Hello and welcome to the Vergecast, the flagship podcast of trade secrets. I'm your friend David Pierce. Nilay Patel is here. Hello, sir.
How are you? That was tired. You okay?
I just had this wave of like, I spent the early part of my career at The Verge explaining patents to everyone through Apple, and we're like 15 years in, and I know I'm going to spend the next part of this journey explaining trade secrets to people. It's come. We're going to do it.
Yep. But that realization had not occurred to me until just this second, and we're going to emotionally process that together as a family over the next 90 minutes.
You really had a literal visceral reaction to the phrase trade secrets just then. That was really great.
We did copyright. We did patents.
We're still doing copyright. I don't know if we actually did copyright. We tried to do copyright really hard.
We've reopened copyright in a meaningful way.
Yeah, we're certainly doing it again. Lots to talk about today. We have a bunch of Apple news. We have Apple versus OpenAI, which happened last week, but we haven't really had a chance to talk about it. Lots to get into there. We have a bunch of gadgets. There's some OnePlus stuff. There's teasers just everywhere for everything. But I think we should start with the Apple betas which came out this week. I am desperate to know, you as America's most steadfast avoider of macOS Tahoe, have you upgraded? Are we on Golden Gate now?
It takes a lot for me to upgrade my Macs. They're both production machines. I should upgrade the MacBook Neo, which I bought for no reason. I will do that. I just literally don't ever use that computer. I own it for no reason.
That's what it is. It's a Tamagotchi that I have. It is very charming. Every now and again, I have to charge it. You know, so the Macs are hard for me because I don't like screwing with the production setups, especially this Mac, the podcast Mac. We just don't touch that one. I did update the phone, which logically makes no sense because the phone is the thing you take everywhere with you in case of emergency.
I was going to say, not an important device, your smartphone.
Yeah, I should have upgraded the Neo first, but I updated the phone. I'm on the Siri beta. It's good. I think some of the animations in iOS 27 can get toned down. There's a lot of flipping around the screen for no reason. I know there are...
Siri in particular, it's always doing stuff. The new Siri looks at me like all the time.
It's just like not what I need. Apple always tones that stuff down as they get towards release, so take that with a grain of salt. But it works. And I actually think the screen reading is interesting. I don't know if it's good yet. I know every single person with an Android phone is going to tell you that Gemini could do this forever and it's vastly more useful, and it is. I agree with all of those things. I just, Apple was so far behind, and then with the power of Google, they have managed to catch up in some meaningful way.
Yep. Have you been using it?
I have. I've been using it since the first developer beta, which I do not generally recommend. I've had a really interesting experience with Siri AI in particular, which is that it has felt sort of like all of these new LLM-based smart home systems in that the ceiling of what they can do has gone way up, right? Like, David Pierce wrote a great piece for us about the new Siri and did the thing where I think his example was basically like, I'm looking at a Spotify page that I know they're going on tour. Go figure out when they're going on tour and add the ticket sales date to my calendar. And that's like a multi-step AI process that I think is ripped basically directly from a Google demo, which is very funny. But it works. That thing works. The ceiling of Siri has gone way up.
It kind of worked for him. He had to say magic words in the right order. David's piece is great, by the way. Also, shout out. David's joining us for like six weeks or so while Allison is out on leave. But I encourage everyone to read that piece because it is true. The ceiling is higher. If you issue the right incantation at the right time and the system understands it in the right way, you can in fact trigger a multi-step agentic workflow.
Yes. Neat. And then if you don't, it's like, I don't know what you're talking about. Right. So ceiling's super high, but basic things that Siri used to do very reliably, so far I have found some of them to be actually substantially less reliable. So like a thing I do all the time, if I'm out walking the dog or in the car or whatever and I have a reminder or just something I want to write down in such a way that I will get back to it, I use Apple Reminders for that all the time. It's just like my inbox of stuff on my phone. So I'm constantly saying like, you know, remind me in three hours to switch the laundry, though. I did this morning. Remind me this afternoon to check the dryer vent. Like that's just a thing that happened this morning because I need to check the dryer vents. And that is worse. It is noticeably worse than it was in the past because, and we've talked about this many times, right? Like it used to be all of these systems were essentially a series of really elaborate if-then statements. And they didn't work for anything except the things that they worked for. But the things that they worked for, they worked for. And so you learn to understand the boundaries of the thing. And if it was all I needed to do is turn on my lights, it would turn on my lights. It wouldn't do anything else, but it would turn on the lights. And I think we're doing a little bit of throwing that stuff out in order to do more complicated stuff with Siri also. And so I've found like, it still seems to set timers fine. That so far has gone very well. But it does things like constantly ask me, 'Did you say this or did you say this?' and they haven't yet solved the it needs to start immediately listening again to get your answer. So I have to dig my phone out and press the button again to make it work. So there's just like all these little things. All the basic interactions of Siri are worse, but the abilities of Siri are vastly better.
Maybe. I mean, hit you with the maybe. And we're in the beta and it's like not fair to overjudge it, and sure. We'll see what happens in a few months when it comes out for real. That replacement of the predictable deterministic system with the vastly potentially more powerful stochastic parrot system, right?
Yeah, you can see those edges everywhere all the time. I'll give an example from a totally different product category. We have a Cadillac Vistiq, a Cadillac EV, and it had Google Assistant in it. Like the whole thing runs Google built in. There's no CarPlay, which the only thing that happens in the Facebook group for this car is someone being like, 'Where's CarPlay?' The only thing people talk about. So they replaced Google Assistant with Gemini. And okay, here's my favorite example of this. I could probably write a PhD thesis about this. Okay. Specific button in the car. It doesn't have a physical garage door open and close button. There's not one in the headliner, wherever. It's a button on the screen. So it's in the car software, right? You push the button on the screen and it opens and closes the garage door. With Google Assistant, I could just say out loud as we are backing out of the driveway, 'Hey Google, close the garage door.' And this is gold. Until you've been so lazy that as you are literally backing out of the driveway, you are no longer pressing a button. You're just saying, 'Close the garage.' You're like, 'I'm the king of the world. Like, I have a butler now.' Like, whatever that feeling is. Amazing. And I have asked executives at GM. I've asked executives at Google. My Google just went off. The garage is already closed. Google...
That took a full like 10 seconds, too. We're doing great over here, guys. Anyway, so I've asked executives at Google. I've asked executives at GM. I've asked Deere. What's happening when I say close the garage door?
Is that Google Assistant going to the cloud? By the way, I asked Cer just to annoy him. He's like, we don't talk about working. But is that going to Google Assistant in the cloud because Google Home knows about the garage door opener? It's in that. Or is this thing locally pushing the button on the screen? Because what it should do is just push the button on the screen.
Mhm. Okay. They upgraded to Gemini, which theoretically has like an MCP server inside the car, right? Like when you would tell Gemini to change the climate control, it literally pops up a bubble that's like connecting to climate system, like...
It wants you to believe that it's like operating the car.
Now Gemini is like, I don't have permission to do that. It's not secure enough. And I'm like, dude, I'm in the car.
I am the security here. Like, we're good.
Like here I am. I can see the button.
Anybody in this car can push that button. They can just reach forward and push that button.
But it's definitely going to the cloud and it's saying I cannot accept a cloud command to open. And that's just the brittleness. Like we've replaced a system that should be able to do all the things you expected it to do with a newly architected system that is simulating what it should be able to do or what it used to be able to do.
But actually needs to find all kinds of new pathways to do it. And Siri, the garage opener example is really stupid because I just can push the button. At the end of the day, it doesn't matter. But that experience and then having the new Siri for a few days is exactly the same experience.
Right? The AI wants to tell you they can do anything at all. And they might have the capability, but in reality, the barriers and roadblocks to actually executing are all over the place.
Yeah, I think that's right. And in a funny way, it is not all that different from the very beginning early experience people had with Siri like a generation ago where there's a set of things that you're like, oh my god, I can't believe it could do that. It's like sincerely moments of that is incredible. And Siri has done for me some personal context stuff where it knows what's on my phone and it can find things in my email and it can find things in like, I can be like, oh, what's the door code for the Airbnb I'm staying at this weekend? And it just goes and finds that and tells it to me. And that is incredible. Like that's a literal thing I did and it worked and it's awesome. But for every one of those, I say, 'Remind me to check the dryer vents.' And it goes, 'Did you say call the Laura friends?' And I'm like, 'No, I didn't. I super didn't, Siri.' Like, what are we doing? And so I think the risk here as always with Siri is going to be that people run into the edges without understanding what those edges are, right? Which is like when all these things first came out, it was pretty easy to understand what it couldn't and couldn't do very quickly. And I actually think in a certain way, the answer to what is this thing good and bad at is vastly more complicated now. In part because it's good at so many more things.
It's good at so many more things, but it's not predictably good at anything.
Right. And that's just a weird one. Like there are examples of this just again across ecosystems where like search my, I'm late turning in the draft of the Decoder book. So I ask Siri like go find this contract and see how late I am and how much trouble I'm in. And it like did it, you know, like the answer is I'm not under arrest yet, but everyone is very worried about the timeline. We're gonna get there. Siri didn't say that, by the way. It was just like you're late. It's good at that when it can be good at that, right? When there's like a string of text it can go look for or it's indexed all the stuff in the right way. And then there's, I think Apple's going to run into this much more than Google does. A bunch of Google AI technology doing weird stuff is par for the course. Apple shipping this Siri to millions upon millions of people, hundreds of millions of people and it being rough around the edges is a danger zone. And I think that's probably why they waited so long.
Yeah. But I'm very curious to see how that broad response is because enough of it is different. And you mentioned, you know, a generation ago with Alexa, people very quickly figured out what it was good at and what it wasn't good at and they stopped trying to do anything else, right? And I have no idea where that boundary is for this Siri once, you know, hundreds of millions of people have it. Yeah, it's going to be really interesting and especially given all of the work that's happening with app intents and stuff like, I think more than usual, the fate of Siri AI is going to actually be about developers because if Instagram can figure out how to plug into Siri in a way that is predictable and useful in that way, that changes things. I don't know that Instagram will ever do that.
Instagram doesn't have an iPad app.
Yeah, that's okay. Fair point. But if Uber can figure out how to do it in a way that's practical and useful such that I can say, you know, hey Siri, get me an Uber and it works semi-reliably, like you can start to turn those things on for people in a way that becomes habit forming really fast. So I'm actually still optimistic about what Siri AI could be, but it is, I have been frustrated in the ways that it has actually backtracked in some small ways.
Yeah, David's again, David Pierce's piece is great. The thing that jumped out to me, I actually saw David because we talked about it for a minute. App developers on iOS are more excited about this than you would expect because small app developers know that people never open their apps.
So their hope is that if by integrating with Siri AI, it will surface their apps more often, they will get more usage and something good will happen.
That's a really interesting point. That's a good theory. I hope that pans out. A lot of optimism in the ecosystem there.
Indeed. You know, where there's not optimism. How's that for a segue? In the Apple and OpenAI legal teams these days. We should talk about this lawsuit, right? It's a week old. I still feel like we need to talk about it because I'm still wrapping my head around it. Basically, the short version of this lawsuit is Apple sued OpenAI for allegedly basically creating an elaborate system by which Apple employees would quit their job and take material confidential information to OpenAI. There are a couple of specific people named in the suit. There are a couple of OpenAI adjacent companies like IO, Jony Ive's hardware startup, named in the suit, but the idea is essentially there are a bunch of people and a bunch of ways in which hundreds of Apple employees stole information from Apple and took it to OpenAI. I feel like the caveat here is this is the first filing in a lawsuit. There will be presumably much time and much discovery over the next period of time, but you and I have both read this lawsuit. What do you make of what Apple is alleging here?
This is where I offer you the big sigh and we talk about trade secret law. The first thing that is absolutely important to say. Every single one of these big tech companies has realized that the complaint in the lawsuit is a thing to be written for the general public. The audience for this complaint is not the judge. It's not the jury. It's you and me and everybody listening and everybody who might read it and everybody who might write a blog post or do a dramatic reenactment on TikTok. They have all figured out this first step that the complaint is a narrative and they all need to sell a narrative. The Department of Justice and the Biden administration figured out their antitrust complaint against Apple needed to be a narrative. And we have talked to those officials about that specific thing in that specific way because everybody understands that the technical requirements of the law are irrelevant to the vibes around lawsuits. It's just the way it goes. So Apple wrote this to be read. It is worth reading. It's a fun read.
Yeah. They lay out how a heist was taking place in specific with a bunch of characters. Trang Tan I think is the main character. Mark Gurman at Bloomberg has some reporting. It says even at Apple he was known for playing fast and loose and the quote is flying too close to the sun.
Which is pretty good. There's other folks in there. There's a guy you're like literally connected to the servers when he had left Apple and was downloading all this stuff. And then the problem is the story can be whatever the story is. And it is certain that OpenAI will craft a compelling narrative in return when they file their reply. And it'll look something like Apple is late to AI and to be late and catch up. They're trying to stop us, the winners. Like you know what they're going to say and they're going to write it all out. So we have to wait for that. And then in the middle of that is the actual technical requirements of the law. And I'll just make another OpenAI comparison. We had Musk versus Altman and that stretched for what, nine months from the first complaints being filed to a jury trial and then what happened? The statute of limitations had most of the case thrown out. Right. Like the technical requirements of the law is actually what won that case for OpenAI in like big meaningful ways. There's other parts of it where they won on the substance, but it was a statute of limitations got most of the case thrown out. Okay, like this is, we just have to put ourselves in this framework.
Sure. The thing that is just like particularly funny to me about this is Apple established the boundaries of copyright law in tech against Microsoft in the 90s and 2000s with like the famous look and feel lawsuit where they tried to copyright the way Mac OS looked and they sued Microsoft for having Windows copy it. They sort of lost and they sort of won but they mostly lost, right? Like Windows looks like the Mac.
That's the way it goes. And when Steve Jobs came back to Apple, Microsoft paid a bunch of money to Apple and they tossed out all these claims. They said, 'We're settling this and moving on because this is dumb.' Okay. Well, a bunch of years later, Samsung made a phone that looked an awful lot like the iPhones. And Apple filed a bunch of patent lawsuits against Samsung, against Motorola. Everyone except Google, by the way. They did not want to be in a fight with Google over this stuff. They filed the lawsuits against the OEMs and they fought those to the death and at one point they were kind of losing and Tim Cook was at a Code Conference and I literally asked him like from the audience, there's video of me asking him why are you doing this and he said I think the patent licensing schemes for 3G standards are a little out of whack and this is how we're going to fix them. And I'm like you're really doing this to fix patent licensing schemes? And he was like, 'Sure.' And it was just like very obviously that like...
No, he was just mad and that was cover to fight the fights, right?
And they fought them to a bitter end, right? The patent on slide to unlock. We again, we made an entire video of this. You can watch me being a baby at MWC. I walked around to every booth at Mobile World Congress in Barcelona and did slide to unlock on every single Android phone and said, 'This is how they're getting around the slide to unlock patent.'
Yep. And there was technical requirements there and there are technical requirements now. But the ecosystem responded. Samsung paid what? A billion dollars and essentially got away with it. And if you're Samsung, paying a billion dollars to make the Galaxy line of smartphones is totally worth it.
Correct. Totally worth it. So Apple filed this massive intellectual property lawsuit. They upset the ecosystem for however long they upset the ecosystem. And the end result was there are Android phones. I don't know if you've looked around. There's quite a lot of them.
Yeah, that turned out okay.
So, okay. So in the 90s, we had a massive intellectual property lawsuit against Microsoft based on copyright. I will point out Windows and Microsoft continue to exist. In the 2000s, 2010s, we had a massive intellectual property lawsuit against Samsung based on patent. Seems worth noting that Samsung and Android continue to exist.
Yeah, sure. And here we are in the mid-2020s. We have a massive intellectual property lawsuit against OpenAI based in trade secret. Now, the third one, it seems a safe bet that AI will continue to exist.
Mhm. I would be 50/50 on whether OpenAI continues to exist.
That was going to be my next question. I think it's obvious this is not existentially terrifying for the AI business, boy does it feel like of all of the companies to not want to be at war with at this moment, I mean, frankly, OpenAI is kind of at war with everybody else in some way already, all the time.
Yeah, I mean this is, OpenAI is permanently making hundred billion dollar deals and also in giant fights with everybody all the time.
OpenAI keeps a war on its own executives. They keep leaving.
Yeah, but this is just, this is the last thing OpenAI needs. And I will say I think one thing that is different about what you just described is Apple has picked a lot of fights about products. This is fun. Like so far, unless OpenAI is very close to shipping a thing that looks an awful lot like the iPhone, this is not about products.
It's about what manufacturing techniques and what Apple has in the pipeline.
Sure, it's about products to come. It's about how to make products. But I just mean even if you paint all of this for Apple in the most sort of negative light and you're like, 'Okay, this company feels threatened and they throw a bunch of their lawyers at their threats to try and make them go away in one way or another.' This doesn't feel exactly like that to me. I don't get the sense that Apple is threatened by OpenAI and there's certainly no earth-shatteringly fabulous AI device that is sitting there threatening Apple at this moment unless again there's something currently being manufactured that is going to blow everybody's minds and we just don't know about it yet.
Well, I think that's the whole game here, right? Apple knows that AI might be an input mechanism that creates a new class of products. Certainly Jony Ive went to go work with OpenAI to create a new class of products based on this new input mechanism. Jony Ive loves the new input mechanism. He loves the click wheel, right? He loves the multi-touch, the scroll wheel on the Apple Watch, as we all know, revolutionized the entire world. Everybody loves spinning in a circle on the Apple TV remote. We, I mean the culture really rotated around that one. But Jony like loves this. So if you're Apple, you're like, 'Okay, he's hired all of our designers. He's hired a bunch of our best manufacturing people. It's like 400, 500 people that they've taken over there. They know what's in our pipeline and they know our suppliers. They know how we do things. And now they're trying to steal our stuff. We have evidence at least one of them is trying to steal our stuff, right? We have evidence that at least one of them is trying to steal our stuff. Okay. What are we going to do about that?' And you can't file a copyright lawsuit because we don't know what they've taken. It might not be code and they certainly haven't redeployed the code. It isn't a patent because there are no products yet. So there's no patent infringement. To infringe a patent, you have to make, use or sell something, right? So there, we haven't done anything yet. There's no patent infringement. You're left with trade secrets. And you know, trade secrets are pretty fuzzy. The definition of a trade secret is like, I'm real fuzzy on this. Someone check me with AI, but the trade secret as far as I remember is it has to be something with economic value because your competitors don't have it, right? It can't be known. It can't be like public information. And then the really fuzzy one, and I'm going to end up having to do so much trade secret... This is where the big sigh came from. I'm going to do so much trade secret research. You have to take reasonable measures to keep it a secret.
This is all over the lawsuit, by the way. Like those three things are all over the law.
Those are the three elements of a trade secret.
Yeah. A huge amount of this complaint against OpenAI is Apple explaining all of the things it does to try to protect it because it's written for the general public, right? So it has to explain the technical foundation of the law so it can say and OpenAI went around this stuff. And again, Apple pushed copyright law to the limit with Microsoft. It pushed patent law to the limit with Samsung. We're about to be in the middle of Apple pushing trade secret law to the limit. If you worked at Apple and you know how they make the iPhone screen and then you get another job, you can probably help people make an iPhone screen. And particularly if you work not for Apple but you work for Foxconn in China, there might be some supplier agreement, but I don't know if you've looked around. It turns out Foxconn makes things that look like the iPhone screen for lots of other people now because that is the entire Chinese manufacturing ecosystem and supply chain. And there's books about it, right? Apple and China, is that what Patrick McGee... there's literally books about it. Yep. You can go read it and you're like, 'This is just going to happen to Apple whether they sue OpenAI or not.' So again, the lawsuit is great to read. I just look at Apple being like, 'They're going to bring everything to bear to push the cutting edge of trade secret law in their direction.' And other big companies that made a lot of money were able to essentially fight them to a draw.
OpenAI is not a big company. It issues a new memo about needing to focus and calling a code red every six weeks on the dot when another executive leaves.
Yes. And its revenue model is totally uncertain.
And up until now, the revenue model has been we're going to compete with Google and search in the consumer market. And we're going to put out a bunch of new devices to compete with Apple by hiring Jony Ive, the guy who made Apple devices what they are, to compete with Apple in the consumer space. And Apple's like, 'What if we killed you?' So you really think it's that simple? This is Apple essentially trying to kill OpenAI in the cradle.
I think, well I think Apple hates anything being stolen. I think the things aren't being stolen by Foxconn or an ecosystem of Chinese companies. They're being stolen by an American company that poached its most well-known design executive who then poached a bunch of other stuff. And it is like deep in the DNA of that company to have an emotional response to that. Like over and over, if you take some information out of Apple Park, that is maybe the thing that is most likely to quickly bring the full bear of that company down upon you.
Yeah. I mean they are absolutely nuts about security. I left my jacket at Apple Park at WWDC in one part of the thing and I wanted to just walk from where I was back to the seat I was in and they're like we're going to have to send four people with you and I was like it's a path, like I can just see the path. And I just left the jacket there. I was like, 'This is too much production for me to find a $15 denim jacket I bought 10 years ago.' And I was like, I just don't want to do this. But that's where Apple is. Just to put it in that framework, the merits of the case are not in this complaint. The merits of the case won't be in OpenAI's reply. It'll be in this long process we're about to go through.
There's also a lot of reaction so far that has said this in trade secrets terms is a relatively benign set of trade secrets accusations. But is Apple going to turn the heat up all the way and get themselves in front of a jury and be like, 'We're at Apple, the company you love. Here's OpenAI, a company in the AI industry, which all of you hate, and which is foundationally built on unauthorized taking of information, and Apple's the company where artists love to show off how they use Macs to make the songs you've always loved the most. Hey, did you know this music video was shot entirely on iPhone?' Like, you know what's going to happen here?
Yeah. Apple is going to bring the full weight of its cultural capital to bear on OpenAI as a proxy for how much stuff the AI industry has taken from people without compensation or permission. And whether or not it's a benign set of trade secrets or whether it's the full catalog of YouTube music on behalf of Suno, OpenAI is going to stand for all of it in this lawsuit.
Interesting. So would you handicap this as likely to go to ground in a real way? Like is this the kind of thing somebody writes a check and it all goes away or is Apple gunning for we want to do this in public?
If OpenAI had a lick of sense it would write a check and make this go away.
But does Apple want a check? Does Apple need a check?
Does Apple trust OpenAI to keep the promises it's making as it hands over said check? No. So and I think OpenAI does not want to write that. They want to prove that they did nothing wrong because probably they also have the same view of this which is this is just the normal course of business.
Right. We hired a bunch of people and said what are you working on? What do you know? Do you know how to make an AI gadget? Because we don't. Right. And they're going to want to fight this to the end.
I mean to be fair, do you know how to make an AI gadget? Bring us parts from the office to your job interview? Not a normal course of doing business if that is in fact what happened. Look, I've been running a small AI hardware startup on the side and I'm not saying I didn't try it.
Listen, it's good getting if you can get away with it. You know what I mean?
Rather unfortunately, I've only talked to people who used to work at Humane. So, bit of a problem. But yeah, I think they're going to take this one. This is years and it's years and then years of appeal and whether OpenAI is stable enough or has the finances enough to just absorb it the way that Samsung could just absorb it and then pay a billion dollars and be like that was totally worth it. Wide open question.
Yeah. This one I suspect we will hear a lot more. We have not as you said meaningfully heard a response from OpenAI. We got a comment from OpenAI spokesperson Drew Puseteri last week who said quote, 'We have no interest in other companies' trade secrets. We remain focused on building innovative technology that empowers people everywhere,' which is just the least you can possibly say in a statement to The Verge about this particular issue.
Yeah. Especially when the claims and the complaint are like, 'Dude downloaded all the presentations about how to make circuit boards,' right? And it's like, well, maybe you did have an interest.
You might have some interest.
Yeah. We're going to come back to this a lot, but for now, let's move on because we should talk about gadgets. There's actually an OpenAI gadget worth talking about. So, let's get to that. But first, let's take a break.
This is advertiser content from IBM. Cold take. When it comes to quantum computing, if you need a million qubits to calculate...
Anything, it sounds like you need a quantum computer the size of a football field. So, why are we even trying?
My name is Olivia Lanes. I am the global lead for content and education at IBM Quantum.
So you will not need a quantum computer the size of a football field. The quantum chip is really, you know, the size of a stamp. So the chip itself is never going to be the size of a football field. What could be a very large footprint is all of the infrastructure that goes around cooling it, housing the chip, and controlling it. IBM has really been, I think, the biggest pioneer in terms of making this technology accessible. IBM has deployed at this point over 90 quantum computers worldwide. So if we talk about scaling, I will just say we have a plan. We have what it takes engineering and science-wise to make these computers work and to build them up to the scale that we know they will be useful. If you would like to learn more, you can do so at ibm.com/quantum.
All right, we're back. We have just a slew of gadget leaks to talk about, but let's start with this OpenAI gadget, right? So this idea of OpenAI's big hardware play has been looming for a while. They released a device this week for Codex that everybody's like, 'This is the first OpenAI.' It's not. This is like a teeny tiny brand integration with a company called Work Louder that makes keyboards. I don't want to talk about that. But there was reporting from Bloomberg that the thing coming out first is going to be a no-screen but sensor-and-camera-having smart speaker. That this is going to be OpenAI's big first device, a ChatGPT thing for presumably your home, but it'll also have a rechargeable battery, smart home controls. This thing seems so incredibly unfocused to me. It's like Alexa, but you can pick it up and carry it around and it has a camera on it, but no screen.
I just want everyone to stop and imagine a team at OpenAI building a full-featured smart home platform.
Listen, Matter has solved all of this for us. It's no problem.
I just, what, just like what are you doing, man? Google hasn't built a full-featured home platform.
Yeah, they've been trying for years. I don't know if anyone has ever looked at the Google Home app. Uh, sure. I don't even... That thing doesn't even look vibe-coded. A computer would have more coherent and crisp ideas about how to organize that app than whatever it is that Google ships in Google Home.
I just, I fine. I think based on the constraints of the market, this is the only piece of hardware OpenAI could ship.
Just run down the Meta glasses problem that we run down every week. If you want to have an AI device that is consistently doing things for you that isn't a phone, and I think for a lot of reasons, OpenAI cannot make a phone.
I think it's going to... it's going to happen. I just want to be on record. I think it will make a thing that is, they may not call it a phone, but it will very obviously be a phone.
A lot of people are about to make phones without Instagram. That's, I'm, I'll put... Yeah, that's where my stake is. A lot of people are going to back their way into making phones without Instagram and then realize that they've shipped Windows Phone, and then we're all going to see how long it takes for them to realize they shipped Windows Phone.
Um, but if you don't want to ship a phone, you're like another form factor. You're competing with the fact that people have a phone.
You have to be smaller than a phone, which means you need less battery, but you still need connectivity, which means you still need a data plan, which means you still need to run that battery down. You still need to decide what happens locally or if your device is useless without a connection of any kind, which means you need to put processing in RAM and whatever. Your costs have already begun to skyrocket and now you're OnePlus and you've gone out of business in the United States. It's just anything that moves is hard for that reason, especially if you want to put cameras and sensors and all that other stuff on. So, of course, we're going to take it home, plug it into the wall, absolutely reduce the challenges of mobility, let whatever processor run as hot as it might need to run so it can do whatever it needs to do locally, and have a fixed connection over your Wi-Fi to OpenAI servers, which at least now is solved.
Right. The connectivity issue is solved. If your Wi-Fi goes down, everyone knows nothing works, right? But I'm on a plane. My device is useless. Unacceptable.
All right. Now, we've made an Alexa, right? We've made an Alexa device or a HomePod or whatever. And then you've got to build the ecosystem around it that does all the other stuff around it, including the smart home ecosystem because of course you want to turn off the lights and play music, which are the number one and number two things people do on these devices. Oh, set a timer, and unclear if ChatGPT can reliably count to 100, let alone set a timer that operates consistently. So, here we go. Yeah, like this is the only solution. And whether or not the ChatGPT platform is being built to work in this environment, in these constraints, and solve these problems, or whether they are totally focused on writing code and revamping the entire Mac app to be focused on writing code and whatever healthcare initiative they're doing. This is a lot of spaghetti to throw at the wall.
Yes. And there's no chance that any OpenAI engineer has been tasked with building a full-featured smart home platform that can compete with Apple Home, Google Home, Alexa, and Home Assistant, which will be every nerd's fallback for this and is happy about it, right? Imagine you're an engineer and you're like, 'I work at the hottest startup on Earth,' and they're like, 'Make the lights turn on and off. Uh, go deal with the Matter implementation.' Like, I quit. Like, I'm going to go work at Anthropic like everybody else.
I think this existing next to what they've been doing to the ChatGPT app is super fascinating to me because the new ChatGPT app, the so-called super app, is trash. Everybody hates it. It's poorly designed. It doesn't run well. It took a native app and made it a non-native app. Like, it sucks. It is universally disliked as an attempt to put all of these things together. Yet again, OpenAI demonstrating it has lots of ability to make very good frontier models and absolutely no product sense. Like this company just cannot figure out how to put this technology in front of people in a way that they like. The only thing it has done is copy Claude a couple of times. Claude Code worked and they said, 'Oh, we'll do Codex.' And that worked. Like, there's just not a lot of evidence that this company knows how to figure out these new kinds of things that you're talking about. But the thing that is so interesting about that app is it kind of deprecates chat. It is very Codex-centric, the new ChatGPT app. And the idea that they're leaning into, okay, we think our technology is for tools, right? This is not a thing to hang out with. We're getting past the 'it's novel to talk to a chatbot' phase and that just burns money anyway. Let's give enterprises enterprise tools. That is clearly the business for this company. That's the business for all of these companies at least for now and possibly forever. So then to put that next to this,
the only thing I can think of is like we talk a lot about New York versus California, right? I think if you are in or near the OpenAI offices, it might be very possible to convince yourself that a lot of people do just want to sit and talk to their agents all day. That like this is a thing that is running rampant around startups all over Silicon Valley. The idea of that being something mainstream enough to be a really sort of meaningful hardware play seems dead wrong to me.
Yeah. It's just, it's the same as Siri. It's brittle.
Yeah. So if your whole idea is we'll make a beautiful piece of hardware and it will have great far-field microphones and whatever we're going to do with the battery to make it portable around your house and then the core technology in the back end is ChatGPT is there, well, everyone already knows how that thing is going to work. Like, I don't have to describe it to you. It's going to be broken in very specific ways. And then when you're like, 'Why'd you do that?' it's like, 'Oh my god, you're so right,' you know, like...
That's a great point. Yeah, right. Like you can just put your phone on the table and have this experience today.
Yes. We have these things, like everybody who is like, 'Oh, it's an Alexa.' We have Alexas. People have this experience. We have a decade of this experience. And people play music and they set timers. And I think what even Google and Amazon have run into over the last 12 months is the idea that we're going to plug an LLM into it and suddenly unlock vast new kinds of consumer behavior is not happening.
It's just not happening. It's actually slightly worse at turning on the lights. It can answer more complicated questions. That's great. So as a sort of simple information retrieval tool, great. Fine. ChatGPT is not going to win that particular game. It's also less sophisticated in its hardware than Google because Google's been at this a longer time. So like what race is there to win unless you assume there is this giant group of people out there who desperately want an easier way to talk to ChatGPT all day. And I just don't think that exists.
Well, they do, but they think it's their girlfriend. Kevin Roose, come on down. I'm sorry, buddy.
We are friends. I'm just... You know, he knows how I tell the joke. Um, I mean, maybe everyone's going to think we're being overly cynical, and I'm confident we'll get those notes from people who think we're being overly cynical. That cynicism is probably born of the fact that we started The Verge at the dawn of the smartphone revolution. And the questions about this new class of devices are exactly the same questions about new smartphone platforms and new operating systems, but maybe with slightly different applications. So, when the JooJoo tablet came out, oh god,
Remember the JooJoo tablet?
No one remembers this stuff. It ran a thing called Fusion OS and it was going to be a revolution. And boy, did that cause a lot of problems for us as gadget reviewers. Uh, including one threatened lawsuit. Um, and we would say, are you going to build an email client? Is your email client going to be as good as Gmail? Right? Because I'm not going to switch away from Gmail as my email provider, right? And Google made native apps for the other phones. Is your email client going to be as good as Google's native client? Because they're not going to ship a client for your dumb idea. And then these products would die. And you would run into, are you going to build an email client? It suddenly becomes this like defining question for a new generation of smartphones. Okay? Are you gonna have Instagram? Hate it or love it. Not having Instagram killed Windows Phone. People would get those phones and they would say, 'There's this app I want and it wouldn't be there.' Yeah. And Microsoft spent a lot of money on app developers. They tried to bribe what at the time was known as Facebook into making an Instagram client. They just didn't do it.
They had no interest in this. And then nobody wanted a Windows Phone. Palm Pre, like down the line, right? Like all these things killed new smartphone platforms over and over and over again because they could not meet the table stakes. I'm telling you for this class of devices, can you turn the lights on in the kitchen is going to be table stakes. Can you do... Do you have a Spotify app? Do you have a Spotify integration that reliably works?
And if I don't have Spotify, do you have Apple Music? And if I don't have Apple Music, do you have YouTube Music? And if I am an insane person, do you have Tidal? Are you playing Tidal losslessly? I have a Sonos system. Can you play Tidal over my Sonos system because you're an all-powerful AGI robot and all these devices are on my network and you have my credentials and I expect you to just do it. And the answer in every case is going to be no one at OpenAI wants to work on these problems.
Oh, I don't think they're stupid. I don't think anybody wants to work on these problems. I think they want to work on true AGI. I think they want to work on healthcare. I like...
Sure. But it is also true that some of those things already exist. MCP solves some of those problems. Some of ChatGPT's own, you know, integrations solve that problem. Like, will your ChatGPT smart speaker be able to connect to Spotify? I'm so confident the answer to that will be yes. Will it do a good job?
I'm not confident Google Home can connect to Spotify.
So this is what I was going to say. If you connect to Spotify, is Spotify going to want that to be ChatGPT talking to the Spotify database or is it going to be ChatGPT talking to the Spotify chatbot that talks to Spotify? Like, is this product going to work well?
And is any of that going to generate enough revenue to be worthwhile to OpenAI, a company that is lighting money on fire every single day and has no plans to stop, right?
I mean, this is my thing, is I think on an ongoing basis, ChatGPT technology continues to get better. Like GPT Live, the new voice model which will presumably be behind this, is quite good. It takes turns more naturally. It's more natural to talk to. It doesn't sort of interrupt you and let you interrupt it in weird ways. Like it feels better to talk to. I think the idea that that is enough just falls on its face as soon as you put this device in my house for all the reasons that you're talking about because it needs to do things. These things... It's like I feel like all I talk about with AI is how many things give great demos and are bad products. And this is the kind of thing that, yes, there are people who just enjoy talking to ChatGPT as an end unto itself. I am on record as thinking that is bad for everyone and the world and society, but I understand that those people exist and those people will probably get great utility out of having a thing they can just sort of yammer away to as they walk around the living room. Fine. That is tiny next to the number of people who need this thing to do something and they will need it in fact to do lots of things and they will expect it to do lots of things. And I just, I'm with you in the sense that I have a hard time seeing how that happens.
And consumers have no economic incentive to overcome failure.
Right. Regular people in their homes, they ask Alexa to do something, they can't do it, and they will never ask it again.
They will give up. And they did. Like we have the data. We have the experience. They just gave up on Alexa in meaningful ways outside of timers and music.
Okay. In the enterprise context, you keep saying AI is business software. In the enterprise context, there's tons of incentive to overcome failure. Uh, we were watching a TikTok clip today. Like ad agencies are like, we're going to automate all of our Meta ad spend with Claude MCP servers.
And like maybe that didn't work for a while. It certainly works now.
And that is true. Like the internet is automated in massive ways with AI. Businesses are getting automated because the incentive to overcome the failure and the rocky starts and they keep plugging at it. Everyone can see the efficiency gain. Everyone can see the business results. There's a lot of incentive to overcome the failure. A regular person is like, I spent how much money on this weird OpenAI HomePod and it doesn't work. I'm just never going to use it again.
Right. And like, they've got to... This is the Johnny problem. It's got to do exactly what it says on the tin and a little bit more in a way that's delightful. And AI is maybe never going to be predictable enough to actually pull that off.
And the physical limitations of something like a smart speaker makes that hard. Like genuine kudos to everyone involved if they can figure out how to add that hardware turn that makes it delightful in the way that you're talking about.
Oh, it's supposed to have some mechanical thing that like smiles at you. This is the other part of the German report.
That ain't it, buddy. I got to tell you, it has some mechanical... It's unclear what it is, but there's some part of it that moves in response to you to like make it have emotions.
Yeah, GPT doesn't have emotions. Never. Ever. Ever. Ever.
Hold. A lot of people disagree with you on that one, my friend.
They're wrong. Let's move on.
Claude loves me. Uh, what are you talking about? All right. No one else does, but Claude does.
It thinks all of my ideas are great, and I've been, you know, you're right to call that out. I've been looking at this the wrong way. Let's start from the basics. Honestly, here's what I recommend. All right, let's move on. That's enough AI gadgets for one day.
Wait, can I say one more thing? I had Claude write an email to my solar panel installers about the thing that went wrong, and I read it, and I was like, they will come to my house and tear the solar panels down before they address what in this email.
Yep, that sounds great. Uh, yeah, that's sometime we should talk about the AI tools that write email drafts and the deeply terrifying things they have taught me about myself. But that's for another day. Um, two more gadget things we should talk about. One very quick. Have you looked at the Pixel colors? The Pixels, as they always do, have leaked to absolute death. Um, there's a bunch of new Pixel phone colors in particular. You, I feel like, have never really been a phone color guy.
No, I've got an orange iPhone. It's just in a gray case.
Well, yeah. You're like, I feel like in general you're...
This is the only one worth buying this year.
Oh, 100%. I mean, I bought the green one, which is also terrific, but the orange, the orange on the Pro was fabulous. Uh, but I just want you to look at these Pixel colors and tell me if they do anything for you.
Oh, I like this pink a lot.
Yeah, Google's really going for it this year. There's a pink. There are a couple of shades of green.
It's pink. Uh, there's a green, there's a black, there's a light blue that's, or like a greenish blue that's called Fog that I quite like.
I can't say... You should say all these names. There's Olive, there's Fog, Frost, Canyon, which is also known as Dune.
I wrote a story recently about the Google Home speaker. Um, and I got a request from a comms person at Google who, in this particular case, I won't name, who asked me to change the reference where I called it red to its official name, which is Berry. And I said, no thank you, I will not do that, uh, because it's red. Berry is not a color. Red is a color.
Berries do come in a variety of colors.
Yeah, they sure do. Uh, but stupid names aside, these colors look great.
Like I'm into this. I think a pink phone genuinely might really have it. I'm psyched about this.
This is finally the thing that will make me switch to a Pixel 11. This gives... Was it the Nexus 5 that came in that rad red color?
Yeah, it's like orangey-red. We have so many arguments in the office about whether it was orange or red.
And if I remember correctly, Josh Topolsky was very much 'it's red' and I was very much 'it's orange'.
I think I would, if memory serves, I would be on team red on that one.
Uh, but either way, that phone was hot and this one kind of looks like it has the same thing going on.
It's got some vibe. I like it. Um, but the other actual news we should talk about on this front is the OnePlus news from this week. OnePlus essentially announced that it is done attempting to sell phones in the US and Europe. And actually, there's been some reporting that this is just the beginning, that OnePlus is in the process of pulling out of everywhere except for China. We've been covering this company for the entirety of its existence. And there was a time at which OnePlus seemed like it had a shot to be a really powerful upstart brand in the phone market and it just never quite happened. What do you make of this complete sort of retreat from OnePlus right now?
Well, one, I think you have been calling out the end of cheap devices because of Ramageddon for months now.
And it is hard to see how this is anything but one culmination of Ramageddon, right? OnePlus could not compete at the high end. It was not making flagship devices. It was part of like a larger company that makes a bunch of phones.
It's part of OPPO and it's like has always sort of pretended that it's not part of OPPO.
