Open transcript
I don't know about you guys, but I miss the good old days of Apple when they would drop a new device and it truly felt magical. There's a little bit of the sparkle in some things, but it hasn't quite felt the same for a while. Many attribute this to Steve Jobs passing, but I attribute it to something a little different. I think the loss of Joni Ive was a huge hit for Apple. And there's a lot of people who agree, including Sam Altman, the CEO of OpenAI, who just posted a kind of weird video of him with Joni, who, if you don't know, which apparently a lot of my audience doesn't know who Joni Ive is, which is kind of wild to me. He was the lead of all things product and design at Apple, famous for doing crazy things like carving the Apple Watch prototype out of wood. He just weird dude. He invented the app grid with the little squircles with the corners and all that. He invented the way Apple did design, which kind of brought us into this new post skeuomorphic world. He's undisputed one of the greatest product designers in history. But he also has had a weird history since leaving Apple. From his weird design collective, Love From, to his work at Humane building an AI pin nobody liked, to his new consulting company, I.O., that just got acquired for $6 billion by OpenAI, despite not having any product. This is a weird one. I wasn't going to cover it. But then I got more and more interesting details. Then Minshi Kuo leaked what the device would look like, which, by the way, if you're curious, probably going to look something like this based on the leaks. There's a lot to talk about here, where this fits into OpenAI strategy, in particular how it contrasts with something like the Windsurf purchase, what this means for the nonprofit status, what the hell's going on with Joni Ive. There's a lot to talk about here. While you can't use any of what they're building for a while, do you know what you can use right now? Today's sponsor. Be right back. Today's sponsor is a product that I'll admit I was really skeptical about for almost four years. I only recently took the plunge, and I'm so happy I did. That sponsor is Convex. I could tell you all about their fancy AI code gen stuff with Chef that they built because of me, but I'd rather just show you how cool it is to use. They really are the best backend for your AI apps, and certainly the best backend for your React apps. Here's my actual T3Chat dev environment. It's running locally on my machine, but this will behave exactly the same in the cloud. I have this thread that I made, T3Chat homepage redesign. I'm going to open this in the Convex dashboard. Let's hop over to data. Let's hop over to threads and see the thread named T3Chat homepage. I'm going to hop in here and change this to say something else. Something else. I just pressed enter to save it. I went back to my UI, and it's already renamed. I must have done some crazy complex stuff for that to work, right? Nope. I literally just call the query the way you would with React query in TRPC, where you pass it API.whatever you defined. Everything, and I mean everything, lives inside of the Convex directory. So getting threads is super simple. I call context.db query threads with index user ID and updated at so we can sort it. Order descending, take 200. Oh no, I need to get the pinned threads too. So I grab those separately. I de-dupe them, and then I return them. No SQL necessary. It still runs super fast because it transactionalizes all of this, and then calling it on the client is just as easy. Here's how you do it. Threads from Convex equals use session query, because I want to use their session stuff, api.threads.get. It's so good. There's nothing special that I had to do in order to make this automatically update. That's all just part of the platform. The amount that Convex was able to simplify our code and actually make our app faster and less buggy is still kind of unreal to me. If you use the T3 chat beta, you know the difference. It's insane. And if you want to feel that difference yourself, you don't even need to add it to your app. Go try out Chef. You can generate a full stack app from scratch with everything from auth to file, upload to sync, handle in literally just one prompt. Thank you for bullying me into trying you out, Convex. I'm so happy I did. And all our T3 chat users are too. Don't wait like I did. Try them today at soidev.link slash Convex. Let's break down Joni Ive's history. The famous story with Ive when he was at Apple is that he was in an elevator with Steve Jobs. And Steve Jobs was just talking about some design thing. And Ive said something he thought was clever. So at some point when the elevator stopped and it was Joni's floor, he started getting off the elevator. And Steve said, no, you're staying with me and brought him to some crazy exec meeting as a random designer at the company. And he very quickly took over all of design and product at Apple. He's also the calm voice you probably remember hearing in the Apple ads for everything with the white background and the calm, slightly British voice talking about how cool the metal was on the device. That was Joni. Legend. He kind of invented the modern world of Apple. But after like, what, three decades probably of working there, he decided it was time to leave. And he started a new company, Love From. Love From was a consulting firm that's focus was helping businesses build really good designs. And he even committed to continuing work with Apple through the Love From brand, which he then proceeded to not do. As far as I know, Ive has had zero impact on Apple since he left. Yeah, hard to know. But as far as I could guess, that is definitely the case. Since then, he helped a lot at what was his role at Humane? Humane wasn't Ive. Humane was another ex-Apple person who left. My bad on that. But I am sure I've paid close attention because it was another person from Apple. If you don't remember, Humane made the stupid AI pin, spent an absurd amount of money. Like Humane raised $230 million to make a pin that sucked in their Series C stage. And they ended up selling to HP. Not necessarily the company you want to work at or be acquired by, especially because the acquisition was $116 million, which meant that the people who invested got back at best half the money they had invested. And the founders got zero. So Humane was a failure, as seems to be the Rabbit R1. The Rabbit was an interesting device because they paid a design firm to do it too. Teenage Engineering, who is somewhat well-regarded despite being kind of weird and doing things I don't love with creators, including like DMCA-ing creators for using their footage and reviews of their products. So Teenage Engineering, I have weird vibes with. But they're generally well-regarded in the art and music worlds, which is why it was strange seeing them partner with an early-stage AI device like R1 that got roasted hard because it sucked. As MKBHD put it, it's barely reviewable. I'm pretty sure they announced that they're like hard shutting it down. Not positive, but relatively sure. It's been over a year and no one's touched it since. So those AI assistant hardware devices have just kind of been failing. I was interviewed by Bloomberg recently about this and said that he considered those other things to be very poor products, which honestly is true. They're just not good. But rather than see that as a failure in the market, it seems like he's taking the opportunity to make something better. And that's what they were building at his new company, I.O., which is focused on doing something with modern I.O. and put output, a new way of thinking about compute and hardware. I can actually do this, but I don't know if this is the right path. The idea of an AI dedicated device still feels strange. In some ways, this is similar to the BlackBerry thing, where there was a lot of people with BlackBerrys, but it was still like less than one in 400 phone users at the time. The iPhone was betting not only that they could kill BlackBerry, but they could make the market itself comically bigger if the phone that was smart was accessible enough and useful enough to the average person. And that bet paid off in spades. It felt so different using a BlackBerry versus an iPhone at the time because one was built for everyone, and one was built to have a huge feature list that you could show off in advertising. And guess which one the iPhone was? The last time I felt this with Apple was actually the Vision Pro. I know it gets crap. I understand. But I've been a VR diehard for a very long time. I own way too many headsets. The Vision Pro made all of my other headsets feel like BlackBerrys. The improvement in user experience is it's impossible to put into words how big the gap is. It felt like VR that my mom would be able to use. And it had so many little magical interactions. Like I remember the first time I opened up a bunch of windows in it when I was sitting on the couch and I got up to go get a drink. And when I got up, the windows just stayed there. They didn't follow my vision as I looked around. They just stayed exactly where I put them. And when I was walking back to the living room from my kitchen, I saw the three screens over the table right where I left them. Incredible user experience. Not just like the hardware engineering to be able to position things in the real world correctly like that. Shut up, Siri. And speaking of which, Siri going off my watch randomly. The quality control at Apple is in a wild spot right now. And it's not great. The point being, I was so impressed with what the Vision Pro's UX did that I'm bringing it up here for a weird reason. If the Vision Pro wasn't enough to make VR successful, I don't know if it's a quality gap. Even I don't find myself using it that often, even though I love the thing. It turns out that VR itself might be a bad product category. It could have turned out that smartphones were similar, where most people just want their phone to have a simple keypad and the ability to text. That wasn't the case. We thought it was and it wasn't. And now everyone has a smartphone. VR seemed similar, where if we had easier to use, higher quality VR experiences, more people will jump on it. Did not happen. I really thought it would. And now the only hope, and my editor is even calling it out, is that augmented reality and virtual reality will do better in a smaller form factor. I don't think that's the case. I think we have now exhausted the limits of what should prove the success of VR and AR. And the investment into the return out ratio is just not there. Meta has lost so much money on Quest, it is unfathomable. Its software published for Quest just cannot make its investment back. It's the only reason there's anything to do on Quest is because Meta pissed so much money into it. And the Meta Ray-Bans in particular don't do much augmented reality stuff at all. They don't have a display. They don't augment your reality. They just talk to you about stuff sometimes. I know that we're not close to the peak of what mixed reality could be. I don't think it fucking matters. That's the point I'm trying to make. Apple raised the bar so much. We went from 2 out of 10 to like, I don't know, 7 out of 10. And the adoption didn't go up meaningfully. That's the thing I'm concerned about. And this is the hard test. Like before Apple made the iPad, tablets existed and people used them, but not really. The iPad almost immediately started to kill all of the netbook market and started encroaching on the general computer market within like three years of its first release. Vision Pro has had nothing resembling that splash whatsoever. There's nothing really worth disrupting on the VR side, and it's struggling to disrupt anywhere else. It's good enough that if the market was good, it would be doing better. And that's why I find IO and what Ive is doing here to be interesting. We do not know if dedicated AI devices are a real market yet. The only way you can know for sure is if you make something really good and it flops. So now we have an interesting question. How much money is it worth to get an answer to that question? How much money would a business like OpenAI be willing to spend to know whether or not hardware focused on AI interactions is a worthwhile expenditure or worthwhile industry? Is this a thing consumers are going to want or is this a thing consumers are going to be annoyed by, make fun of, make funny YouTube videos about, and then move on? Because that's kind of what's happening in VR right now. We don't know yet, but it would really suck if we made something that was just slightly not good enough so it can't go over that threshold. It would have really sucked if Apple was just slightly off on their investment on the iPhone and they made it too cheap by using a bad screen and then nobody liked it as a result and it never caught on. It's important to invest enough to make sure that first attempt that the company makes is good enough to get over the threshold. That doesn't guarantee it will be successful, but it guarantees that you get an answer. That's what I think OpenAI is doing here. They are spending an egregious amount of money, $6.5 billion on a company that's founder only had like, I think it was 500 mil was his understood valuation and like worth at the time, just 12x his net worth in order to see if the best of the best is enough to make one of these devices successful. And if it is, you now have the benefit of being the one with that device. And if it isn't, you spent $6 billion on an important lesson. That's a lot of money for the lesson, but that's the reason that I think this is happening. I also admittedly think a little bit of this is Sam trying to cosplay as Steve Jobs. Like, come on. Come on. This is an extraordinary moment. Computers are now seeing, thinking, and understanding. Despite this unprecedented capability, our experience remains shaped by traditional products and interfaces. Two years ago, Joni Ive and the creative collective Love From quietly began collaborating with Sam Altman and the team at OpenAI. Collaboration built upon friendship, curiosity, and shared values quickly grew in ambition. Tentative ideas and explorations evolved into tangible designs. The ideas seemed important and useful. They were optimistic and hopeful. They were inspiring. They made everyone smile. They reminded us of a time when we celebrated human achievement, grateful for new tools that helped us learn, explore, and create. Yeah, this is them cosplaying as Steve Jobs and Apple to an extent. And I want to be clear about that. You have to remember that OpenAI was started as a nonprofit research company looking into what AGI could possibly be. But at the same time, I do believe Sam Altman has formed a genuine friendship with Ive. Any attempt for executives at a multi-hundred billion dollar company doing a multi-billion dollar deal will feel cringe. And they did not do enough to reduce the cringe. I'm not going to pretend otherwise. But I did get a sense that they have a real friendship. And that will lead to making irrational decisions. Just imagine now that you're Sam Altman. You went on this crazy journey to research AI. And now, just 10 years later, three years after your big break with ChatGPT, you have a $400 billion valuation. You just raised, what was it, like $36 billion? No, the numbers are backwards. It's a $300 billion valuation. You just raised $40 billion. These are numbers you could not have fathomed this all being worth. And there's one other note. Sam has zero. But Sam does control the business. He has nothing here. He wants to figure out if this hardware thing is worth it. He wants to make sure if they do this, that they can get a definitive answer of yes or no. And if the answer is yes, that they win. He has no equity. But he has full control over all of this. He wants to know for a fact what AI's role in the world is. This company exists to figure that out. He also likes to support his friends and work with his friends, these people that he cares about. And most importantly, he wants to build real products. Even though OpenAI was originally founded to do research, and it did a really good job at that, ChatGPT was kind of this transition into OpenAI now being a product company, where they're hiring as many engineers to build the mobile app, the website, all the crazy things they're doing with Codex. All of that is product, not research. There are research parts of it, but it's separate from product. Is this question of the success of AI product worth 2% of OpenAI? This is what Sam asked himself. Knowing that more and more, OpenAI's product and user-facing experiences are important, if hardware ends up being the next frontier, how do we win? And how do we know if it is going to be? Is that worth 2% of OpenAI's current value? That's a tough question. And it seems like Sam Altman, when he considered all of the biases, all of the options, all of the things going on, and also that Windsurf was worth around 1% in comparison. This is a company that has built something developers are actually using that has good penetration in the enterprise world. Their valuation is around $3 billion. Spending more than double on a company with no product and no users is terrifying. But they also said they've been collabing for a while. They see the potential. And if the thing we talked about before is true, that Ive and Sam are friends, remember that Sam owns 0% of OpenAI? That is a pretty bad thing as they move out of the nonprofit because the nonprofit still has majority control over the company. They still own the majority of it. As percentages get taken out and given to people like Ive, that 2% is no longer part of the nonprofit. That's 2% of the company now in control of somebody that Sam is friends with. That part makes it make even more sense. If Sam's trying to build a team of executives that can make good decisions for OpenAI, as far as he is concerned, having his friends come in that have lots of experience in the industry to run more and more of the business helps a lot. And when I read this on an HN comment, I don't feel like finding it, started to click for me how Sam is really trying to play chess here. So when you recognize that existing products are so bad that they haven't really taught us anything, that the structure of ownership at OpenAI is weird, that Sam's desire to build product is real, it's super real, and that Ive and Sam are friends, all of this starts to make a lot more sense. So let's take a look at the leak. What are they actually building? This leak comes from Minshi Kuo, who's one of the most legendary leakers in the history of Apple leaks and news. Very trustworthy source. Mass production of this new device is expected to start in 2027. Assembly and shipping will occur outside of China due to the tariffs and all of that chaos. It'll be in Vietnam. The current prototype is slightly larger than the AI pin, with a form factor as compact and elegant as an iPod shuffle. That can say a lot of different things because there's a lot of different iPod shuffles. The design specifications may change before mass production. One of the intended use cases is wearing it around the neck. It will have cameras and microphones for environmental detection with no display functionality. So it will not have a display. It is expected to connect to smartphones and PCs, utilizing their computing and display capabilities instead. In my view, one of OpenAI's motives for announcing its collaboration with AI right now is to shift market focus from the recent Google I.O. Google's ecosystem and AI integration showcased in the I.O. keynotes pose a challenge that OpenAI currently struggles to address. The challenge being Google has a vertical integration between product, inference, model generation, APIs for people to access, and actual hardware design, both on the Android and Pixel side, where they can make their own glasses, make their own platforms, make their own OS, but also on the TPU side, where the chips doing the inference are owned by Google. Google's vertical integration here from the thing that the user wears on their face to the CPU that's actually running the inference is insane. Google is very, very, very well positioned right now. OpenAI is less so. They need to win the vertical integration if they want to keep competing in the current market. This was obviously a play to do that, not necessarily to compete directly with Google, but to have more product under their umbrella. And announcing it now also makes a lot of sense if they want to really compete with Google. AI integrated into real world applications, often termed physical AI, is widely recognized as the next critical trend. I'm not necessarily sure about this one. I would like to see where it goes. I think that's why this investment is happening, but we can't know for a while. This partnership also recalls Alan Kay's well-known adage, people who are really serious about software should make their own hardware. I don't know, man. I'm pretty serious about software, yet here I am. Really, though, makes a lot of sense. And while this leak looks really Apple, and I would be surprised if they went with such a blatantly Apple-adjacent design, it does show that OpenAI is taking product almost too seriously. To spend that much money on effectively a pile of talent, not even on a product, is nuts. But OpenAI has always been a kind of nutty bet. So for better or worse, I think I see why Sam did this. It's a chaotic move to answer a chaotic question. Will dedicated AI devices be a real market? And if so, can OpenAI win it? I'm excited to see where this goes, but I have no idea how useful these things will be. I'm curious what you guys think, though. Is this all going to blow over and be forgotten? Or are AI pins and AI devices the future of how we interface with the world? I have literally no idea where any of this is going, and I doubt they do either. But we all want to figure it out, right? Let me know what you think. Until next time, peace, nerds.
