Apple at 50 - Ben Thompson interviews Horace Dediu

Ben Thompson of the Stratechery interviews Horace on the Stratechery Plus Podcast to discuss Horace's journey to become an influential tech analyst: early career in research, business school under Clayton Christensen, and his time at Nokia during the rise of the iPhone. 

The conversation further explores Disruption Theory in the context of both smartphones and AI, why Christensen misjudged Apple, and whether AI could challenge today’s tech giants. It also marks the launch of Dediu’s new paid service Asymco One and reflects his continued conviction in Apple Inc. at its 50-year milestone.

Audio

https://asymco.com/wp-content/uploads/2026/04/Apple-at-50-Asymco-Stratechery.mp3

Transcript

---

Ben Thompson • 00:06
April 2nd, 2026. Good morning. This week's Stratechery interview is with Horace Dediu of Asymco. Dediu started Asymco 16 years ago and was a major inspiration for Stratechery. While Asymco remains free, Dediu recently launched Asymco One, a paid offering primarily focused on Apple that includes additional analysis, research, newsletters, a conference, and more. In this interview, we discussed Dediu's fascinating background, including coming to America as a refugee, working as a researcher at an offshoot of Bell Labs, and what he learned and didn't learn at business school. It was there Dediu met Clayton Christensen, who we worked with closely. We discussed disruption in the context of both the iPhone and AI, and why Christensen got the iPhone wrong. We also covered Dediu's career at Nokia, and why the company was so oblivious to the iPhone threat. Then we jumped to AI. How long will it take to become profitable and fully diffused? And is Apple the only big tech company that might be disrupted? Finally, we discuss Apple at 50 and why Asymco One is the ultimate evidence that Dediu continues to be an Apple believer. On to the interview. Horace Dediu, welcome to Stratechery.

Horace Dediu • 01:09
Great to be here. It's nice to see you, Ben, long time, but I've always admired from afar.

Ben Thompson • 01:17
Hey, well, no, it's a real, the honor is mine. You know, it's a real honor to have you here, not just because you're one of the preeminent analysts covering Apple, whose 50th birthday was this week. So it's a great time to finally have you on. We'll get to that in a little bit. You're also a major influence on Stratechery. Your site, Asymco, was one of my favorites before I started Stratechery. A great site name that I kind of tried to mimic and mine was terrible. I'm still very jealous of Asymco. Great name, great logo. You got it all working. And you sort of gave me real confidence that there was a market for the sort of analysis that I felt I could provide. And when I started, I was a big reply guy for you. I was super excited. You would compliment my posts on Twitter occasionally. And you gave me a bunch of links on a Simcoe. So thank you for that. I'm here to say thank you.

Horace Dediu • 02:04
Oh, you're very welcome. I was always happy to see your work. And actually, I was very grateful that people picked up the torch and ran with it. it was a beginning of an era at the time where analysts could actually have a voice. but suddenly going out on your own and doing it online was a big new thing with blogs, you know. And I was happy to see more and more people take up that challenge, you know. My inspiration was John Gruber also because he was doing it, but he was doing it as a like more of a tech, let's say, product and not so much business. It was more like technology analysis, but and obviously product review and all that. But those are amazing times. And by the way, just to plug myself a little bit, although Asymco has been around ever since, well, before it's trajectory, we just relaunched into Asymco 1, which is for the first time a subscription version. 80% of what we do is still public and is all readable. There's a little bit behind a paywall, but it's like mostly that's intended for people who are investors who may want to get some of the data, some of the numbers. And so we keep some of that a little bit more private. I'm still very willing to share a lot of information, though. So we're still publishing a lot. And we just launched it a couple of months ago and kind of excited to get back in the game and writing on two to three times a day.

Ben Thompson • 03:41
So it's super exciting. Yep. Well, that is the other reason to have you on. You are, like you said, sort of back in the game. Let's get to that a little bit. Before we get to what you're doing now, though, this is the first time I actually appeared on your – you had a podcast called High Density that I appeared. I think I was the third guest that was on there. So we did talk once. That was in 2013, I think. But you have never been on Stratechery. And before – whenever someone comes on the first time, I love to get the back story. And I think you have a particularly interesting one. Lots of pieces I sort of want to dive into. But let's go back to the beginning. Where you grew up, not the United States. How you came to America.

Horace Dediu • 04:22
I think your education path is particularly interesting.

Ben Thompson • 04:24
Let's go back to the beginning.

Horace Dediu • 04:25
I'm going to try to make it real fast. First of all, I'm older than Apple.

Ben Thompson • 04:31
I'm not. That's how I started my article this week. I'm like, most tech companies, I can remember the origin story. Apple is not in that camp for me. Okay.

Horace Dediu • 04:41
You know, we're separated for a few years. Actually, I was born in 68 and Apple in 76. But the thing is that I was born in Romania at a time when it wasn't part of European Union or it wasn't it was behind the Iron Curtain. And my father was sort of wanting to leave and became a dissident. at that time, wanting to leave a country that was communist was considered a crime. So you were persecuted. And so it was pretty tough times. But we made it out. We were refugees for a while. And we first moved to Italy as stateless persons. I didn't have a passport.

Ben Thompson • 05:23
I was not technically a citizen of any country. Is there a dramatic story about how you actually

Horace Dediu • 05:29
made it out of Romania? Slightly. I mean we, we, they took away all our possessions. So all we had was one suitcase per person. And I was like nine years old. So I had to carry my own possession with me. So like four people, two parents, two children, having to leave a country with nothing, no money, no cash, no jewelry was allowed. It was very tough to, to sort of, but we left in an airplane. So we didn't have to like fight our way.

Ben Thompson • 05:54
Yeah. Like not quite Andy Grove style, but yes. Yeah, exactly. So,

Horace Dediu • 05:58
So but we made it out in Italy and then we applied for immigration to the United States under some allowance for people who are refugees. And we made it. And we got there about 78. So I was a year in Italy and I learned Italian, by the way. I was like thrown into a school because it was like we left in August and in September the school starts and I had to go to attend a school in Italy. I didn't speak a word in Italian, but I learned it pretty quick because Romanian, if people don't know, is actually a Latin language. So it's pretty easy to pick up another one. And at that age, you pick it up real easy. Anyway, so I stayed in Italy. We stayed for a full year and I took a whole year of schooling there, which was great. It was elementary school, but still. Then we went to the United States and we had to find sponsorship. It was tough to find a place to live and so on. And we ended up in Cleveland. We thought we'd be in Boston at first because there was a church willing to sponsor us there. We ended up in Cleveland. And that's where I grew up, more or less, like up until high school or late high school, actually.

Ben Thompson • 07:02
I didn't realize you're a fellow Midwesterner. See, all the good analysts come from the Midwest. It's great to know. Oh, absolutely.

Horace Dediu • 07:07
And yeah, no, Midwest made a big impression. That was my first real view of America. And so we're talking 78 to 84. I lived in Cleveland, or it's suburbs. At the time, it was Lakewood, Lakewood, Ohio. It was right next to West Cleveland. And then we moved to Boston at the time, 80s. You know, we're talking 84 now, right? Classic birth of the Macintosh-type time. But it was also booming in tech, especially around the 128 belt in Boston. My dad got a job there.

Ben Thompson • 07:40
So was your dad technical as well?

Horace Dediu • 07:42
He was a mathematician, actually. So he was a professor in university. So very, very math-oriented family. My mother also was a teacher of mathematics. So I was brought up in that tradition and plus you're not really educated in business or finance. You know, people don't realize this, but when you grew up in those places at that time, you didn't know what a credit card was. You didn't know what the Wall Street Journal was. You didn't know what a stock price was. And you weren't taught this in school. You didn't go have a summer job or learn any of these things. A lot of Americans learn by osmosis. They learn about business. They learn about every aspect of finance and accounting and all these things. You kind of get into it. For us, it was a very abstract thing. And so I actually applied myself to try to figure it out. But that's a story slightly later because I took up engineering as my undergraduate degree. And I did that at Tufts University, which is near Medford or in Medford, where we lived in Boston area. And then I actually did a master's as well. And I got a job in the meantime, between those two degrees, working at a place called

Ben Thompson • 08:51
GTE Laboratories, which was kind of like a mini- By the master's, are you referring to your MBA or you also got a master's in engineering?

Horace Dediu • 08:56
No, no, no. I got a master's in computer engineering. Got it. Completely nothing to do with MBA. This was 92. Okay. And so I had an undergraduate degree in computer science and another one in electrical engineering. It was a complicated situation. I got two undergraduate degrees.

Ben Thompson • 09:14
You could make the whole computer, not just half of it.

Horace Dediu • 09:16
Yeah, I kind of had to, yeah, long story. But anyway, I also had a master's in computer engineering from the same university. But at the meantime, I was working. So I was working through my master's. I was working at the GTE Laboratories, which was actually pretty prestigious because this was like a mini Bell Labs. This is like where you were a researcher doing really basic stuff. And at the time, the research I was working on, I was a junior guy, but it was a senior what they call member technical staff and the senior uh project uh leader was working on information retrieval now information retrieval just to give you background was the name of what search became search was born with essentially the internet or even a little bit after that with you know altavista which was one of the first search engines and you had other you know other alternatives, but Google is the one that ran with it. But basically, I was doing this 10 years earlier. So we were not yet in the internet world, but there was still a question of how to retrieve documents from large data sources, mostly textual. And was this structured or unstructured sort of data that you were working with? It's unstructured. So we were training ourselves on things like 20 years of the Wall Street Journal, but it wasn't hyperlinked data. It was like plain text, huge amounts of text. But the big thing that happened between when I was there, which was 92, roughly, and 2000, was that the internet took off. Now, we were using the internet in a much, much simpler way without the worldwide web. It was the internet back on, it was just FTP and Telnet and other things that was done in laboratory type environments or academic type environment. So I was very much into all of this stuff because I grew up with, I shouldn't say I had a personal computer at home, but I grew up with mid-size, mid-so-called mini computers, which were from digital equipment or from IBM or from various like workstation, Sun, Apollo, these things. So I was like a command line guy using Unix. And for me, PC was actually passe by about 1989. I was already jumping into these computers that were much more powerful. And so we had the X-windowing system, which was very cool. It was like today, you would say it would be like a Chromebook. It was like a terminal, but it was graphical. A lot of stories. But basically, I was in that world for doing research for six years until I kind of had, like I said, very curiosity about business. I just didn't understand how it worked. And I thought, hey, the way you learn business is by going to a school that teaches you business. That was my naive idea. I said, well, what schools teach business? It would be business schools. And so I applied and I didn't get in the first time I applied. And the feedback was that I didn't really have business experience. You needed a little bit. You needed something. And all my work was like being a very technical researcher.

Ben Thompson • 12:26
I'm the guy that went to business school from being an English teacher. So I know exactly the challenges that you're talking about.

Horace Dediu • 12:34
But it was like, again, they just gave me nice feedback. And then I said, okay, I need a job in business, quote unquote. So it turns out I was headhunted by someone to chase Manhattan. Now it's called Chase, I guess. But it was back in, again, 93. And I went to New York and I worked in effectively an IT role. So it's not very challenging technically, but it was at least in a business environment. And then I reapplied the year after. And lo and behold, I got in. And so I had to cut short my New York career, my IT career, my banking career as it was. And I went back to Boston and I did two years at Harvard. And that was when I was kind of, I was partly blown away, but also partly disappointed because actually there was no theory of business.

Ben Thompson • 13:29
Right. No, no. Tell me about this. Because you said before you were naive. In what way were you naive?

Horace Dediu • 13:35
Well, I thought there would be a theory that because when you learn engineering, You learn first science and you learn basic physics and chemistry and you learn mathematics and you learn that things are axiomatic and things are built on top of each other so that there's consistency all the way up the stack. So the way a computer works can be traced all the way down to the physics of subatomic physics, if you wanted to VLSI circuits, all of these things are based in physics. And so there was this notion that, okay, we start with the axioms and you build on those and you make proofs or you establish causality at least, like A implies B implies C, therefore A implies C things of that nature. So you do a lot of math and you think that that's what business does also depends on these things. But then when you go to a business school, you realize the way I put it retrospectively is that it's like equivalent of sitting around a campfire telling stories.

Ben Thompson • 14:34
Especially at Harvard where everything is a case study.

Horace Dediu • 14:37
Everything's a case study. And they got their idea, by the way, of case studies from the law school, which was just across the river. Because when the business school was started, they didn't know how to teach business. And they just said, hey, they have this thing called cases in the law school. And basically, if lawyers can learn law this way, then maybe business managers can learn business this way. And that's where it came from. And I think Harvard was like the second school. So at the time, it was like 1903. It was the birth of the business school was like not that long ago. Anyway, so you go through the case method. And yeah, there was an inkling, however, of a theory. And it was maybe some heuristics and there were some rules of thumb and things like that. And in finance, you do a bit of math. Okay, that's fair about that. But the thing that really influenced me was my professor at the time who was teaching us operations. And his name was Clayton Christensen. And he stood up and then said things to me which resonated deeply, which were that, hey, we don't know what the causality is, but we are going to find out. And there are times in history where businesses or parts of business management were completely unknown, but we figured it out. And the example you would give was like in operations is that the factories used to be either good or bad. And the only determination we knew whether they were good or bad is whether there was a good manager or a bad manager. So a good manager was worth his weight in gold and was always getting poached to run another factory because somehow he had the magic touch. But that was like in the 50s. And by the 70s, we knew or even 80s, we knew that you could manage factories to be reliable in their output without having to rely on some genius. And the Japanese, partly Japanese, are responsible for teaching us this, although they themselves learned it from Americans much earlier. Long story, but basically at the business school, I was like, wow, hey, this is actually exciting because this is an emerging theory. So maybe I can help even figure this out, you know. And so at the time, Christensen was writing his book that became a classic, which is The Innovator's Dilemma. He was actually writing it while he was teaching us. So he was even using some of the materials on us. And we didn't know at the time, but they came out as chapters in a book. So and two or three years later, I think it was 98, well, I graduated in 96. But in 98, the book came out, the internet blew up and suddenly everybody was doing startups. It's funny because in my graduating class, normally you would go do management consulting or you would do finance. These were the two major categories of employment for graduates of the business school. Historically, it was more like general management, but that had gone out of favor by the 90s. Nobody really went to become a CEO of IBM from there. But finally as the late 90s rolled around, suddenly five years after everybody graduated, they're all working in startups.

Ben Thompson • 17:36
But I actually went straight to a startup in 96, which was very unusual. Well, I want to ask you real quick about the Professor Christensen. it was there an aspect where it was actually less the theory that compelled you and more the fact he was trying to sort of systematize and produce a theory in the first place it was just that there

Horace Dediu • 17:58
was a hint of a theory that he he proposed that there are heuristics that work um for example the toyota production system um tps famous the machine that changed the world you know he he went through that. We went through that. And you explained, for example, just asking five questions why is better than asking one question or no questions why. And the Japanese kind of had this rule of time that you should ask five whys because you're going to eventually come to the real cause of a problem. And although that's not really a very strong theory, it's actually an extremely powerful idea that you should seek out the causality. And then you ask the same questions about why people buy things and as hard as it is to believe that the determination of why people buy things was a complete unknown and probably still is but to solve that they propose a theory called the jobs to be done theory as opposed to the kind of like women buy this and young people buy that it was not

Ben Thompson • 18:56
the attributes of the person not the attributes of the thing they're trying to do which was completely

Horace Dediu • 19:05
different it's like your outcome based your circumstances based and and all you had to do was be patient enough and and ask the people who were making the purchase decision follow them not not to ask them why they do it because they themselves don't know that was another piece of the puzzle you can't just ask anyway so so what he had was he called it like drawers you would pull on the cabinet you know and inside each drawer there was a theory so kind of you try to pick and choose from a bunch of ideas and stuff so it's the basic beginning of empiricism. It's like whole pattern recognition. Oh maybe we should categorize things. These seem to go, if things go together and these are separate, this is very basic stuff. And I with my own background, I wouldn't say it was like very deeply scientific, but at least you get a bit of introduction to the scientific method. And it was like, wow, actually, this is exciting, because there's probably a theory that's out there, which may or may not be true, by the way, in retrospect. But I was excited at the time to think that there was a theory. And secondly, that it seems like this is the way to discover it. And so, but I wasn't yet ready to dive into being a researcher on business theory. What I was hoping for is to get a job and pay off my loans.

Ben Thompson • 20:19
Let's circle back to Christian Sid and sort of some of the hints you just gave about the theory and how applicable it might be. Certainly something that I've written about and thought about a lot over the years. But did you go straight to, you mentioned you went to a startup. Where does Nokia come into this journey here after business school?

Horace Dediu • 20:36
Okay, so Nokia was like my third job. I did one startup with a partner. Then I went on my own. They did another startup. This was like about 99. I hopped out and did my own. This was going to be in the e-book space.

Ben Thompson • 20:51
Got it.

Horace Dediu • 20:51
So then my first startup was document analysis. We were using today, we wouldn't call it AI, but back then it was called AI. We were trying to look at a document and break it up into its components and then try to reconstruct the document. And that got applied to PDF, by the way. PDF extraction is something that people are kind of doing now. But that takes essentially, it took an expert system to do that. So that was the first start. Second one was actually with the knowledge of PDF, I was able to take all printed materials and make e-books out of them. Essentially, a production system for e-book publishers. So they could take their pipeline, which is print-oriented, and then reoriented to this new thing we call e-books. And at the time, these were going to be read on computers mainly, but on the horizon, I was watching and talking to a lot of people on the horizon with devices. They were, initially they were PDAs, but we were looking for the done. And I said, phones are going to be huge. Already dumb phones, cheap phones were quite popular at about 98, 99. And millions and millions in the world, right, were sold every year. And I said that little screen could get bigger. That was all my, that's all I observed. And then I got on an e-book standard committee and met somebody from Nokia there. And I told them, guys, and this was in the U.S., but the Nokia guy was finished. But I told them, guys you're the ones that are going to dominate this because the phone is going to be the reading tool of the future. And they're like, you're the only person who agrees with us. Everybody in the U.S. Well, that's interesting.

Ben Thompson • 22:37
You're the ones that agrees with us. That implies that they saw the future. Oh, absolutely they did.

Horace Dediu • 22:45
this is 98. In 1996, there was a movie called The Saint, in which it featured a smartphone. This is 1996. Val Kilmer was in it. And he had this opening phone, which they called it the brick. But it was a phone that opened that had a screen and a keyboard inside a phone in 1996. And by 1998, they introduced a new version. And by 2000, they had a color version. And that was called the Nokia Communicator. Already on the third version. And that was like such a cool product for me. It's what I carried for many, many years. And I read books on it in 2000. I read the Harry Potter series on the phone in the year 2000, 2001. Think about it.

Ben Thompson • 23:28
No, we're two pieces of the pod here. I remember I got my first cell phone. I want to say 99, 2000 somewhere. But I quickly figured out how to get an RSS reader. Or I think it was like some like website where you could like get your feeds. And I would just read on there. like four lines of text all the time, which was great until I got my phone bill, which was astronomical. I'm like, oh, shit, the data was quite expensive back then. But yes, this is the things we we we I

Horace Dediu • 23:58
mean, and the thing was, OK, so just a little back to the story. So now people were like, this is only guy in America who seems to agree with us. The Japanese had ideas, the Europeans had ideas, but I was like in the US at the time. And so I said, all right. And then they said, would you be willing to just be a consultant for us? We want you to write a report on how phones could be used to deliver content in the form of written content. It could be newspapers, could be magazines, could be books. But these were all businesses. These were publishing businesses. And I said, as far as media is concerned, this media problem can be solved with phones, media distribution, how do you make a profit, what's the margin, how do publishers work, how do you get deals with them, how do you protect it, just the whole business plan for distributing. And I wrote it up about books. And then they said, great, now we want to actually implement some of that, which would be willing to come on. And by now, we're looking at year 2000. In 2001, as you know, as we crossed into 2001, the great crash occurred.

Ben Thompson • 25:05
Yep. And suddenly a job offer from a big phone company sounds pretty great.

Horace Dediu • 25:09
Yeah, but the crash occurred and it impacted my startup because I couldn't raise money anymore. And that's why I said, OK, I'm throwing in the towel. I'll take this idea to Nokia and build it inside of this company. So for the first year, I actually worked out of the U.S. and then they invited me to come to Finland. And I fell in love with it, not to mention finding someone to fall in love. Yeah, I'm still here. But let me, so we created an internal venture called media distribution. And it struggled for a couple of years now. We partnered with a publishing company here in Finland that is the biggest newspaper and magazine company. And we worked with them. Again, 9-11 happened. We worked with them throughout that period to test this idea of distributing content on phones. And that's how I met my wife, actually, because she was working for the publishing firm. and we were in the meetings together. And my point is this, that remember when Apple News launched, like I don't forget how many years ago it was already. It was five, six years ago. And she texted me during the keynote. She goes, you realize that's what we were working on together in 2001.

Ben Thompson • 26:22
Right.

Horace Dediu • 26:22
It was called Project Melissa. I can save that now. Project Melissa was this idea of Apple News today. And that was 25 years earlier. Yep. Or maybe 20 years. And that just goes to show you sort of the story of my life. I was doing search 10 years before Google existed. I was doing e-books 10 years before Amazon launched the Kindle. I was doing smartphones and media distribution on smartphones 20 years before Apple more or less showed how it's done. And so my problem has been that I kind of see things happening or see where things are going, but way too early. And that's why one of the theses I have over time is like, I don't know what to call it, a quote I have. It's the idea that the difference between a futurist and a billionaire is knowing when one thing happens. So it's not about knowing what's going to happen. A lot of people are very good at that.

Ben Thompson • 27:31
Including Nokia, to your point.

Horace Dediu • 27:33
Nokia, science fiction writers, Elon Musk, a lot of people predict this. But getting the timing right is where the money is made. And so that was one of those lessons I learned.

Ben Thompson • 27:45
Well, when you talk about the publishing and Nokia and the reading and all those sorts of things, what it makes me think of is when the iPhone comes along in 2007, And I'm very curious to get your perspective on what that was like inside Nokia when the iPhone launches.

Horace Dediu • 27:59
I'll tell you exactly.

Ben Thompson • 28:00
But before I get there, but to this publishing point specifically, the key bit was the Internet was permissionless in terms of producing content. Like, it's great that you met your wife, but the fact that you needed to be in a situation where you could meet your wife was actually the reason why it wasn't going to happen. Because if you have to make a deal to do everything, it's just never going to get there. Whereas if you can just surf the Internet, the real Internet, and no one needs to do anything to adapt to it, suddenly you get it all for free. And then you get the virtual cycle going where they're making stuff for you. And that was sort of the, that's the timing bit is like the, the external structure around what you're doing.

Horace Dediu • 28:41
Yeah, it's, it's, it's complex and it's multi-causal and you're right. It's multiple, the reasons why things happen the way they are. And it's just not foreseeable. The internet was, and by the way, the thing that really caused the project to fail and be abandoned, this media distribution side note, it was actually the fixation on DRM. And DRN actually killed a lot of great businesses with content ideas back in the day, whether it was music or video or textual. It was people just obsessed about protecting it before they obsessed about distributing it and making it great. And this is why when iTunes came along, even though there were some circumstances there with being on a Mac and not on the PC, that the publishers finally said, why don't we go ahead with this? So long story there. But as far as what Nokia was doing, so that was like for four years I did this startup inside of Nokia that was kind of abandoned. And then they switched me to, because I was kind of, okay, he's the guy with an MBA, he's the guy with an engineering degree. He kind of can speak both languages. Let's put him in charge of doing some analytics. So finally, I got into the job of essentially developing a strategy analysis and doing competitive analysis initially. So it was like they said they asked me to do Microsoft. So Microsoft was the biggest worry for everyone because they had intentions of being in the phone business. They made no secret about it. They had prototypes. They were working with competitors of Nokia to give them the operating system so that they would swap Nokia with like Samsung running Microsoft, essentially mobile. Went for many names, but essentially that was the effort. They were pushing on all fronts, except Nokia was the one that was holding out and saying we've got our own operating system and so on. So they gave me this job because every other company was hardware. All the analysts were covering hardware. That was the only guy who was covering what essentially was a software company. And in doing that, I was also sort of nibbling at Apple a little bit because the iPod had come out. The iPod was doing amazing, but it wasn't a phone product. And I was kind of looking at it a little bit, because now Christensen comes back in the picture, because his book had taken off. He has already written the second book, which is The Innovator Solution. And I was like, hang on a second. A lot of stuff that I read in the book, and I remember back from my days at school, applies to Apple. here because they're doing a lot of interesting things with being disruptive as the case may be. And Nokia itself had been disruptive in its early years. So, but mostly I'm focused on Microsoft. And my point was that we can fight this battle and win, but we got to be very careful. And there's a lot of twists and turns there, but I took it upon myself also to understand what's going to happen with this media product from Apple. And then I began to write a few internal notes because nobody was covering Apple. This is really interesting and important. No one's not coming. It was no cover. No, no. Imagine you have a team of a dozen analysts, you could put on multiple jobs you assign people to multiple problems. No one thought that Apple was a problem to be looked at even like, okay, just put it part time when you're on your plate. No, there was nothing interesting about it. And part of that had to do with the fact that they were one foot in the grave already in the 90s. And now they were more or less like this little toy media company. They didn't really have scale. Phones were on the orders of billions of users already. And this was iPod was in the tens of millions, maybe a hundred million. So it was orders of magnitude. There was no real worry there. But I was really annoying at this thing. And I started building a model. I remember distinctly opening a spreadsheet, looking at, I had access to a research report. I was looking at research reports and I was thinking to myself, okay, let's look at the history of the iPod, how it's growing. And I started to match that to the Razer, which was the most viral product at the time, the Motorola Razer. It was like, even Nokia was jealous about this thing. And so, and they were like, Damn it, they sold 50 million in total. That was like, at that time, they had a cumulative sales of 50 million, which was considered way too scary. And BlackBerry was another one, by the way. So I'm doing this and I'm like, hang on a second, what if they do a phone and it's a combination of the Razer and the iPod, could that resonate? And I just put it out there. I said, it's possible. And there was a rumor of the iPhone coming out. Right.

Ben Thompson • 33:31
So Apple did do a Motorola Razr with iTunes functionality. Did that actually end up, because that was a flop. And did that actually make people not as like more relaxed?

Horace Dediu • 33:46
No, it was, if anything, because we knew that it didn't run an operating system. It didn't have a developer API. Okay. There was nothing, no hooks in it. It was just an appliance. And as such, it was like a feature phone. And that was not interesting. We had been talking and pounding the table on smartphones already since 2001. Remember, when I was joining, we had a communicator. And I was also, one other job I had as an analyst is to project the smartphone market. We had Symbian, Series 60. We had the interfaces. We had all picture phones, music, etc.

Ben Thompson • 34:20
I was a user as well.

Horace Dediu • 34:22
And I had done the numbers, by the way, and by 2005, more or less people were burned out on smartphones. Because the projection was like this, but it was more like this. It wasn't growing as much as you thought it would. And so a lot of people kind of said, especially the top manager was like, you guys are very enthusiastic about the software stuff, but it's just not working out. We're going to double down on hardware. So there was a lot of burnout. So, but I kept looking at the phone at Apple and I suggested, hey, what if this thing scaled away? Like I said what if they could sell as many as 50 million? That was my provocation. And people were like, dude, I was essentially taken aside and say, tone it down, okay? First of all, you're not doing yourself any favors by being so provocative. People, you're just going to lose credibility. my boss, who's a great guy he just took me aside and said, just don't do that. You know, it's bad for you.

Ben Thompson • 35:22
And this is like 2005, 2006, around then?

Horace Dediu • 35:25
Well, already 2006, because the iPhone was rumored. It wasn't out, but it was like, if you're going to have iPod success and you're going to do this crap phone with Motorola, you're probably thinking, maybe they're going to do something interesting beyond that, because we knew they could, in theory, do something. We didn't see the touch interface, but we saw maybe like a scroll wheel or something. You know, Tony Fabdel was running the thing. And anyway, so bottom line is like when it finally came out, it was like about this time. No, it was like almost January, I think. Yeah, January 2007. Yeah, January 2007. It came out and this was not for sale, but it was like on stage. And I was so excited. And I said, this is unbelievable. I remember running into the office the next day. Usually you socialize at lunch hour, sitting in the cafeteria, talking to your friends and asking them what do you guys think about this? And most people just said, oh, it doesn't look that interesting. aspects of it were copied. There were copies from here and there. There are other people that don't touch. You know, a stylus has been around for a long time with BlackBerry also doing amazing things with keyboard and so on and so on. So there was a lot of cynicism and sort of not really interesting. I said, man, it's running OS X or Linux or Unix or whatever you want to call it. But it was basically running a full operating system. And I said, this is going to get developers working on it. But again, the other thing, the other head fake from Apple was that Steve Jobs said, we're not going to put any apps on this thing unless we write them. So that like, oh, man so I was actually anticipating far more than even was being talked about. And for that reason, it was just super frustrating. Anyway, even two years later, I was still working there by 2009. I left in 2009. But by then, they were still not following Apple.

Ben Thompson • 37:13
How is that possible? Like even two years on, this device is out there.

Horace Dediu • 37:19
Because it still didn't have distribution. And this is the thing that consumed everyone following the iPhone in the beginning. It's like they had one launch partner. It was initially Singular became AT&T. Don't forget that the old AT&T had nothing to do with this AT&T. It was just a rebrand of Singular, which is a crappy, very small, not very small, it was maybe second tier.

Ben Thompson • 37:47
Verizon didn't want to do it. Apple did this again and again. Verizon was number one.

Horace Dediu • 37:50
They just watched with second tier. Verizon was number one. Absolutely. But Verizon was in CDMA and all these issues. But the thing is that it was a crappy network and they didn't have a lot of points to sale. They didn't have distribution. and everybody said this, you need 500 network operators to saturate the global market, right? And how fast are you going to be able to obtain distribution was the question. And so they're so cantankerous. You know, Apple is known as being like my way or the highway. And in fact, it's a very crazy thing for singular. Yeah, Verizon said no for years. And so Apple was like, this is, I'm sorry, Nokia was like, this is not going to fly. These guys are, if anything, they're naive. Don't forget, by the way, there's a separate story on how BlackBerry dealt with the iPhone. There's a movie about it, by the way. Beautiful movie. It's a great movie, yeah. Everybody should watch it. But basically, everyone had the similar responses that this is not kosher. This is not the way things are done in this business. It's all about relationships with operators. It's about getting shelf space. It's about getting salesmen in the store who are bribed effectively with a commission to tell you what to buy and like effectively yank the phone out of your hands and give you another one. So this was this was it was a very kind of dirty business at the ground floor at the street level.

Ben Thompson • 39:19
So so what do you know what was the first what was the first article of mine that I think you ever linked to? It was about precisely. No, no, you don't need to know. I know because it was very formative to me. That's what I'm going to tell you. It was basically about talking about this was how distribution worked. And what Apple changed about the whole thing was that customers actually switched networks for the iPhone. And it was doing just sort of a traditional sort of business school analysis of and pointing out the power shifted to instead of distribution to the buyers. Basically, what I ended up writing about for 13 years now is demand mattering more than sort of the supply bit. But it was the exact same thing. That's why you linked to it. You are you are already on board with it.

Horace Dediu • 40:04
But it was also the thing that resonated for me personally. Not I knew something about the phone. This is mostly I knew how big it could get. I knew that there were billions of people, many more billions than used PCs, which was by far the most dominant influence in technology was the PC. and most people oriented themselves, like even Gruber and others who were like, okay, Mac first and then all these accessories. But I knew the phone would be a couple of zeros more, number one. Number two, I also understood distribution, like you said, but also there was this problem of like the incumbency, which were all the phone makers, were so far from understanding what the software dynamics of platforms were. By the way, platforms had been more or less, platform theory was developed by Microsoft. They pretty much invented the whole idea.

Ben Thompson • 40:50
Was it because you were analyzing Microsoft that you understood the software aspects?

Horace Dediu • 40:54
Yeah, but I knew Microsoft in my gut already in the 90s. There was this little when I was at the laboratories and we're using very powerful workstations. And this like crummy PC is just outselling every great machine out there. And it's worse in every way, and yet it's still killing everyone. It crashed all the time, and yet it ran Windows 3.1, by the way, even before we got to Windows 95 and then NT as a better version of it. And all of us were like, this is awful. This is just, how could this win the computer game? The computer race. Mac was elegant, but Unix was a real operating system. And then these PC guys have a non-multitasking non-real-time, just really bad technology-wise, right? So I need enough engineering to know. And then my mind was blown, like, these guys are winning even though they're not good. All these things, you can't imagine how you have to go through multiple disillusionments before you realize enlightenment. You have to be hurt badly. This is what people don't understand, especially young people. It's like success doesn't make for greatness. Failure does. And this is, so when Apple was doing it, I also knew they had been themselves through the wringer. And I thought, how do they think? What do they think? How do they do things? Why are they focused on the customer in a world where the customer was number four on their list of importance, right? You know, distribution, technology, lock-in, platforms, developers. Microsoft in the 80s and 90s was all about those things. customer experience was not high. And those who did focus on customer failed. And I kept going back to the theory, kept going back to Christians and I kept going back. And I I had some instinct. I was not, by the way, a Mac user. I was not, I had used one yet before, but it wasn't like my main machine. I had not even spent my own money on one until about 2004. Yeah, same year I got a Mac. Yeah. Yeah. I, in 2001, I did get an iPod, but again, I was using it with, uh, well, initially with a Mac, but I had a Mac. I remember my first Mac was actually in 2000. I slightly misremember. I had a Cube as my first Mac, but my first work computer was 2004 with a laptop. And I had the first iPod in Finland and I brought it to work and I showed it to people and people loved to look at it. It wasn't available to buy. So nobody had even seen one. Nobody had seen a BlackBerry for years there after it was launched. So unless you were a competitive analyst, like you had the budget to go buy one somewhere, people just weren't exposed to these things. Anyway, so this is one of those hard lessons about understanding the response mechanism of incumbency, the response mechanism, especially these are not stupid people. These are the best in the world after what they do. The best executives at Nokia were the best people to understand phones, including engineers. You know, they need distribution. They knew how to segment properly. They knew how to sell billion phones. It was incredible. But at the same time, they had their blind spots. And so this is where you kind of have to have a little bit of faith. You're not going to work with data. That's another thing Christian said, his famous quote was like, God is very cruel. He gives us no, none, no information about the future. How could he do that? So I thought about that pretty hard. All the interesting stuff that we need to know is not available. The most important data is the one you don't have. And so then you have to rely on theory.

Ben Thompson • 44:39
To the theory point, and you mentioned Clay Christensen. Why did Clay Christensen get Apple so wrong? He predicted the Apple was doomed again and again.

Horace Dediu • 44:48
If we fast forward to 2013, I actually went to work for him. so i left at 2010 i started a simco yep exactly a couple years i did that on my own i was getting a couple of gigs out of it and stuff but uh but basically he approached me people who knew me and knew him uh had told him about it hey do you know one of your students is doing this stuff on my name he wasn't probably even reading my stuff but people recommended it and he said would you like to come to join the think tank I'm forming. We're going to get corporate money to fund my research as opposed to academic money, but it's going to be from businesses. And we're going to have a couple of fellows that are going to sit and write and think about the future. And I said, well, man, this is a dream, working with Clay, getting paid. So I took a job in 2013 to work at the Clayton Christensen Institute. And the idea usually was that you work with him for a couple of months or so, and then you decide on a topic, and then you plan on writing a book with him. So you become a co-author on the book. And so he had done this with collaborators over the years. And he had a pipeline so that a book would come out every year. His publisher essentially ensured that. So I went to him and I said, maybe we should do something. So why was he wrong on Apple? Well, the reason was the price. At the beginning of the disruption theory, his definition of the low end was that it was like a cheaper product. So the idea was that low end meant almost always a cheaper product, a more humble product. But the problem is if you measure performance, it's all relative. So an iPhone. Sorry, let's start with the iPod. an iPod, although $400 in the beginning, it was not competing with a cheaper MP3 player, which was 50 bucks at the bottom, right? Very, very cheap stuff. But it was competing against either home stereo or something like a Sony…

Ben Thompson • 46:57
A huge binder full of CDs and a Discman, which is…

Horace Dediu • 47:00
Yeah, Sony, whether you had a DVD player or whatnot, So it was very small. It was very beautiful and had amazing design and a thousand songs in your pocket. And people, that was just nowhere to be found, a thousand songs. You know, you would have a library at home fewer than that. So my observation was that what Apple was doing usually is they were introducing product that was relatively expensive. So a Mac may be more than a PC, but it was much cheaper than getting a supercomputer or something. And so the relative low-end was important, as I put it to him. And I said this, when the iPhone came out, by then it was already five years out, I said, the iPhone is a low-end computer. Now, it took an engineer to decide this because I said, it's got an operating system just like a Mac. It's got the memory and storage and even the display that are pretty good compared to a computer of a few years ago. And now we're saying, by the way the Mac Neo has got a phone processor in it. So that's the whole idea. So I said, if you compare the iPhone to a computer, it's an amazing low-end product. And that convinced him, actually. He said, yes, we were wrong to categorize only on price. We have to understand that performance can be relative to something much bigger. This is why when Apple usually introduces a very expensive product, I don't look at the other PCs. I would look at like a monitor. Let's say the Vision Pro. The Vision Pro is $3,500. But if you're a surgeon and you're putting that on for help in the operating room, which I know surgeons that do this, and there's not a piece of equipment in their operating room that costs less than that device. In fact, you're looking at 30,000 for a monitor, you're looking for 100,000 pieces of equipment everywhere. So for them, it's almost a disposable item. It's so cheap. So it all depends on what it and it gives them huge superpowers. this is an amazing product. So there's a lot of relative, you have to think relatively. And that was my point to him. And I said, we need to maybe rework a little bit the theory to do this. But what I settled on working with him was to figure out mobility and transportation and particular energy. Because remember, 2013-14, a lot of green tech was coming out. A lot of people were worried about the environment. Global warming was a new thing. And then I said, maybe we can look at the whole transportation. Apple, by the way, 2014 was when the rumored Apple car project started. So I thought everybody's going to get in this game. So we're going to have to look at cars. And that's what I, but the long story short is that after about two years of pounding the table on that, I realized that it's not the cars we've got to fix. We've got to actually find lower end vehicles, which became micro mobility. So I created the idea of micro mobility and I left to do another startup basically because I believed in so much of this that I thought rather than write about it, I'm actually going to build it.

Ben Thompson • 50:17
Well, you're still, you're still working in the micro mobility space, right? Do you have a big conference coming up in a little bit?

Horace Dediu • 50:22
I started two companies. One of them, unfortunately, ended because of COVID. The other persisted and was sold last year. But I'm still working with, I still have a relationship with the buyer. So we're still helping out. But I feel very good about it. it's again a question of timing. I'm afraid I'm a bit cursed that if I say something's going to be huge like the Vision Pro or micromobility, you probably…

Ben Thompson • 50:53
It will be, just who knows what. It's my time frame.

Horace Dediu • 50:56
It's going to take a while longer.

Ben Thompson • 50:58
Well, in this context, you saw all this stuff coming, and the question of timing is a good one. Where would you put AI in your personal framework? What is a bit of a surprise? Did you see this coming?

Horace Dediu • 51:12
Again, as an old computer science guy, having taken the AI classes in the 80s, you don't get very excited about every new AI that comes along. There's been so many. But this one was interesting and was surprisingly rapidly adopted. And it impressed a lot of people, including me. But it was very difficult to go from that to a business model. And again, the question of distribution. So I keep asking these dumb questions like, are these things going to get into the hands of people? Are consumers, what's the job to be done? What's the consumer market versus business market? Is this going to commoditize? Is it going to be horizontal? All this, by the way, is going to get written up in great detail on my website. And I've already started and I'm asking some tough questions there. But the thing I want to point out is the old rules of business haven't disappeared because we ended up with a new technology. They never do. Every new technology obeys business rules. And what are those business rules? Well, things like I said you have to have profitability. You have to understand the customer. You have to have distribution. You have to decide if it's modular or integrated. You have to ask yourself, do I stick to this piece or do I integrate to other pieces? you know, should I make my own microprocessors or should I why should I not be in the silicon business? Why should I not be in the why should open AI do its own operating system or processor? The fact that Google, a search company, made an operating system, this is not obvious. This is not obvious at all. In fact, you would have been bet against that idea back in 2000. So when you study the problem, you have to understand that in order for this company to be successful, to have the fuel it needs to grow, to become dominant or whatever its target is, it has to have some profit or it has to be, you know, you couldn't go without profit for a long time, but then you have to raise money. You have to figure out where it comes from and how to deal with the whole economic question. So technologists and business people tend to be oil and water. And historically, I think you can't usually think, like I said, because one theory is very different than another. And so now we're in a world where many are expert in both, but it's still two things in your head that are often in conflict.

Ben Thompson • 53:51
Well, you told the iPhone story before. And what it sounds like to me is you're inside Nokia, even though you're a business analyst, you were the technologist. And you were dealing with the greatest business people who were thinking about distribution, were thinking about profitability and all these bits and pieces. But because you understood software, you understood platforms, you could see what was going to happen. When you raise objections like business type objections, do you ever feel concerned? Am I a Nokia executive right now? Or am I young Horace who sees the iPhone coming and it's going to be a big deal?

Horace Dediu • 54:24
You always have to be full of doubt. I think scientists have their superpowers. their superpowers, they just are skeptical. You have to be skeptical. But here's the answer to the AI question. I did do a piece that I didn't publish, when I will publish soon, about how long it takes for AI to reach saturation, where we have to define what that means, first of all. But while I was at the Institute, I took it upon myself to ask this timing question, because I felt I was so bad at it. I wanted to understand how long things take. So I stumbled upon all this data about s curves s curve is a phenomenal uh once you understand what's behind it and how it works it's it's a phenomenal uh uh revelation it's it's it's it's an epiphany and what the s curve tells you uh or the diffusion curve or the logistics curve um what it tells you is to uh understand and jeffrey more popularized this idea of sort of that there are stages from early adoption to later or laggards. And that's based on even earlier work in the 1960s. But basically what you understand is that some things are fast and some things are slow. I know it's shocking. But in the world where we live, where everybody assumes everything's fast, there are things which are slow, like automobiles. The speed that that's changing is far slower than anyone in 2014 imagined. As a quiz, ask yourself how old Tesla is. Tesla is from 2003. It's already 20…

Ben Thompson • 55:58
23 years old.

Horace Dediu • 55:59
You know, 23 years old. I don't know exactly when it was born in that year, but roughly.

Ben Thompson • 56:04
But we consider it new because automobiles were like stagnant for like 50 years before that.

Horace Dediu • 56:09
But we think of it as a startup. We think of it as a sort of an up-and-coming company. And the incumbents are very old. But still, it has like 3% or less market share in terms of all cars sold. Actually, it's a bit even less than that. I think it's 2%. You know, the electric car, you can build an S-curve. So I did this for AI. And I did this like three years ago, almost. Two, sorry, two years ago. And there wasn't that much data to put on the graph. But what I used as a proxy, as a way of sort of calibrating this curve, were previous inventions that I felt had a similar feel to them. That is to say that they were mostly trying to do intelligence. And the two examples I used were one from the 15th century, which was called the printing press, and another from the 19th century, which was called the automatic loom. the machine that makes clothing. And those were both, I felt, intellectual machines. One was replicating intellect. The other one was encoded to create a pattern based on code, based on data. And the people who were involved in that adoption, who took it up, who invested in it, who ended up having many times been fighting over it or getting killed over it. There was a dynamic there very similar to the tension in society that we see with AI. And my conclusion based on that and a lot of others, 140 different diffusions, is that AI will take about 50 years. Now, the problem there is when you mark the beginning. And the beginning would have to be not just all NEI. I think it would have to be about, I'm still debating this, whether it's about 2003 or 2004 when the algorithms were first proposed or 2017 when the first implementations happened. And I call this as market zero. It's like the point of invention also is not the point usually where a market begins. So it's very tough to pin that thing. And then you have to ask, what do you define as adoption? and it could be many things. It could be space, like railroads is a network or it could be users, but you got to be careful because they have to pay something usually to be counted. Free stuff is, of course, easy to adopt. So maybe I'm going to modify my thesis, but I think it's a 50-year cycle and we're going to be measuring it in terms of AI being in the hands of everybody and they're using it every day.

Ben Thompson • 59:02
Well, you said you did that two years ago. Have the previous two years affirmed your thesis or has stuff maybe happened faster than you expected or slower?

Horace Dediu • 59:13
I think maybe if there's a bias, it would be maybe it's a little bit faster, but it is still not yet clear. I had the same struggle, by the way, with crypto. Everyone was saying it was going to take off. And it was people were counting wallets, but there weren't a lot of practical applications. And I kept looking for the killer apps. I kept looking for this. And although crypto is established now, I don't think it rose to the potential of replacing the things I was hoping to replace. And I was, again, the guy was a little bit maybe too conservative about it. I'm still maybe feeling that I'm not sure yet how AI sort of begins to resonate with consumers and whether it's actually going to deliver on the promise of value.

Ben Thompson • 01:00:02
it resonated in a way that makes, is it really about profitability? OpenAI is doing billions of dollars in consumer revenue, even setting aside the enterprise aspects.

Horace Dediu • 01:00:13
Yeah, but the problem is there's a huge, huge bill in CapEx that we're having to work through. And I actually wrote the piece this week, which is going to come out maybe today, tomorrow. where the valuations in AI shouldn't be PE ratios, they should be EV over FCF. And that takes into consideration the CapEx. And when you look at it that way, even though pricing has come down on the hyperscalers, they're still quite expensive. But I think the cash flow question is the fundamental question for AI. It's not just revenue. You have to be able to prove that you can generate free cash flow. And so I'm a little bit reluctant to commit to this because I don't feel I'm an expert on AI. But I think that my instincts and my data tells me that it's not going to be as fast as… Again, please tell me I'm wrong, but I don't understand the consumer business yet. And like, what is the main job to be done? I hear about consumers using it for chatting and being a companion. Is that a big job? Is that the main job? What other jobs do consumers anticipate? Or what do we expect people to do different? That's not clear.

Ben Thompson • 01:01:44
Yeah. Well, I think this is a very interesting point here, because I think the AI has clear productivity gains. But I think the history of software shows that consumers don't pay for that. To the extent they pay for productivity, it's they will buy devices. That has been Apple's sweet spot. It's software differentiated hardware. But every sort of startup that sort of starts out trying to serve the consumer with a productivity application ends up pivoting to enterprise. Because the enterprise is paying employees, so they will pay to make those employees more productive. And that's probably a big thing that Anthropic has sort of gotten right, is being sort of focused on that. And if anything, OpenAI has probably been distracted and has been hurt by the consumer uptake. At the same point, that's also why I take their sort of device forays somewhat seriously, just because they may be coming out of backwards, as this is actually the only way to make money from consumers in the long run. And if AI is that good and integration still matters, maybe there is a bigger opportunity here than people might expect.

Horace Dediu • 01:02:50
I agree. And I just feel uncomfortable about calling it this early. And I feel uncomfortable about how it's going to end up. It's like it's not even the phone business, I would say. It's more, remember, because phones were around a long time before 2007. We went through multiple, multiple huge debates about spectrum, about operating systems, about who controls network operators versus devices and so on. It wasn't at all clear until 2010 or 12, around then, what started in the 70s as far as what that industry would shape up into. And also the same you can argue for social media where the early pioneers were all disappeared. And so even though that's a behavior that's now extremely saturated in society and extremely consequential. So the other bigger question, though, I think that you need to ask the right questions. The real question to me for AI is, is it modular or is it integrated? And I think you hinted at that with, is the device together with the software? But the other layers would be like, if it's like infrastructure, and if I think it was Bezos, it's like a layer. If it's a layer, it's okay. It's like operating system, super important, super valuable, but maybe not easily valued by the user because it's background. And it's not something that you decide with you don't, it's kind of given to you. So there's all these questions about, does it commoditize rapidly? How soon do you reach a point of good enough or a point where you can't go beyond because you've reached the limit of the algorithm? So there's a lot of debates. On one hand, you're watching all these, reading all these papers, people making claims and counterclaims. And then at the same time, you're trying to figure out the economics. You're trying to figure out how long will it take for these chips to depreciate? So what is the horizon for a lot of the capital that's in use? So there's a lot of questions I still don't feel comfortable with. And so I think the safe thing to do is to kind of assume it's going to take longer than anticipated.

Ben Thompson • 01:05:15
That's my assumption. This is a point I've been coming back to again and again. What's so intriguing now and also very uncomfortable as someone who has to, like, give takes multiple times a week for weeks on end is the high degree of uncertainty. And I think it's layered on with everyone has opinions. But also, I think there's so much tenseness in the air because of the amount of money that's at stake.

Horace Dediu • 01:05:43
Yeah. And that, by the way, makes me a little nervous with the Christian philosophy. Let me put it this way. There's a couple of tests, litmus tests. And one of the things I immediately came to as a conclusion with AI was that it was very much sustaining. Now, the question is to whom. It's certainly disruptive to some job categories, but is it disruptive to technology companies that are extremely wealthy today and have a lot of money? or a lot of capital. So the test for that is, is this something that repels incumbents or causes them to embrace it?

Ben Thompson • 01:06:17
Oh, that's great.

Horace Dediu • 01:06:18
And control it. Because the key thesis of disruption is that incumbents are allergic to the new thing. They'll either say like, okay, this is uninteresting or not for us and not for our customers. Or if they look into it and say, We can't make money with this. The components are too expensive. It's the bill of materials is too high. We can't build millions of these profitably. None of it makes sense. We know the price is going to go down because of competition. So our margins are going to disappear. You keep rationalizing why not to do it. And so my take on AI is like when people said that it costs hundreds of billions of dollars to create a leading model, I said, well, what startups can do this? I will put it to you. I did this with mobility. I would ask myself every time, can a company in Africa today become a global giant manufacturer of mobility vehicles? Well, if it's bicycles, yes. If it's scooters, probably. But when you start to talk about cars, probably not. now when you when you then ask the same question who can make this thing now there's some hints that hey open source guys are doing something in china and like i get interested i'm much more likely to get excited but if you're telling me it costs half a trillion dollars to be in the game i say well this is a big boy game let them play at it but that doesn't make it disruptive no exactly i think

Ben Thompson • 01:07:52
it's such a compelling point that all the big players are spending so much money, which affirms this was my initial take at the beginning, too, is this seems like it's going to be mostly a sustaining technology. But you've sort of pointed out, along with a lot of Apple observers, Apple's getting the greatest deal in the world here. They don't they they don't need to invest

Horace Dediu • 01:08:13
all this money. Yeah. So that was I was getting beat up for two, three years about this because

Ben Thompson • 01:08:18
speaking privately mostly. But does that suggest it might be disruptive to Apple also? that it's not just that they're getting a good deal.

Horace Dediu • 01:08:26
So I was speaking in New York to investors once in January recently, and I said this. What Apple is not allergic to, what they're most attracted to, is having the interface between humans and computers, having or controlling, whatever you want to call it. But they've always lived on being right there. That begins with the Apple II, where they had the interface, the first personal computer. The interface happened to be a keyboard and a monitor. Then they moved into being a mouse and a bitmap display, and then they moved to a scroll wheel, and then they moved to multi-touch and gestures. And all of these things, for each of these major leaps forward, platforms were built, empires were built. Now, they happened to be competing ones, but they usually settled into one or two or three. But in the case of, so we had the DOS era, we had the Windows era, we had the iPod and we have the iPhone. And I said, if Apple wants to be on the next thing, it's going to try to figure out, well, how are we going to communicate with computers better in the future? And a lot of people said, oh, it's voice. And that was all the Alexa thing. Some people said it's watch. Well, that kind of helps, but it's very short duration, glances only. So we kept debating and debating. And then the big thing, which also captured the imagination of Zuckerberg, was like, what if it's a visual interface, but it's really head-mounted? And so I still am a firm believer that that's the next frontier is human interfaces. And then in the middle of that, somebody comes along and says, hey, how about we go back to the command line?

Ben Thompson • 01:10:08
Right.

Horace Dediu • 01:10:09
It's like 1976 again. I remember those years fondly, well, at least the years when I was using a command line for everything on a computer. okay i'm still know by heart all the unix commands i could right now open a terminal and do anything i want the point is though that that's what the state of the art of ai interfaces was at least a few years ago and then it's slowly moving into the voice but when i when i when i asked why would apple want to go there the answer is they wouldn't there's no challenge in figuring out anything well

Ben Thompson • 01:10:44
So here's sort of the counterpoint, though. But if this is a completely new type of computing, right, and it's almost like AI is like business versus like traditional computing is like engineering. It's very deterministic versus probabilistic, right? And if you have to go back to the beginning, Apple did start with the command line. Like the Apple II, you had to type in what you wanted it to do. And I guess in the short and medium term, I think Apple is clearly fine because you need a device to access AI, and they make the best devices. The question, though, is if the way to interact with AI has to go through that progression from command line to WIMP interface to multi-touch to whatever it might be. By the time we get to multi-touch, could you have gotten there without having gone through the whole path? And at that point, what would be a problem?

Horace Dediu • 01:11:32
Yeah, it's unlikely. And that's unlikely. And just to complete my theory, I said Apple would lose AI if AI becomes enabled through a new interface that is low end, that is voice only, which I create an equivalence here. Voice, the bandwidth of voice, even of type, it's a little bit more than typing, but it's like a 1200-baud modem.

Ben Thompson • 01:11:58
Yep.

Horace Dediu • 01:11:59
Your eyes have the interface of megabits per second. The bandwidth of the optic nerve is astronomical compared to at least the parsing necessary to decrypt words. Okay? The brain has to actually work very hard in cognition to understand what's being spoken.

Ben Thompson • 01:12:26
There's an analogy to AI here where we're stuck on these tokens that are like words. It's like it's very inefficient compared to how this could be done.

Horace Dediu • 01:12:32
Exactly. Words are terrible for the interface because they depend on cognition, not on perception. Perception always wins against cognition. This is why people are not thinking, they're acting and reacting. So there's a huge gulf between these worlds. And Apple's always very conscious of the perception, very, very much from the beginning. It's about really triggering people and perceptions. So my point is this. I'm not dismissing the low-end disruption, but it has to evolve. So it has to convince the world that you need to put aside your phone and we're going to work with you as an agent. And the metaphor there is like, what if you had, there's always this great thought experiment, ask what rich people do today and ask how can that be done by billions of people? And almost everything in history has gone that way. So we live, an average person lives far better than any king would have in the Middle Ages or even 200 years ago. And today, if you think about rich people and how they lead their lives, and seriously, somebody should do jobs to be done analysis, just follow a person who has at least net worth of 100, maybe 200 million, and see what they do all day. And to see to what extent they depend on other people. Now, I'm sure, for example, for travel, they don't go on search for their flights. I'm sure when it comes to getting to the airport, they don't figure out, should I drive or take an Uber and deal with parking? They have an assistant to do all that. And they used to have an assistant to do a lot more, but like that became part of the phone. But the thing that is disruptive is that the interface for most people will not be a screen, but will be effectively like almost a year piece with maybe other displays around you. Maybe every display is your display. And then it becomes more like, hey, I'm your best friend. I'm your companion. I'm your assistant. I'm maybe even your lover. And people will totally trigger on that because companionship and all those aspects of interacting with other humans, those are far more powerful hormonally than any cognition or perception. Then AI has a shot, which is why I always think that the AI boyfriend or girlfriend is a disruption potentially. But we got to be thoughtful about how we get to transition through that into something that you trust, something that actually, you know, you will be divulging information to this assistant that is so personal that they'll know more about you than you know yourself and more than any doctor. So we got to be very careful to earn that trust. But my point is that, yes, if that interface becomes so pervasive or so subliminal or so non-visual that Apple would find it like, we cannot value here. We just don't, there's no surface for us to paint anything on. Then it's a worry. On the other hand, I also say that the smart in the smartphone can get smarter. There's no reason why the phone couldn't be the best AI device the world could imagine. Maybe it'll take 20, 30 years to move beyond it. And even so, I still think putting a 4K display on top of every retina is more interesting than putting an AirPod in your ear. But that's just me. I could be wrong about that. I have a blind spot because I love visual stuff.

Ben Thompson • 01:16:10
Well, this is the 50 question, right? You talked about AI in 50 years, and if you mark it from, say, 2003 or 2017 or whatever you might do, we're here in the context where the reasons was Apple is turning 50 years. And the MacBook, to your point, has been the device all along. MacBook Neo, arguably the most perfect Mac yet. Where is Apple in 50 years? Are they still around? And what's –

Horace Dediu • 01:16:34
Yeah, so again, I'm writing all this stuff on my blog, and I'd love to have conversations there. But I even asked today, I asked the question, what are the odds that Apple survives? And the data, again, I go back to the data. And I point out that actually being in a Fortune 500 is very fragile. In other words, I call it the unfortunate 500, those who are there. Because you tend to live a very short life in a Fortune 500. But if a company reaches 50 years in the Fortune 500, it's most likely it's going to stay forever. Interesting. Because there are a few what I call immortals.

Ben Thompson • 01:17:21
Yep.

Horace Dediu • 01:17:22
They've been around since like a century ago. They're still there. Oil companies, IBM, a bunch of railroads, stuff like that. And Apple actually was born in 76, but it didn't join the index until 83. So I'm kind of using the clock from then. So there's this sort of, as opposed to human beings who live long lives, but we know what the limit is. With companies, they're very short lives, except for a few that are becoming more. And most people are fixated on those and don't realize that the tens of thousands of bodies that are buried to make those other ones survive. It's not at all guaranteed that if you have even a big business, you're going to survive. So there's myth about corporations being somehow superhuman and all that. So there's a duality to it. So my first observation was that already by the fact that it's made it this far, it's kind of buying itself a long life. But the more important thing, and I think there isn't really a good theory of Apple that's been published, but I think the company is an exception. in many ways, every time I dig into it, and I've been digging in 50 more years, is that there's an operating system there, mostly put in by Jobs. The Jobs, not the original Jobs, but Jobs 2.0, which came in in 97. In fact, Jaws, Josie Black, said that all these people are trying to write thought pieces on Apple 50, say anything before 97 doesn't matter. It's true. There's two apples. And the second apple is the one that actually they thought hard about, how do we make this survive? Because they've been to near death. So they started to think about, become introspective. You know, they say that unexamined life is not worth living. When I was at Nokia, I even pointed out to people that because of turnover, there is no institutional memory. There are some old timers maybe who remember things, but most people have not been there long enough to know what was the causality of success and failure for the firm from the beginning. They might read about it. There's a lot of lore, but most of it is written by journalists. No offense, but it's written by people who want to tell a good story. and that's how you get published.

Ben Thompson • 01:19:48
Yep. Well, to that point, when I was at Apple, I was at Apple University, which was, this was sort of the, at that point, this was sort of the goal was to sort of, what makes Apple, Apple was the point.

Horace Dediu • 01:19:58
So that's the evidence of introspection. So they're trying to think, what do we do to pass this on? And I think they were conscious of it in Steve's time and I think Tim has had this on his mind for the whole tenure. He knows that he's the custodian and he wants to make sure it doesn't get broken by the next. That's another thing, entropy. Companies lose their souls and that happens naturally and it should be somehow, you should try to preserve it. I even wrote a piece years ago about what makes companies that last hundreds of years. There are companies, they're hundreds of years old. They're all private. In fact, they're all family owned. So it's genetic. It's like I pass it on through my blood. That is what keeps the thing together. And the oldest are Japanese, and they actually have a trick up their sleeve. The Japanese last for hundreds of years because they adopt people as adults into the family because they don't trust their own children because they feel that they're born with a silver spoon in their mouth and they're not going to fight very hard. But if you brought into the company as an adopted son, then you might, with a daughter, you might fight harder. And this is, so it's fascinating to understand longevity, mortality, and all these things. So Apple, I think, is cognizant. I'm not sure I'm good that you understand the problem, but at least they're aware of the problem, and they're trying to figure it out. There's other things about the story of Apple, which is that the focus and the sort of the idea that the customer is the most important thing. One of the mantras I have is that the value of the firm should always be the measurement of the creation of the customer. This comes from Deming. It's not particularly novel, but he had the great phrase, which I remember, which was that every number on a financial statement needs to be multiplied by something called customer satisfaction, because that tells you whether those numbers are going to happen in the future. So the customer comes back based on their satisfaction. There's nowhere in the financial statement where that number can be read, however. And by the way, there's a numerator. You divide everything in the financial statement by the customer dissatisfaction. And so those are the two most important numbers that companies should have. And everyone should look at them as a KPI every single day. And yet, hardly anybody knows what they are or what they should be. So the customer, and that brings you back to the hard question of like, is Apple measuring its success based on how much customers are satisfied? And you know what? If you listen to Tim Cook and the CFO every quarter, they always bring it up, probably five times. And not once is it ever written about in any of the commentaries that follow. So analysts are not paying attention. Pundits are not paying attention. But Apple somehow continues to pound the table on this. It's just because we've become kind of, it's glazed over. It's like, yeah, yeah, everybody says they want to have happy customers and stuff. But I do believe that Apple cares about it a bit more than others. And that, again, unlocks a lot of puzzles because when they went through all these weird decisions and other companies went different ways. Microsoft was an example. You know, Google was an example. And I highlighted some of these facts. You know, the fact that Google doesn't have people paying for its product. People are its product. So all these things affect your meta. Meta is the worst. They affect how everyone in the organization behaves.

Ben Thompson • 01:23:43
Well, we offer a point on this, which I think is interesting. We've got way long. I have 47 questions, so you have to come out again. We got to talk about more about this. But one of the cases I've always made about meta, and I think this applies to Google as well, is the reason why they are much stronger than people think is precisely because they have created their own customers. and in that a huge portion of Meta's revenue base are advertisers who only exist because Meta's advertising exists. And so that's why they're almost anti-fragile in a way where you go through something like COVID or you go through a large CPG boycott of Meta and they get stronger because that's just more opportunity for the customer base that they've created.

Horace Dediu • 01:24:28
Yeah, I don't want to pick on Meta. Honestly, I…

Ben Thompson • 01:24:31
But I'm saying, I think your point is a good one. It's just like, it all depends on who you look at as the customer.

Horace Dediu • 01:24:36
You've got to be careful. Yeah, totally. You've got to be careful about defining the customer satisfaction. I do believe there are many people are very happy being users of Meta.

Ben Thompson • 01:24:46
Well, the advertisers are very happy being advertisers of Meta. That's the key thing.

Horace Dediu • 01:24:49
Yeah, there's value. There's definitely value there. I'm just pointing out the fact that if Meta wants to make a device, you might think about it differently.

Ben Thompson • 01:24:58
Right. Well, that's probably why they're bad at devices.

Horace Dediu • 01:25:01
Exactly. But no matter how many great people you hire, you poach the best designers, you poach the best product marketers, you poach the best engineers, industrial or otherwise, to come to your firm to repeat what Apple does. Somehow, it doesn't come out as good. Now, there's a question about this. Why? And even many people at Apple, in particular, Johnny Ive is famous of saying that you take me out of Apple and I'm completely useless. He said this. And then, of course, he goes off and works on Ferraris and works on devices for opening eye. But there are many people left who said the same thing to me privately. They said it's a system that somehow brings out the best in people, but it's still your cog in a very big machine.

Ben Thompson • 01:25:53
It's also a unifying theory. Some of the points, your concerns about AI, which is there's a bit where it's precisely because Apple devices are so expensive and they make so much profit per device that that's actually a key to their long longevity because they have to continually earn it as opposed to we're going to sell at cost just to like expand our user base or whatever it might be. It's sort of counterintuitively increasing the challenge and having to earn money makes you last longer.

Horace Dediu • 01:26:22
Yeah, in many ways, I'm always impressed by, I forgot who wrote this, but basically the idea of Apple being the underdog is a superpower. For a long, long time, they were the guys who were always on the siege. They were always, some of them, stronger than them. First, it was because they were the rebel against IBM, then it was the rebel against Microsoft. Even Android versus iOS, there's a sense of like, okay, we got to fight our way against commodity. And I think there's a sense also that when Apple makes a mistake, it gets punished much more severely than when other companies make exactly the same mistake. I give the example of Samsung Galaxy, which I forget which year it was, in 2016, 17, 18.

Ben Thompson • 01:27:19
Yeah, the phone would really blow up.

Horace Dediu • 01:27:22
It would catch fire. And every time I got on the plane for about a year, I was told, if you have a Samsung Galaxy device, please make sure it's turned off. And so you had a situation where a product was effectively declared a menace to society. and i was watching carefully the pnl of samsung throughout the period and nothing happened they didn't lose a beat um they it all washed over but if apple a couple years earlier had the antenna gate but nothing really major happened except there was attenuation of the signal because of the way the antenna was positioned. You're holding it wrong. And they had the lawsuit and you had to sell bumpers to solve it. I'll give these bumpers away. But more or less people had freaked out and said, that's it, it's over. Apple is dead. They just shattered their credibility, their brand, everything is destroyed. How could they? How could in the mockery and just piling on? And now we're seeing the same thing. Oh, they're going to have advertising. They lost their way. I don't know how many times I hear this, but actually I think that's a positive. I think for the company, whenever they're beat up so much, it puts them always on their guard and you will have an instinctive intuition about making sure what you ship is not going to be damaging to the brand. It's one of the reasons I think they got out of the car business is possibly, I know that's one of the reasons they wouldn't get into the micro-mobility business, even though it a sense for Apple to make a scooter or a knee bike or something really neat. But if somebody falls, breaks the bone, everyone will say Apple is the worst company in the history of the world. It's for sure. And I said this to an interview with CNBC recently. I said, when the new theory shifts, it will be declared the worst AI in the history of the world, even though underneath it's Gemini, whatever it is, it doesn't matter. It will be for sure declared pathetic. This is true of Maps. This is true of Siri. This is true of any product that Apple ships. It's always the worst in the world. The critics are absolutely vicious. And that's their superpower.

Ben Thompson • 01:29:54
Well, that's a great note to end on. I literally have like, we got through like a quarter of my questions here. But the good thing is you are back in the analyst game, as you mentioned, a Simcoe one. There is a to your point. You can go there, start reading. There is free stuff that you could pay. I like how you're doing it. You have lots of snippets of articles. And then if you pay, then you get your sort of your commentary on it.

Horace Dediu • 01:30:17
Yeah, there's basically two categories. One is like reactions. So I quote an existing piece and I have a take called Nota Bene. And then there's research pieces, which are longer and more in the traditional style of me kind of coming up with some original material. The difference was that because of the days when a lot of my articles would be like, well, this is an inflammatory or this is an outrageous comment. Let me dissect or rebut this point. Instead of doing that as an article, I sort of categorize this as a reaction to the news. But it's not a new site and it's not. And I think my takes are particularly different, I would say, hopefully different than most takes. So there's hopefully value in the reaction, but also value in the research, of which there will be plenty, a few hundred a year at least.

Ben Thompson • 01:31:12
With your Asymco One plan. So I'm actually a big proponent of the way to think about business models is on the Internet, you need to maximize revenue per user. If you're not going to be like a massively scaled player. And you're taking that to the extreme. It's $100 a month, $1,000 a year. How did you think about pricing when you were launching this?

Horace Dediu • 01:31:32
Well, the reason is because, again, we had a lot of feedback over the years. And I realized that the core audience were people who were very interested in investing. And although there are plenty of developers, there are plenty of enthusiasts, there are plenty of journalists and others. And we can probably segment pricing for them as well. but the the i think the core job to be done was that i was i was i was i was being thanked for um essentially preserving wealth for people um and and the fact that that a lot of people weren't it wasn't about saying buying this or that thing but it was more like holdings for tough times

Ben Thompson • 01:32:13
don't sell apple that's how you make money is just hold on to the stuff yeah it's a funny thing

Horace Dediu • 01:32:18
He even told me things like if one of the best performing types of portfolios out there are estates. So when people die, their money gets held up in an estate for a while while they settle it. And they outperform everyone else because nobody's doing anything. So actually being holding is a very good strategy. Anyway, but the question of pricing, let me say this. There's a dichotomy people have in their minds between those who have and have not. And this becomes a political discussion. Because people say, oh, there's all these billionaires, and then there's everybody else. Well, the fact is there's actually millions of millionaires. Actually, in the U.S., there's 24 million people with net worth between $1 and $10 million. Now, a lot of that is in real estate and businesses. These are the primary asset classes. But there's a lot of money also being held by retail, what's called retail investors, in stocks and funds. And the thing is that they don't get access to great financial advice. Most advisors are positioned for people above 10 million or even above 100 million net worth. And those, by the way, the numbers of those people goes down exponentially. So there's like 900,000 people who are 10 to 100 million. Then it's like 40,000 people that are above 100 million and so on. So you're like dropping off a zero from the number of people when you get to the higher numbers. But if you're going to be a fund manager who wants to advise people on wealth, you're looking for someone 10 and above. So there's a sweet spot, in my opinion, between 1 and 10. And of course, there's below 1 as well. But those are people who are usually younger and are looking to start and are looking for buying opportunities, not holding as much. So the thing I was puzzled by is that if you look at the instruments, at the information sources out there, you have institutional, which are the usual Wall Street suspects. And then you have CNBC, the Wall Street Journal, business magazines, business newspapers, and so on. And their business models are advertising. They're not trying to help you out with, and the narrowness of my scope is like, I really want to just talk about Apple. And I found that there are people who were grateful, like I said, and I figured if there's a handful of those, there's probably out of, like I said, by the way, Apple is one of the most widely held, both for retail and institutional. So just institutionals alone, there's over 6,500 institutional owners of Apple. That's public information. I don't know how many retail investors there are, but there are about 60% of the money invested is retail. So there's millions and millions of people who have millions and millions of dollars of Apple. And if there's so many people like that who are not getting access to great research, We're not getting to communicate, except, of course, they're seeking Alpha and a bunch of other places. But it's all like it's not high quality. It's not concentrated wisdom.

Ben Thompson • 01:35:50
Right. Well, to your point, they're the ones that are the most susceptible to the underdog narrative. And so your your case, I guess, is yours talking it with your own book that you do believe in Apple's future. And the point of this is to convince other people that as long as you believe they should believe, too.

Horace Dediu • 01:36:08
Well, I got some skin in the game, but the main point is that I just Warren Buffett said, if you're going to put all your eggs in one basket, you better watch that basket. And Philip DeWitt taught me another thing. He told me about his, he has a good audience, okay, who are Apple investors. And he said, although he's more news driven than analysis, right? So he does a lot of news every day. He goes through and exposes the most interesting views, which are very valuable. But he said, when we sat on stage with Laura Martin in New York, he said almost all of his audience has all of their money, like 90% in one stock, which is very highly concentrated. And so, again, if there's a few there, there's plenty of them everywhere. And I just feel that this portfolio is not what I do. And I don't try to tell people to balance things out. And I will do a little adjacency. I'll look at I have Google and the usual suspects, but mainly it's Apple. So, yeah, I think the idea of having a deep, deep 15-year track record where you've looked under every stone and you look at it at every angle. You've looked at culture. You've looked at product. You've looked at history. You've looked at financials and technology and competition. You looked at everything. I think there's some value in saying, hey, if there's something about Apple, there's probably I've probably known it or forgotten it.

Ben Thompson • 01:37:51
It's great that you're back writing regularly. Not that you stopped, but the pace is very high. It took me quite a bit of time to go back even just to February when you really launched. And given that, now that you're back in the game, I would love to actually get to the rest of these questions sometime. It'd be great to talk to you again.

Horace Dediu • 01:38:09
Absolutely. Happy to connect again. Appreciate your continuing efforts. I think I really feel like we're sort of birds of a feather. Absolutely. It's wonderful. Thank you.