Pod Insights Sharp Tech with Ben Thompson (Preview) Should Apple Buy Perplexity?, What Apple Could Learn from IBM, Vision Pro and Its Vaporware About this episode Answering all your emails in the wake of Apple's intelligence catastrophe. Topics include: whether Perplexity should be an acquisition target, the limits of local AI, the value of long tail product development, Apple's parallels to IBM, the future of Tim Cook, and a report from Ben on his latest experience with the Vision Pro.
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Andrew Sharp? Hello and welcome to a free preview of Sharp Tech.
0:08 Ben Thompson? But you go back to Apple and this question. And it's, okay, what exactly are they? What do they need to do, right? The reality is the iPhone has a lot of lock-in, remains a compelling platform. The like, all the AI developers are mostly using Macs. Like, like, like, are you actually the ultimate? People talk about build picks and shovels in the gold rush. The biggest pick and shovel maker, the most profitable one is Apple, remains Apple. It seems like it's going to be Apple for a very long time to come. Even when it comes to things like if there's goggles or visions, they're, they're well-placed, at least in theory, to, to succeed in that area.
0:50 Ben Thompson? So what's actually wrong with what they're doing? And not just, maybe nothing, right? Like, that's the whole thing. And, and I think what, what people missed about Gruber's article is he wasn't. Yes, part of it was critiquing their failures in AI. What he was actually talking about was their dishonesty with end-users.
1:16 Andrew Sharp? I mean, and that would be the problem with what they did.
1:18 Ben Thompson? Right, and dishonesty with themselves.
1:20 Andrew Sharp? Last year, yeah.
1:21 Ben Thompson? Right, that, that's what his finger was on. It was remarkable to see some motivated readers miss the point. Like, like, like, that, that was his whole thing. And, and he, as he put in the article, it would have been kind of mock-worthy if Apple comes out and all they have is like basic image, image generation and ask ChatGPT a question. And it's like, really, Apple, in 2024, this is what you have?
1:47 Andrew Sharp? This is the best you got, yeah.
1:48 Ben Thompson? It would have been fine because you still need a phone. It would have been, yes, you could sit there and make fun of them. You're so far behind OpenAI, you're so far behind Google or whoever you want to point to. And actually, from a business perspective, it wouldn't really matter because, and the whole thing that I put after that was, you know, they could be the aggregator for all these other models. Now, I did assume and grant them that they were going to actually do better on delivering the personalized intelligence bits, but you know what? Even if they don't, it's still pretty compelling. Now, that, that.
2:23 Ben Thompson? Position may diminish over time. Again, I think the the strongest assumption should be that Google is going to nail a lot of this in a way that Apple isn't sort of in the long run. So then how does Apple need to, you know, in the very long run? So assuming they're
2:37 Andrew Sharp? No, in the long run, but in the short term and the medium term, we've been saying for the last three years, maybe Apple sitting on the sidelines and letting everybody else spend themselves into oblivion to try to be at the leading edge of this stuff. Maybe that's strategically the right decision. And then ultimately, it's all going to come back to the iPhone. And so absent the ambitious announcements that then never materialized with actual products, Apple would be fine. Yeah, I think that's an important addition to this conversation.
3:08 Ben Thompson? Right, we would probably be sitting here.
3:09 Andrew Sharp? Essentially what Grover was writing about.
3:10 Ben Thompson? Yeah, we like me, I think both John and I would be sitting here on dithering, saying like just poo pooing everyone's concern about Apple if they had shipped relatively basic updates and say, look, where are you going to run these models? You're going, where are you going to access these apps? You're going to access them on an iPhone?
3:26 Andrew Sharp? I mean, every time Zuckerberg comes out and talks about AR glasses, or this, that, or the other, he's always careful to say, we don't think phones are just going to go away. I mean, even the people who want to.
3:37 Ben Thompson? I'm sure they sort of wish they would, as long as they could keep all their revenues along the way. But yeah.
3:41 Andrew Sharp? But he's not lying to people. And he frankly is more credible because he says, look, we don't, we're not going to displace the phone overnight with some of this new hardware. And all of that is obvious. And so it's really a credibility issue more than a software issue.
3:58 Ben Thompson? Exactly. That that is the key thing that that that is the problem that is revealed by all this AI stuff is they started lying to themselves. And that's the, that's what John was putting his finger on for sure. So where should they go now? Well, look at a company like Microsoft. Microsoft is backpedaling furiously from the idea they need to own and control their own model. They have decided for lots of reasons, in part because the OpenAI partnership ended up being sustainable. I think, I think the core reason is because ChatGPT is too popular. It wasn't tenable. The way that relationship was structured is ChatGPT does all the research and Microsoft Harvest, sort of the end, end, end use game, like the end users.
4:41 Ben Thompson? And ChatGPT is in a position where I think the long-term route to profitability is owning the end user relationship. It was just a fundamental structural conflict between the two companies that I actually don't, I think has less to do with, with, with the whole Altman situation. It was just once ChatGPT was this breakout hit, the relationship was doomed. Just because you can't, you can't like be serving two masters in, in some respects. And Microsoft's like, well, okay, that's, this isn't sustainable in the long run. And also, by the way, we're looking around and there's like 47 GPT four level models out there, models are a commodity. Guess what? Models being a commodity? Fantastic. Our entire business is predicated on computing be a commodity.
5:25 Ben Thompson? We treated processors as a commodity. We treated PCs as a commodity. We get to treat models as a commodity. Fantastic. Well, let's get in a situation where we, what do we do? We leverage our distribution, our, our relationship with enterprise customers and we, and we'll make, they don't need to worry about the model. We'll take care of the model. And by the way, we'll switch in models in and out. We'll figure out how to make it totally, you know, a black box to, to the end user. They will trust us and we will figure out a solution that, that, that maximizes our profitability. And we're not going to spend on Stargate and this crazy AGI dreams. We'll take what we want and then we'll, again, is that going to work out in the long run?
6:06 Ben Thompson? Might it be the case that AGI completely transforms work and undoes everything? It might be, but if you think about it, if AGI completely eliminates like normal worker PC use and productivity, Microsoft screwed anyway.
6:24 Ben Thompson? There's a bit where there's an ask, and this applies to Apple too. If we get to a world beyond the iPhone, Apple's probably screwed. So why not, why not assume that's not going to happen?
6:36 Andrew Sharp? Apple's only chance in that universe would be we're going to invent the next piece of hardware that people are going to use. And at this point, it's purely theoretical, like what that hardware will look like. But they're not going to lose out in that reality. If they don't have software that's good enough.
6:54 Ben Thompson? Exactly.
6:54 Andrew Sharp? There's a sense.
6:55 Ben Thompson? This is, I, I've talked about this for, as long as the right is to check re. There's this sense of, because people do treat tech, and by the way, I'm not complaining about this because it's very good for business. People treat tech as like a sports competition and they're cheering for their team and they want to see who can do sort of X, Y, Z. But from, from you zoom out from a market perspective, like there's no reason companies need to live forever. Like from, if I'm a shareholder, my money is in Apple because I think the device market is valuable going forward. So it's there. If I think that market is no longer viable, I don't want Apple, who's good at making devices and integrated operating system hardware devices to throw that money at something they're bad at.
7:36 Ben Thompson? I would like to make the capital allocation for myself, withdraw that money from Apple and put it into uh, another company that's better at what, at what I think the future is. And, and, but people don't think that way. Like, oh, Apple has to live forever. Apple doesn't have to live forever. I think doubling down on, we actually think devices are going to be important. The phone's not going anywhere. We're going to be the best in that. It's perfectly fine. That's what Microsoft is doing. And I think, and I think it makes total sense. So in that perspective, Apple,
8:04 Ben Thompson? it's, it's not that they're, they're actually close. They're not, they've done a good thing by not spending billions and billions of dollars building their old model or acquiring an anthropic, which I predicted they would do, I think, when chat G P D came out. That hasn't happened yet. It's only increased drastically in value. Probably if someone acquires them, it's going to be a, an Amazon. But, but you know, I think that, you know, they would certainly like to make a go of it on their own. We'll see what, but, but you know, so I think Apple's okay. And in this perspective, a perplexity is not a terrible sort of option because again, perplexity also predicated on models sort of becoming a commodity and owning that user interface layer.
8:47 Ben Thompson? Should Apple do that? I mean, maybe. I guess what perplexity does if that was Siri and you were getting that experience. It's not the worst idea of the companies in New York. I mean, I think the price is it's going to be pretty steep. But at this point, Apple's pretty desperate. I think the the the chances of M and A happening are pretty high. And the question is, does that M and A go towards? We have to actually become better at making models. Or we have to be better at the interface layer and we don't need to own a model. We can bet on it being a commodity. If it is the latter perplexity, at least in theory, certainly makes a lot of sense.
9:25 Andrew Sharp? And so that makes sense because Apple just doesn't have a great feel for how users wanna interact with this technology. And from a developer standpoint, they just lack the chops to build those sort of tools internally. Is that right?
9:42 Ben Thompson? Yeah, I mean, well, yeah, clearly they they they have they're struggling. I mean, the problem, a lot of the core problems here are really about product. Someone posted a screenshot of, I don't know if it was a beta or something about, you know, the Apple intelligence promoting an email to promote like a scam email. Um, or a spam email that is trying to get your information to like elevated that as being important. That's a product problem. Like, like, and what I mean is Google is very good at identifying spam by and large, better than Apple is. And some aspect of that is certainly machine learning in AI in a more sort of traditional sense. But the point, the the problem there isn't like, it's very easy to blame generative AI for Apple just being bad at product.
10:27 Ben Thompson? And and and there's, you know, and this is, this applies to Google as well, by the way. I mean, I'm, there's a point that I'll get to later, but there's a difference between the core technology and the actual implementation of that technology. And Apple traditionally product is supposed to be their strength. And that is a big problem here. Is perplexity a great product? I would say my big problem with talking about perplexity is I basically don't use perplexity. Um, I've found the insertion of search capabilities into regular chat models.
11:04 Ben Thompson? Kind of took away the main motivation I had. I had to use it. I certainly in theory get it. If Apple owns it, they have like, they can take care of the customer acquisition problem to a certain extent. But the alternative is Apple being Apple decides actually our problem is we need to control the core technology, blah, blah, blah, blah, blah. And we have to build our own model. And in that case, that is more of an anthropic acquisition or a Mistral acquisition. And that is, so that's the core. If they do have an A, that's the core question. Do they decide, no, we're in the right place strategically. We just suck at product.
11:43 Ben Thompson? Or actually, no, we need to, we need better model making capability. The models are going to swallow everything. If we're not in this game. We want to deliver an integrated experience and just as the old integrated experience was from the processor up to the operating system. The new integrated experience is from the model up to sort of the user interface and, and this is a question facing AI broadly, right? Is open AI, is chat GPT just going to swallow every chat GPT wrapper? It's just going to become internal to the model and that's the user interface to rule them all. I think Apple should not do that. I don't have faith in them pulling it off. And, uh, and like I said, uh, the, you know,
12:25 Ben Thompson? that's spending a lot of money that I would think as a shareholder, I'm happy to let other people spend because I don't think I'm going to be good at it and I still have a very dominant position otherwise.
12:35 Andrew Sharp? That's the thing. That's the whole privilege of being like the dominant hardware product in America. And you just would be sacrificing that massive advantage to try to keep up in a race that may or may not lead anywhere for Apple. Like that seems crazy to me on its face.
12:50 Ben Thompson? But you could argue this is the future, right? And so you actually need to get in the game fully. Again, I think, I don't think Apple does, but I understand the argument that they do, particularly if you're a really an AI sort of believer and think it's going to. There is going to be like this AGI moment and it's going to sort of consume everything. But so I would fall on the something like a perplexity makes sense. I'm sure you know perplexity and their investors would would would welcome that very much.
13:24 Ben Thompson? And not get into the like weave the hardcore investing model stuff to to everyone else. But I can get the argument in the other direction. I would but for sure, I would imagine I've heard mumblings. There's a there's a.
13:41 Ben Thompson? Apple knows they screwed up. There is from what I what I gather, there is a sense of panic and we have to do something here. So we'll see.
13:47 Andrew Sharp? Well, a little urgency is healthy. All right, more will be revealed in the weeks to come. To keep it moving, you mentioned integration earlier. Jonathan says, gentlemen, perhaps I am mistaken, but I believe some of Ben's previous takes regarding Siri and Apple Intelligence mentioned that Apple had significant self imposed handicaps when it came to machine learning and AI due to their pre existing privacy commitments. All right, and that is the end of the free preview. If you'd like to hear more from Ben and I, there are links to subscribe in the show notes, or you can also go to sharp tech.fm. Either option will get you access to a personalized feed that has all the shows we do every week. Plus lots more great content from Sir Techery and the Sir Techery Plus bundle.
14:33 Andrew Sharp? Check it out and if you've got feedback, please email us at email at sharp tech.fm.
Previous Apple History and Apple Psychology, How Apple Should Capitalize On Its AI Potential, Why It Probably Won't Next (Preview) Deep Research and OpenAI's Business Model, A Counterpoint on White Collar Concerns, The Future of Cognition and Companionship Page updated November 3, 2026 · transcript and summary are machine-generated