from Google this last week is because it's there.

Like they have a lot of products that are good, not just chat GPT. Right. Like, yeah. Sora is not a great video generator yet. So it doesn't exist yet. It's like, it's around. Like some people have used it sometimes.

VO3 is just like, you can just screw with it. I watched Sergey Brin try to generate his own face for a while. At least that's what I was told. Like, it's just people are using it today and it will lead to some outcomes. I think some of those outcomes might be like

negative for the film industry. Like who knows, but the products exist and the changes are going to happen around them. Some of the agentic stuff, like I just literally watched a live demo of project Mariner, like look for jobs. And it was just clicking around a remote desktop Chrome session. And you know, I have a lot of questions about that. Like why would any of those services agree to that? Like, I don't know. Is that pretty brittle? And like, should you just use some of these new technologies like MCP? Like, I don't know, maybe, but,

But it works like it's there. And the rate of improvement is,

is so fast that Google is like, oh, we're going to do this. This is definitely just happening for us. They have the best models, and that is a tremendous advantage. I was talking to someone who works in another big AI lab, and they were saying model quality impacts usage of the product more than anything. We don't think it does, but if you release a shittier model in a chatbot, people use the chatbot less. People can feel how these models react and what they want to get out of them,

in a way that maybe we just broadly discount and how we cover them. And Google is objectively leading in model quality right now. And that was like the first thing Sundar came out on stage and talked about was like, we're number one on all the leaderboards. I'm like, that is true. But is everyone talking about Gemini? Is Gemini like the thing I hear when I go home for like Memorial Day weekend? No, no one even knows Gemini has an app. Exactly.

except like early adopters and certain, you know, pockets like chat GPT is the Kleenex of all this. And that's a problem for Google. Uh, unless we hit AGI like Demis once and none of this matters anyway. Yeah. And Demis, the only one left standing just openly being like, I'm doing AGI. Yeah. That's what this is all for. I'm on stage with Sergey Brin being like the AGI moment is coming. We're, we're taking bets on whether it happens before or after 2030. And they were six months on like either side of the line. Yeah.

You know, like every other company has stopped talking about it because they have to make some products and make some money to justify all this huge investment. And Google's like, yeah, we did it already. You know, I think Google maybe last year, particularly two years ago, was incredibly insecure about having done all of the research investment.

Like we've been joking for years now that if you say chat GPT, a Google comms person like leaps out of a bush and is like the T in chat GPT was invented at Google. Like they're over that. Right. They're like, okay, we're, we caught up. Like we're going to ship products that are good. But I think the interesting part of that, Alex, is maybe chat GPT has the brand name. Google has the distribution. Yeah. Right. So they're loud about AI overviews being the most single, most popular generative AI product at scale.

Hundreds of millions of people.

But this race, this perception of the race, this perception of how far ahead ChatGPT is and just consumer mindshare, it's an ego problem for them. It's maybe a long-term strategy problem. And I think it just hits them more in that realm. I don't think it actually – it's clearly not hitting them on the business. And that's what you and I were feeling being on site. But like –

Is it still a problem for them and that they are not winning at it all? Yes. I was with a Google exec who's been there forever at one point yesterday. And they were like, yeah, you know, just like a thing we really have a problem with is like not being first, you know, and that's like every company. But I think Google for so long just had this perch that was untouchable. Right. They just ran the Internet.

And now they're getting challenged on all sides. And I think that just, that, that messes with the, with the ego a little bit in a way that makes it seeps into the products. It seeps into the decisions they make. But yes, is the business maybe doing better than we thought? Yeah. I mean, I thought it was crazy. Like at the very end of the keynote, Sundar mentions Waymo for like 30 seconds. It's like,

They may have a trillion-dollar company in Waymo that has solved self-driving and is now doing more rides in San Francisco than Lyft. That is happening in one part of Google. YouTube is unshakable. YouTube is probably going to eat the entire entertainment industry. So they have all these products that are just huge and dominant, and I think they were feeling that. Yeah, at the end of the keynote, Sunar just casually mentioned their satellite constellations and text fires. Yeah.

And he literally was like, it will be fully operational by year's end. And I was like, that's how they announced the second Death Star. Those are the words they used for the Death Star, dude. And he was like, have a good I.O., everybody. And I laughed. Also, I think they're feeling really great because Dieter is more famous than Giannis. That's true. Dieter got the single biggest cheer at the I.O. keynote. Shout out to Dieter. It was very fun seeing him.

as a popular Google executive on that stage. Uh, also when he came out, uh, we were, we were all sitting next to each other live blogging and V just goes, Dieter. Like it's like totally quiet, right? Like she just screams, Dieter. Incredible moment. It was very good. Uh,

There's the distribution, right? That we have AI overviews, everyone's using them. Is it popular? Does the brand work? On some timeline, it doesn't matter because they're just there and they get really, really good. And then you just, you know, the next new person never even thinks to get chat GPT. And that feels like a big part of this bet, right? That's where AI mode really comes to play, where you've got this big new search experience. And over time, things will graduate to the main search experience. And then maybe no one's ever using it.

chat GPT because the people who have yet to experience AI are just like, whatever, Google just does this. Like I'm looking at this other, you want me to use this app that's doing the thing that's already happening? It could be like stories. It could be like how meta copied Snapchat and yeah, Snapchat got big and it's still big, but it never became a multi-trillion dollar company because a company with tremendous scale just copied it quickly and leveraged its distribution. Yeah, that definitely could happen. This is where Google shipping its org chart

I think is hurting it pretty badly because the thing I can't figure out and that no one at Google will give me an honest answer to is do they care about Gemini

as a public-facing brand? Is it important to Google that we know and care about Gemini? Yes. It's funny they wouldn't tell you that. I think it's because they know they're nowhere near where they need to be, and they don't want to say that on the record. I mean, they all say yes, but I don't believe them. No, it's actually Sundar told employees at the beginning there that is the top goal for the company is that Gemini wins the chatbot race. They need people to think of Gemini as... I believe that, but I think if it were...

If Google were a different company where it was...

easier to succeed by making things good and not by launching new things, would it have made a lot more sense to just do all of this work inside of AI mode from the very beginning? Yes. And you go back to how can we make google.com the most important thing on the internet again forever? And you just you pull this thing from a tab slowly into search. Like the thing I think they're doing with AI mode makes a lot of sense. And the fact that it's not

called Gemini doesn't make any sense. And so I'm just lost on this thing about like Google is actually trying to point you in so many places that I think it is risking like diluting the actually interesting work that it's doing. It's because it's the money searches the money. They can't they can't do this like they can't disrupt. This is classic innovators dilemma. This is what open AI is attacking head on. Yeah, is that Google cannot change its business fast enough to

For the consumer awareness and just attention sucking that chat GPT is doing right now.

And yeah, that is exactly what it should be. Gemini should be searched. There should be no distinction. The problem is, is that Google isn't a company. It's a combination of like 14 companies. It's their own CEOs, you know, and yeah, should Westeros like run at all? Probably like would things run smoother? Yes. But there's a lot of Lords and they all have their own. I mean, I even think about like there's, there's a, there's the Google app. That was a seamless Game of Thrones reference. That was really good. But like there's a Google app in which you can either do Google search,

or with one button, go to Gemini. And then there's also a Gemini app. And in no sane world do both of those things need to exist. - No, I cannot believe I am the person who is saying this. I don't think any of this shit matters. I really don't. - You might not. - Like I think the main search experience is gonna get better. People are gonna see the words AI mode and like click it 'cause it just says AI mode. And that is a clickier thing than the word Gemini, which means nothing.

Then they're going to be like, wow, people are using AI mode. And that's going to become the main thing. And at some point, they will do the thing that Google is uniquely capable of doing and say, click this link to download the Gemini app. And they will just get Gemini distribution. And that will all just...

Like they're they are uniquely capable of doing that. One of the and yes, they got the innovators still my problem. I can't burn down the most lucrative business in favor of the cheap disruptive thing. That's almost good enough, but works better for some people, right? Like, yeah, they got that problem. I think their other problem is.

They're basically going to torch the web. Like along the way, the web as we know it is coming to an end. And they are, you know, sooner or later I'll tell you it's not a zero-sum game, that the web is growing. They're crawling more web pages than ever. You'll hear that in Decoder next week. Every Google executive there is like more stuff is happening on the web. There are more and better applications on the web than ever.

Like the web is the place where you deployed new apps. High point, like maybe of all time. Yeah. Because of all this app store ruling and the Epic case, the web is a place where you go to buy stuff about to become incredibly important again. Right. All these apps are going to kick you out of in-app purchases to the web and we're going to all be buying stuff on the web because you don't have to pay the Apple tax. Like it's all that stuff is great. But like the web is a media platform where the new information is, it's like,

It's already under pressure from all these tools, right? It's already dying because of these tools. You can see it every day. All that new information is on, I mean, you're listening to this in your podcast player on YouTube. Like it's in these other weird platforms. It's on TikTok. And Google has to figure out how to get that information. Or it has to figure out how its agents can go and look at all those databases and make that worthwhile for people.

And I don't know if that stuff makes sense yet. I don't even know if it makes sense for OpenAI, but at least they're the upstart, right? Like if their core business fails, like Sam Altman will just let a bunch of money on fire. And that's kind of what we expected him to do. If Google can't pull it off, like a lot of things go wrong, but they have to work that curve of like,

The web is the place where the information is, is quickly getting abstracted away to the web is the database that the new AI Google search synthesizes for you. Yeah. I heard that too at IO. I heard execs saying like, no, like the content on the web is bigger than it ever has. We have more to crawl than we ever have.

And I just was thinking like, wait, isn't that all AI slop? Like, isn't the reason you have more to crawl is because your models are producing a bunch of garbage? And I really think they view data like

Like sand. Like, it's just like the beach has gotten bigger, but it's like, yeah, but there's like a bunch of dead animals in it. Like that's kind of how they view data. Look at the amount of rubble we're standing on. It's like, well, this blew up the city. Yeah. So like, yes. In aggregate, is there more data for them to crawl? Probably. How much of that is human? How much of that is high quality? We know, we feel this as people in the media business, like,

All of the best content is going behind paywalls. It's going to be hard for agents to get to this stuff. And how does that work? How do you solve the fundamental business tension of people who make the good stuff cannot support what they do on the web as it exists today? Well, actually, let's take a tiny detour into MCP because David wrote about it in the context of Microsoft Build this week. We had Kevin Scott, Microsoft CTO on Decoder this week. MCP is like a phenomenon, right? It's a standard from anthropic

that is basically here's how an agent shows up to a website or a database and interacts with it's basically just an api replacement like it's it's it's not more complicated than like we made some apis for agents to go do stuff and if you just run it all the way to the end it's kind of like oh you don't need websites anymore you just have you need off and maybe payments to databases and then the agents can just like go and api into the databases and come back and give you stuff yeah and i

I just, I don't know how the business works, but maybe you wrote about it this week. I mean, I think built into the way that you're talking about it is the assumption that everyone will access every part of the web via some chat bot, uh, which is what a bunch of these companies would like you to believe, but I don't think is true. And I don't really think has ever been true. Um,

The theory is that like, okay, what we can do in one of two ways is help plug AI into the web, right? And I think that's like MCP stands for model context protocol. And basically the idea is it's like a structured way for me to say, I have a bunch of information on my website, right?

Here is how to give the model access to that information in order to be able to do things. Fine. Microsoft is also working on this thing called NL Web that I think is very cool, which is basically an open version of a protocol like that that says, OK, what if I could actually run all of that stuff on my website?

And so now instead of going to chat GPT to talk to my website or go to Claude to talk to my website, you come to my website to talk to my website. And that is like as somebody who believes in the open web, that is a future I am much more interested in over time. But I think to Alex's point, I don't think these companies can...

No.

to care that much about what happens with news content because it doesn't matter that much to our platform. And I think most of these companies feel the same way, that actually most of this is just headache. And what people come to Google for is to like find Adele music videos and chocolate chip cookie recipes and not read politics news. And there are all kinds of terrifying downstream effects from the fact that they don't really care, but I think they don't really care. And the idea that what they're gonna do is set up

A world in which everybody who owns a website has to stop thinking about Google distribution and go build themselves into a destination people go to on purpose. And that that's going to require reprogramming 20 years of behavior of how we do the Internet. Like it's a mess and it's going to kill a lot of the Web. And I just don't think these companies care because it is there's new stuff to do. I think Google wants you to believe that it cares.

But on the longest timelines that Google thinks, it does not. So I would say it the other way. I think Google cares a lot about the web and very little about websites. Like making sure that websites that exist today get to continue as they exist today, I don't think Google cares that much about that. Oh, sure. No, we're agreeing. But I think Google cares about the web is this like really –

The web is a miracle. Like I just like straightforwardly, the web is a miracle. We have in the world, a giant interconnected interdependent, mostly open application platform. That's weird. Yeah. Right. Thank you. Government spending. You can just, yeah. Like, right. We, like we just horsepower this thing into existence. Yeah.

Because honestly, Marc Andreessen got a bunch of government funding in the middle of the country. Like, that's a thing that happened. Like, that's Mosaic. And it turned into Netscape. And like, it's all here. And now Figma exists. And like, Johnny Ive, legendary designer, probably opens a web app to do a bunch of design work.

That's crazy. Now he uses pages for everything. Yeah, it's funny you brought up Andreessen. I actually talked to OpenAI to their MCP lead this week because they joined the steering committee for MCP. So they're now working with basically every AI lab has agreed that this is the new standard. And I think it's a great way to start.

I actually came away thinking we are actually still pre-Netscape on agents as it relates to the actual plumbing that it takes to hook them together. So this open AI lead was like, there's still no registry. Like there's no discoverability mechanism for a developer to even know who has an MCP server. Like we know there's a few of them. There's like Stripe and

Zendesk or whatever, but no one's even built the interface for then the developer to go, okay, X model, go make this happen with this server. And that, that will happen. It'll probably happen this year, but we're not even at the browser point. This is funny to be talking in these words because like, yes, agents are supposed to like replace browsers, but like we are not in the browser phase of agent building yet. Like,

This has not even been connected in that way. It's still very, very early and it will be enterprise and very developer focused first. It's not like we're going to be getting some all powerful agent that can just like go and order a bunch of stuff for you and talk to a bunch of apps like,

anytime soon even though google showed that off like this is going to be a developer replacing function calling with telling a model to go talk to an api yeah and that's really interesting that's interesting yeah exactly you can see there's a new web you would build on that like yeah and that's what i mean like i agree with david i think they care a lot about the web because this big interconnected application platform where you can like instantly deliver complex applications

to a runtime like Chrome is a miracle. And I think Google is very invested in that particular miracle. I think as a media platform where you read a bunch of text,

And it's a they don't care about that at all. And like, I know this because Alex and I were sitting in the audience for the Dennis Asabas and Sergey Brin fireside chat with Alex Kantrowitz. And Kantrowitz asked, what do you think the web is in 10 years? And Sergey was like, who knows? Right. He was like, the world will hit the singularity. Like, I will be one with the sun, like off in space.

And Demis said, it's a good question. And then quote, I think the web in the near term, the web is going to change a lot if you think about an agent first web. Do the agents have to see renders and things that we need to see as humans?

Do we need to actually render web pages? Do we need websites? That's the question. And he's, I mean, his answer clearly is no. His answer is no. And then I asked Sundar a riff on this question. I read him that quote. And he was like, what is the web but a series of databases? Which, first of all, is the vergiest verge answer of all time. Yes. Like, I literally started, you're going to hear him. I just started laughing. I was like, thank you. Like, I feel very validated. Like, what are we but a series of databases? Dying out on that one for a while. Yeah.

But like if you reconceptualize the entire interconnected application environment, the web is just a bunch of databases that you can go ask for stuff. And some of those databases have Toyota's Camrys in them and some of them have vacation houses in them and some of them have sandwiches. It's like, oh, those are just businesses that are going away.

Or they're businesses that get funded through API calls instead of ads on their web pages. Like, will the incentives figure themselves out because do agents need things to do? Yes. So like, if everything that an agent would talk to suddenly goes away because its business model is destroyed, like, guess what? Agents don't work. So will this get figured out? Will it be super messy? Will a lot of companies go away? Will whole new categories of companies get invented? Yes. Right.

Yes, that's all true. But do I think like the fundamental building blocks of the plumbing of the web, like content and things that an agent could do will go away? No, because then there's no agent. And what Google and ChatGPT are both racing to be is the interface to interact with all those databases on your behalf, which is like the greatest abstraction of technology I think maybe we've ever seen like in our lifetimes if that happens at scale. And the value that will accrue to the interface is...

is arguably greater than the value that Google gets now from running ads on search. I mean, this is the funny thing. It's like, Alex, you and I have now spent a lot of time on this podcast in the last week talking about display ads. But the fundamental question here is what happens when display ads don't work anymore? What is the business of the internet without display ads for the people who have traditionally relied on display ads? Paying $250 a month for Google Ultra. Maybe that's the answer. Maybe the end of this is Google and OpenAI...

pay websites directly that i don't love that but that is one possible outcome of this but if if the web is a series of unrendered databases the whole business of the web is suddenly immediately gone ben thompson by the way wrote about this this week uh based on mcp and he was like this is what stable coins are for and like a like full body grown basically is yeah i i

Boy, I hate that. Oh, man. This is the idea that we'll have these crypto microtransactions pegged to the dollar in some way.

That make all that happen. I would just say we have, if one more person says microtransactions are the answer, I'm going to send everyone a thousand links to all the times we've tried microtransactions and it didn't work. And then I'm going to heave myself out of a window. I will, again, full body cringe in reading that stable coins are the answer, but he did make one good point. Ben Thompson made it at one particularly excellent point here in that the web already runs on thousands, if not millions of microtransactions every second, because that's how the ads work.

Yep. Yeah, but it's not me giving you... All of those banner ads are microtransactions. Sure, but... But it's there. You just hate stablecoins. I get it. I get you. Whatever those are. That's... Sure, all of that exists, but if you made me type in the IP address of every website I wanted to go to, that doesn't work either. Like, making the users do it falls apart. You're not going to need to.

Go order me a sandwich. And it's going to be like, or, or you have some subscription to one of these centralized indexes that needs to get built. And that subscription like has the weird Spotify model where it's like, you get a penny and you get a penny and Taylor Swift gets most of the pennies. And like, maybe that's how it plays out. But you can see just at IO in the conversations they're having that

The web we have today is reaching its termination point, and this new other kind of web is definitely the thing they all want to build. It's the thing Microsoft wants to build. That's why they're doing NLWeb. It's the thing Google is just sort of openly talking about.

It's what OpenAI needs to have happen to make its dreams realized. Like if you wear your Johnny Ive iPod shuffle and it can't do anything, you're kind of stuck, right? Like it needs that whole ecosystem to be developed and built and for the money to all work for that product to be as useful as they want it to be. And I think this is a marker. I think part of Google's confidence is they think they can horsepower a new existence, right?

while sort of preserving the best of the old web and bridging the revenue into the future. I will say publishers are furious about this. They are super mad. We have a statement on our site from the News Media Alliance, Disclosure, Vox Media News, and the News Media Alliance, along with Condé Nast. It's

It's the business side of the company. But we're also taking money from OpenAI. Oh, it's true. We have an OpenAI deal. It doesn't even occur to me. I also made a Netflix show. You should watch it. It's great. That's relevant, right? It's not just me bragging. Google wanted to buy Netflix. That's true. Google. There you go. Sundar announced that Google wanted to buy Netflix at one point. But the publishers are like, the whole point of this deal was us getting traffic and now that's gone. And then they called it theft.

Like that is, that is what the publisher industry, the news industry is saying about AI mode. And I think Google is like, well, it'll be sad when you're dead. Like that's kind of how it feels.

And I think that will get litigated even harder. Not to mention the fact that they might have to sell Chrome, which I don't know what will happen if they sell Chrome. I don't think they're going to have to sell Chrome. I got no vibe that anyone was worried about Chrome at IO. Well, what are they going to do? They're not going to wander around their own party being like, well, this is going to suck. No, no, no. But you know when you're actually having a drink and you're staring at someone's face a few feet away and you ask them, are you worried about Chrome? Yeah, you're trained, but you do it enough and you kind of get a vibe that

And I did this on a few topics and like, I was very curious, like, are they worried about this? Like up and down the chain. And I think they know they're going to have to stop just like,

mafia-style buying out search distribution. Like, that's going to go away. But do they think they're still the best search engine? Yeah. So I don't think they think they're going to have a divestiture. And that's another topic. I mean, on the website thing, I'm curious what you guys think about this. Are websites just going to become like driving a vintage car? It's going to be this thing you do because...

It's a luxury. It feels good. It's like a really bespoke, unique, like visceral experience. It's not as efficient, but you do it and you spend more money. Like I kind of feel like that's the direction websites are going to go. I do like thinking of theverge.com as like a cranky vintage Jaguar. Yeah. Right. It's kind of always broken, but you love it. It's going great. Yeah. I don't know. There's other kinds of web...

Like, you know, Substack and Ghost are web properties. Blue Sky, like, at Protocol, built on a lot of web ideas. And the thing about Google is that all of these websites are effectively addicted to Google's distribution. That's why SEO has polluted the web so badly. There wasn't another choice. And so, like, yeah, maybe agents are going to turn everyone into databases. Like, I think if you're a retailer, if you are Walmart or Target or Macy's or whatever, and you're looking at Google's new try-on mode, which looks very cool,

Like the promise there is that once you've tried on the clothes virtually, the demo was you set a price and then at some point Google just buys the clothes when it hits a sale. If you're a target, you just totally got disintermediated. Now you're now you're just doing high frequency trading with a Google bot. Like this is a really weird business to be in. And Google's kind of answer there is like, oh, well, we'll talk to them and make sure that we show some of their web pages sometimes.

And I said, well, what's that negotiation going to look like? And Google was like, it won't be a negotiation. It'll be a conversation. This is ice cold, right? Like there's a whole universe of businesses that are going to have to get reconfigured. And a lot of them are going to want to say, well, actually, we don't want anything to do with this. We want you to come to our website. Substack for, you know, it's many, many faults is a web property. It's expressed on the web. There's not like a Substack desktop app. And the innovation is they've solved distribution inside of their own little network.

And that is really, really lucrative for a lot of people on Substack. All of that is the optimistic case, right? The pessimistic case is, yes, websites die and everything becomes the high frequency trading. You're essentially fighting for Google scraps. The upside is for the first time in two decades, everybody's going to have to care about their website.

Because it has to be, I mean, this is, this is the verge bet, right? Like this is, this is the thing we've been talking about for a while that like the bet is we have to make something that you come to on purpose or else we can't be that upset when nobody comes. I think maybe that's a place we're going to land. And, and a lot of that's going to make things harder for a lot of the weather's going to be a lot of pain along the way, but like,

If we do this correctly, I think, and this guy Guha, who I was talking to about all of the NL web stuff, he's like, the fact that the web trends towards centralization is bad. And it has been bad several generations in a row, and we have to stop it. And the only way to stop it is to distribute the good technology evermore.

everywhere. And I think the way we think about the federated social networks, the way we think about MCP, the way we think about NL Web, bringing some of this AI stuff to websites is like, maybe there's a chance that actually this ecosystem gets bigger instead of you just going to google.com, landing on a webpage you've never heard of, and then never going back. Because that's what we've been doing, but you could argue that's not the correct outcome.

But it's like it's an outcome. I mean, what is this index of MCP servers except incredible centralization on the web that will reemerge? Right. Right. I mean, there's a reason the AI labs are the ones doing this. This is not Tim Berners-Lee being like, this is great for the Internet. This is anthropic being like, this is how we get Internet access. Yeah, because they can't. I mean, you know, their first shot at it was what if we click around your website, which is just incredibly brittle. Yeah.

It just, it's never good. That is, that is never going to work. That will be the backup plan for all of this. And MCP will be plan a we'll see. I, again, we, like we walked around Google IO and they were all feeling it. Like they are riding high. Like their theme, they kept saying was research into reality. And it's like, what they mean is you didn't believe us for a long time that we had all this stuff in the labs that we were working on it, that it was coming to fruition. And now here all like all at once, here's all of it. And, and,

You know, so it has to ship and, you know, people are, I'm confident we're going to get feedback on this part of the episode that's like, they're stealing our jobs and boiling the ocean. There are some really weird downsides to all this stuff happening. Yeah, both of those things are true, by the way. They're true. They are true. But the weirdest one is that we're staring at the beginning of a wholesale re-architecture of the internet. Yep. And that we should probably pay a little more attention to that to make sure it's good this time.

Because the last time, like four people just got to be in charge. Yeah. And like, I might take the trade off if you can actually decentralize a little bit more. We'll see. All right. We got to take a break. We're gonna come back with lightning round. We'll be right back. This podcast is supported by Google. Hi, I'm Dave. One of the product leads on Google Gemini. We just launched Gemini Canvas. It's my new go-to for real-time collaboration with Gemini. Write docs, edit code, get feedback, iterate,

all in one new interactive space, from a blank slate to a built-out prototype. My favorite part? Ask Gemini to leave feedback and suggestions, just like you would with a teammate. Check it out for free at gemini.google.com. All right, it's time for the lightning round. Unsponsored? For flavor. I said it with a question mark because I actually don't know if it's unsponsored.

There are rumors that we have a sponsor. There are rumors that we have a sponsor, but I believe it's not today. And I wouldn't even know because the whole point is no one can tell me what to do. That's why Verge subscriptions are 40% off this week. Next week, it'll say sponsored for flavor. And then we'll make new t-shirts. It's going to be Doritos. It's going to be great, you guys.

Actually, does anyone know anyone at Doritos? Listen, we should reach out. Yes, I will eat Doritos constantly through an entire first cast if Doritos would like to pay for that. It's funny. You know, our great rivals for podcasts within a podcast.

My brother, my brother and me. They were once sponsored, I believe, by Tostitos Pizza Rolls. They were. They made a whole Totino's episode and it was excellent content, if I may say so myself. We've got a lot to live up to on that front. We do. We will never do that. I just want to be clear. You can't pay us money to tell us what to do. That's where the flavor comes from. I guess in this case, this would be Brendan Carr's The Dummy Brought To by the FCC. Speaking of which. It's time, David. Let's do this one more time just for this week and then hopefully never again.

Every week I root for it to be the last version of Brendan Carr is a dummy for a lot of reasons. And every week it is not. It is time once again for 2026 Webby award winning podcast within a podcast. Brendan Carr is a dummy. Neil, you've slept for three hours. I expect this one to be very, very good. There's there's only there's only two items in this one. The first one is so stupid.

Like so, so stupid. And the other one is so disappointing. And that's really the heart of Brennan Carr is a dummy, right? How can you disappoint me in ever more stupid ways? So this week, Brennan Carr, America's number one censor, who continues to make such a splash as a would be speech cop that like more and more profiles and coverage of him emerge.

Politico wrote a profile this week. John Oliver did an entire segment that is basically Brendan Carr is a dummy. Thank you to John Oliver for validating us. And thank you everyone who sent that in. He is a real test of all press is good press. Our buddy Brendan. It's not. You're a traitorous moron, Brendan. And I know this because this week he has started to open an investigation into NBC over Kamala Harris appearing on Saturday Night Live. Uh,

disclosure NBC investor in our parent company Vox Media just putting that out there you can do it that way you will again I remind you no one can pay me to say anything it's the first subscriptions are on sale hey do you guys know who won the election do you remember who it was that doesn't matter I forget yeah do you remember who lost definitively lost and is not the president do you know who that is uh-huh it was Kamala Harris I believe that she lost

Right. She did not win. We can agree that she is not currently the president of the United States. Correct. The other guys. OK, well, Brendan here in mid-May of 2025 has decided the most pressing speech issue in the country is whether Saturday Night Live violated the equal time rule for broadcast television stations by having Kamala Harris appear on the show. Do you know what I also know definitively is that Brendan knows they didn't.

But yet he has decided here in May of 2025 to reopen an investigation which was closed into whether NBC and Saturday Night Live violated the equal time rule by having Kamala Harris and Saturday Night Live. The equal time rule is like an archaic broadcast rule. It's as if a broadcaster that uses the public airwaves to broadcast content

gives time to one candidate, they got to give equal time to the other candidate. It's a little more complicated than that, but that's right. It's called the equal time rule. Which NBC did the next day. NBC had Donald Trump on the next day. He got a bunch of time during the NASCAR race. I reported on this while it was happening, right? Because Harris showed up on SNL. There was a bunch of like MAGA tweets. And then Brendan was Brendan before he was auditioning for the jobs. He was like, I will arrest everyone when I meet the FCC commissioner. Great. I reported it out at the time.

And what I heard from sources in the FCC was the only program that regularly deals with this is Saturday Night Live. They are the best at it because politicians are always showing up on Saturday Night Live. Oh, interesting. Like they have the infrastructure for dealing with the equal time rule because no one else gives a shit. So of course they filed the equal time notice with the FCC. And of course they offer Donald Trump time the next day.

Because they're good at it. Like, this is a thing Saturday Night Live is the best at, of all of the shows on all the broadcast networks. And so, sure enough, Brendan, his emails were foiled because he was threatening this investigation before he was ever made the FCC commissioner. And he emailed Fox News producers, here's the equal time request from NBC and Saturday Night Live. I know he knows it because I'm reading his email saying he knows it.

They complied. Former FCC chair Jessica Rosenworcel looked at this and dismissed this because they had complied. And I know they complied because Trump was on the air during the NASCAR race the next day. Equal time was given.

Well, here we are, May 2025. Election has been won. We have accepted a jet from the government of Qatar. We're definitively past Saturday Night Live booking Kamala Harris having any effect on anything that happens in this country. And Brendan is running around trying to punish NBC for violating the equal time rule, finding ways legally to reopen this investigation just to put more pressure on NBC.

For its alleged coverage of Donald Trump being last. This is just speech police stuff like it's it's over like the election is over. Whatever harm you may have felt the Trump campaign felt that car felt for his his boss and buddy Donald. It's over. They they won. But like the harm has passed. Right. The opportunity for that to have swayed the election has come and gone and it didn't work.

And he knows. We can see his emails. We reported on his emails that he knows that they didn't break the rule. And yet he's trying to reopen. I mean, this is just moronic, right? Like, he's doing it to be a cop, to chill the speech of these news divisions that are reporting on Donald Trump. Maybe because NBC owns MSNBC. Even though Comcast is spinning that off into a new company called Versant, which we haven't even talked about yet. But, like, that's why he's doing it. So stupid.

So I will I will put it in the show notes. We have his emails where he's telling Fox producers they complied with the rules. And yet he's going to reopen this investigation just because he wants to be caught. He's not even trying to hide the fact that this is what it's about. It's very straightforward. Very straightforward. So that's one. That's very stupid. Here's the very disappointing one.

We've talked a lot about how Brendan threatens deals that are in the pipelines and tries to get companies to comply with whatever he wants, whether it's their coverage, which he says is biased, whether it's DEI policies, which he says are not racist enough, whatever it is. He gets what he wants by holding up deals. So last week, Verizon completed its acquisition of Frontier, another fiber provider, and it got that through the FCC by sending a letter to Brendan Carr saying it was walking back all of its DEI efforts.

Super disappointing, right? Verizon just rolled and said, we're going to give up on our DEI efforts. This is Brendan at work, right? He went and threatened the business over something that has nothing to do with the business. And Verizon caved, like just straightforwardly caved and said, this is a government regulation that we're going to deal with. Even though the FCC telling us who to hire and fire is kind of a straightforward infringement of its rights as a business.

And the thing that is particularly disappointing to me is that whenever the FCC has tried to do something good to Verizon, they have fought it tooth and nail. They're the primary legal antagonist of net neutrality for a decade. Every court case, every challenge, Verizon was there. Whenever you talk about increasing broadband access, Verizon fights that stuff tooth and nail. Verizon takes money and promises Fios will be rolled out.

In New Jersey, they just never do it. And then they get sued and they fight it tooth and nail. We had the CEO of Verizon Wireless on Decoder. I don't know why enterprise software CEOs or telecom CEOs show up on the show, but they came on the show and I asked him.

Are you going to fight Brandon Carr in these weird regulations as hard as you have fought net neutrality? Just run the clip. I want to be very clear, very explicit about this. When the government passed net neutrality rules, it wasn't we have to follow the rules of the land. It was we are going to file lawsuits for a decade.

to get out of these rules because we think they're dumb. And in this case, you're saying Brendan Carr, who has been openly censorious, openly chilling of speech, openly hostile to companies because they have diversity initiatives. You're saying you just have to follow his rules. We have to follow the rules of the land. I don't think those are his rules. Those are rules of the land and administration. Are you going to file a decade worth of lawsuits about these rules?

We don't know. We're going to work constructively with them to follow rules that are needed. But at the end of the day, look, our role is to our stakeholders that we have. You know, my stakeholders are my shareholders, my customers, my employees and society at large. We have to manage and we are going to deliver for those stakeholders what's needed for us. And we will do whatever is needed with the administration to deliver to all the stakeholders. It's a balance when you run a large company our size. You have to balance the different stakeholders and we will balance those stakeholders. We've done it extensively.

extremely well in the last 25 years that we've been Verizon and we'll continue to do that going forward. You got to balance the stakeholders, guys. We have a comprehensive view and we have balancing the stakeholders and you can get out of almost anything. That sucks. That is 60 seconds of him being like, when he says those aren't Brendan Carr's rules, those are the rules of the land. That sucks. I mean, that's straightforwardly a lie. Those are Brendan Carr's rules. He said, I'm going to hold up this deal

Unless you walk back your D.I. initiatives and Verizon, instead of fighting, which they always do when it comes to consumer protection or broadband deployment, they always fight here. They roll and it's it worked. And the thing that kills me is it doesn't have to work. Like if you stand up to these bullies in particular, they tend to roll over. And Brendan getting a win is just like the most disappointing outcome. And it's particularly funny because it's Verizon always fights.

Verizon looks at government regulation and is like, no, no, no, no, no. One decade. And here they rolled. And I just think it sucks. I'm incredibly disappointed that it worked in this way. I'm also, you know, not for nothing. It's another merger of ISPs and it will result not in greater service and lower prices.

You know how it's going to go. We all know how this is going to go.

The timing of this is so suspicious to me because it really seems like AT&T looked at this and said, oh, all we have to do is write a letter about DEI and this will get done. Yep. And I'm like, I hope that's not how this goes, but it sure seems like AT&T is about to just run the same playbook in order to get this thing done. I think there's a renewed sense that some mergers are allowed to happen. You know, there was a lot of excitement.

Even in the tech industry, deals are back. Biden and Lena Kahn were not going to let any deals happen, and now the deals are back. And that actually wasn't the case early on. Andrew Ferguson, the new chair of the FTC, is kind of a trust buster. He kind of hates big companies. He hates big tech.

is just like, give me an ounce of blood. You know, just bloviating away moronically. Like these deals weren't happening. And I think there's people are starting to figure out like there's a path forward and it involves trading your values to get the deals. Right. And over and over they're doing it. They're so happy to do it. I think the next one we're going to see is CBS and

And Paramount settling their case with Donald Trump over 60 minutes to get the Skydance deal done. And that will be a dark and sad day for American journalism. Yep. To the point where it's like, oh, that's over. You shouldn't trust that network anymore. That's brutal. As always, Brendan, if you're listening and I super know you are, you can come on the show.

You know, I got your press release today. I wrote back to your people and invited you on. He put out a press release about how much more spectrum he's opening for satellite networks. Great. He loves it. It's his favorite thing to do. Congrats, Brendan. You can come on. You can defend this stuff. The fanciest CEOs in the industry, telecom CEOs, come on Decoder and answer the questions.

Because they can defend themselves. Sampath, the CEO of Verizon Wireless, he gave maybe the single best answer to the decision-making question on a code we've ever had. Even though I completely disagree with we have to manage for our stakeholders. They can defend themselves. They're smart. I think you're stupid. So you're not going to show up. But if you want to, you can. We're available. You can be on this show. You can be on Decoder.

But for now, that has been Brendan Carr's dummy, America's favorite podcast within a podcast. That's beautiful. I've decided to take a comprehensive view to Brendan Carr. I'll get back to you on what that looks like. I have a comprehensive view of the data. Can I offer you some breaking news as a palate cleanser? Mark Gurman.

from Bloomberg, just published a big story about Apple with two very interesting pieces of information that I think are germane to what we've been talking about. Thing number one is he reported that Apple has been working on smart glasses and has like really ramped up its effort to sell smart glasses and hopes to do so by the end of next year. And he cites one person, presumably inside of Apple, who said they will be similar to the Meta Ray-Bans, but better made. Yeah.

which is terrific. So yeah, presumably he called Tim Cook and Tim told him that and here we are. So that's piece of information. Number one is that Apple smart glasses, which we have been hearing for a while, we're sort of being pushed further and further off, may have been pulled back up in the roadmap. And the other one is that

We've been hearing for a while about the idea of an Apple Watch with an integrated camera, and that is apparently shut down. And that will no longer be a thing, which is a shame because I wanted to know very badly how any of that was going to work in the real world. But...

This is Apple pushing on the same kind of stuff, right? Like multimodal AI assistance is the name of the game here. And Apple seems to be on the smart glasses trend along with, at this point, kind of everybody other than OpenAI and Johnny Ive. Yeah. And I can tell you why they're doing it. It's because these smart glasses are hitting decent scale. They're doing single digit million sales a year. The meta ray bands are huge.

They may get to double digit millions. They're also like good and people like them, which I think is an important piece of this. Oh, I think everyone is super confused. Most of all Meta about why people like the Ray-Ban glasses. Oh, that's true. But they are good and people do like them. Yeah, people are buying them. They are. But what people like is having glasses with a camera in them. Sure. And a shutter button. They don't like Meta AI. But that's also if you're Apple, that's a core competency, right? Like that's a reason to do it if you're Apple. But that's not... I don't think they're an AI distribution platform the way that Meta wants them to be. No. But...

But they don't have a display in them yet. And that's what everyone else is demoing. Apple smart glasses, if they don't have a display, they're one kind of product. If they do have a display, they're a very different kind of product. And no one has cracked that display problem yet. At least about Google. The one we saw, the Android XR...

it is so a prototype it is you you saw these right i'm gonna talk to v a bunch about all this stuff on tuesday in the glasses but i am curious you you got to try them right yeah just really quickly it was like a five minute demo and it was basically gemini on your face um the weird thing about these prototypes is the wave guide is very small and not high resolution but it's in the dead center of the lens whereas i was expecting like a heads up to the side thing like glassy kind of like because it's literally yeah it's literally just a gemini icon and then it

reads back what it's saying to it and what you're saying which is like I don't I don't need that the best thing was the map stuff where like you could turn your head down and it would show where you were your pin would move like it had GPS and tracking and who knows maybe the room was like set up for that but and then the other thing was like oh you when you take a picture with it you see the picture like you can see how you're framing it because it shows it to

it to you whereas the meta ray-bans obviously you have to take your phone out to see what you shot that's cool the hardware is not super compelling yet the software is very bare bones I think they will actually start shipping these next year and they got Warby Parker and Kering which makes Cartier and like

Gucci glasses and they got another cool the name escapes me another cool glasses company out of Asia that V was very excited about they got them on board so basically Meta has hitched its wagon to elsewhere like Sotica which makes Ray-Ban and Oakley and a bunch of other brands and Google tried to get that deal I reported that they tried very hard and they failed and

And now they're going for basically every other eyewear partner that they can to get Android as the platform instead of meta. So I think we're going to see this AI platform, smart glasses battle between

Meta and Google with their partnerships strategy. Gemini is a better AI. Meta has a head start. We'll see how it shakes out. The demo though at IO was like kind of meh. I do want to quickly say though, a demo that was awesome and that I'm very excited about and we have a great video about that I did with Viren is Beam.

which was Project Starline and is basically Google's answer to what if virtual meetings suck less? Starline has been one of those ideas that Google has had that seems to get cooler every 18 months, but also...

and less likely to ever be a real thing. It was really cool. They invited Vera and I to the lab where they made it. We were the first ones in their labs, hardware, prototype room externally and got to see how they moved all the tech into basically what is... It started as like it filled a room five years ago and now it's basically a DVD player and everything is streamed through Google Cloud. So they built this proprietary volumetric 3D AI model that makes you look like a hologram basically through a 2D, very fancy...

I tried to get them, Nilay, to really explain the panel. It's a highly customized panel that's doing a bunch of fancy pixel work, but I couldn't get a lot of detail out of it. But it's a reference design now that HP will actually start deploying into offices later this year. So they've got Salesforce, Deloitte, some other companies that have committed to installs. And I asked, I was like, is this expensive? Because companies will not buy this if it's not cheap.

cheap as their existing video conferencing software. And they're like, nope, that's why we did the AI work. It's all in the cloud and it costs about the same as the video conferencing setups you see in offices today. So I think there's a real chance we start seeing Beam and people who work in offices start seeing Beam over the next couple of years. And it is wild. Like go watch the video. It's on our YouTube. Viren got to try it for the first time. I had done it a couple of times because we had it at Code and then it was at IO a few years ago when it was Starline.

And Viren tried it and it's just this like, holy shit moment. Every time someone tries it for the first time. I mean, Nila, you remember. I do. And...

Yeah, it's cool. It's like one of these classic Google things. Like it reminds me of Wham-O. They just toil away as like with the resources that only a company like Google can to invent this wild new thing. And then five to 10 years in, they're like, it's ready. And it's like, now the cars are driving themselves. Now we have holograms that we're like putting into Deloitte offices. Like it's just happening. So that was pretty cool. Yeah. My understanding of the screen, by the way, is the reason they're not talking about it in detail is...

It is more standard lenticular display than not. I think there is a lot of custom stuff going on there, but usually when you have the big tech innovation, you talk about it a lot. Yeah. And the reason they're like, it's a secret, is like, it's kind of a 3D display. Like, all the actual happening is in the camera, but eventually these are going to be out there and we can go look at them and figure it out. Yeah. But it is very exciting to see that out. On the glasses...

The display is not solved. Meta Orion, they spent billions of dollars on the wrong technology and it didn't scale to make those displays, right? Turns out growing crystals doesn't work at scale. We're just crystal farmers out here, guys. I don't know. I mean, I don't know who's going to build those displays. I don't know where they're going to come from. No one's cracked it yet. And it's just one of these things. What is the core technology that's going to enable this next generation of products?

You need the displays to make the smart glasses. Or you don't. Or you're Johnny Ivan. You make a fancy aluminum Pokedex and that's like...

All right. We are way over. I'm so sorry to everyone for the rambles today. I mean, what did we expect this week? It's the day after now. We got a lot on the site. Tons and tons of coverage. V is going to be back next week to talk more about the glasses demo. She just published a story trying to guess what Johnny was building. So way more coverage on the Verge cast from her coming up. Sundar on Monday. Get ready.

It's not as tense as it was last year, but there are some moments. I think you're going to like it. All right. That's it. That's VergeCast. Bye. And that's it for the VergeCast this week. And hey, we'd love to hear from you. Give us a call at 866-VERGE-11. The VergeCast is a production of The Verge and the Vox Media Podcast Network. Our show is produced by Will Poore, Eric Gomez, and Brandon Kiefer. And that's it. We'll see you next week. This podcast is supported by Google.

Hi, I'm Dave, one of the product leads on Google Gemini. We just launched Gemini Canvas. It's my new go-to for real-time collaboration with Gemini. Write docs, edit code, get feedback, iterate, all in one new interactive space from a blank slate to a built-out prototype. My favorite part? Ask Gemini to leave feedback and suggestions just like you would with a teammate. Check it out for free at gemini.google.com.