Apple's Vertical Integration | iPhone Lead Engineer, Andy Grignon

Now, when we started the phone, it was kind of like I said, we had this kind of half step, right, of let's try to make it. We kind of— but we had in another organization bought this touch— what was it called? FingerWorks was the name of the company. And they had this idea of building a new kind of keyboard as an accessory to the Mac.

Mm-hmm.

That's why we bought them. Somebody— it wasn't me, it was somebody else very smart. It wasn't Steve either. had put the two together. It's like, oh, shit. What if we were to marry this— you know, it was on an opaque surface. What if we were to marry this finger technology with, you know, the screen technology?

Right.

And that's, you know, a couple prototypes later, we got a thing. What if we were to actually make the phone that? Because this other thing we built is kind of stupid. And so that became a big fight. I can't go into all those details, but it was a fight. But we ultimately ended up on a path. Now, when you think about iPhone, this was whatever it was, 2004.

Right.

There is no reason that product should have succeeded from a technological perspective, right? I mean, not from a business, from a technology perspective.

You mean the spec that you were laying out was too expensive? It was—

Think about it. I mean, so normally when you build a new thing, let's say we're gonna build a laptop, you start with a laptop, you know, kind of a design. that you've already had, you change a couple things, add a new this, do that, give it a new color, bam, there's your new laptop. Great. This is the first time we had ever tried to build a phone. So we didn't know shit about that. We started, because we're Apple, we don't start with off-the-shelf chips.

Right.

We don't go to Samsung and, hey, give us the sphincter valve and Newton nut chip. No, we have to say we want a custom chip because we want to put macOS on that chip. And macOS has this really crazy thing. We need 64-bit timers in that chip, and no other part had that. So can you add some of those and 2 of those and give me 3 of those? Like, give me all this other shit. We bundled it into a part. So we have a new chip from scratch.

So you're basically vertically integrated. You make everything from scratch.

Everything from scratch. And now we're going to add some antennas into this thing, right? Now we've— again, we've never built a phone. Let's do all this. Let's put an operating system that's never been compiled for ARM before. It's only been x86 and before that PowerPC. Well, actually, it never really existed on PowerPC. Then on top of that, let's invent a whole new middleware layer, a whole new paradigm of interacting with the screen touch and, you know, kind of desk avoidance. And on top of that, let's invent a whole new application stack that is built for a class of device that sits about this far from you. So every variable in the stack was new. So you couldn't fall back. I can't tell you, there was probably about 3 or 4 times where the program halted to its fucking knees because we couldn't— it was a straight up, we don't know how to solve this. Now, when you have a company like Apple filled with some of the smartest people in the world, the deepest bench imaginable, and you have a problem where everyone's just like, I don't know what the fuck to do, right? I don't know what to do. And those are Yeah, so we used to do these meetings with Steve and we would have a review. When we got closer, we would meet up twice a week. And yeah, I got basically yelled at all the time. So, you know, I was, if you Google me, I'm Fuck Chop. That's the famous name only because I put it on a business card. And, but I was fat fuck, dumb fuck. You were, at one point I was single-handedly ruining his company. I was like, that's pretty good power. Yeah, you should be the CEO. Yeah, shit. I should have been. And so, but you'd go into these meetings and then he would just like, right, he'd just like berate you. And then he'd just give you like a little compliment and you'd be like, ah, fuck yeah, everything's all good again, right? And, but you know, those were— so, but you'd go in because you'd screw stuff up all the time, right? I did certainly. I make mistakes.

People love pain.

They do. I love pain. Yeah, exactly. Hit me again. And then, you know, but the ones that really made you the most scared or nervous Was where you would kind of lay out a thing. You'd, of course, you wouldn't ever blindside him. It was always, you know, we'd investigated every avenue or whatever. And he would just kind of sit there. And then, you know, we had these rocker chairs. And him and Tim Cook would sit right across from each other and they'd kind of, just kind of like this. And they wouldn't say a fucking word. No yelling, no screaming. But that's when you knew you were in real trouble because he wasn't like apeshit insane.

Beyond chill.

Yeah, he was like, we've gone past that. You know, we've jumped that chasm. Now we're just in the crazy man land. And that's— but it was really interesting because you knew, well, on the one hand, you knew, like, all right, this is the time to really kind of think through the problem.

Yeah.

And not get crazy and psycho and lose your cool. But it's also the time when you have to really dig deep. And those are the times where you did things that were unusual. You took advantage, you leveraged, at least in Apple's case, we would leverage the— because of iPod, the connections we had into our vendors. So we pulled in people from— in one particular instance, I pulled them in from Korea. We had our chip guys, and they all sat and they single-stepped through this chip because we actually had a chip-level problem. Again, because we used new silicon, right? So now when you have problems that are unsolvable higher up and you have to have the people who wrote the code that made the chip that's in the thing to single-step through, you're like, what? It was that bad. It's pretty interesting.

You essentially, you guys had access to every part of the supply chain from the software assembly guy—

All the way down.

To the guy who makes the Gorilla Glass.

Exactly. And that was actually, that's not an advantage that most companies have. Nowadays, we've solved a lot of those problems. We kind of know once you pave the road initially, now you actually get to kind of build on the shoulders of giants.

That's why we get to see Android phones and things like So, maybe talk a little bit about just the constant debate of vertical integration versus best of breed. Apple's obviously in that former camp, but many companies now, in fact, most companies, sort of have chosen the latter approach.

Yeah, and I guess, I wrestle with that, because on the one hand, your value as a startup or as a person making a product product, you know, your value is in adding your special sauce. Should the entire stack be special sauce? No, I don't think so. To create iPhone, the entire stack had to be special sauce. Everything had to be custom. But, on the other hand, I'm also vehemently against just like, hey, let's take this part over here, this open source project, let's do that, and then swizzle them all together and let's call it a this. You know, this is Instagram for cats, or whatever. Like, I don't get— I think we have too much of that happening today. where we're not innovating and moving the needle forward. I think we need to pick the areas where we invest in to create the special sauce, but I do not advocate a complete ground-up redesign of the entire thing. You have to be smart about it, and that's kind of a tricky line, I think.

Yeah.

We were told when we were— so I was responsible at the time for iPhone for what I— the way I describe it is I'm responsible for all the things you hate about your phone. Like, at the end of the day, like, if you drop a call, that's this guy. It's because he couldn't pair to your, your shitty Bluetooth in-car radio, this guy. When you can't join the airport Wi-Fi or whatever, this guy. Nobody ever, by the way, calls you up and says, hey man, I just had a call for like an hour and it didn't drop.

Right.

Nobody says that. Everybody says like, fuck you, you drop the calls at this part in PageMill every day, like, fix it. Right? Those are the kinds of problems you get. But to create that, we didn't know what the hell we were doing. The very first one, like, I don't know how to make a phone. Like, none of you are born knowing how to make phones, unless you're some sort of crazy person. But, how do you do that? Why? And, oh, and then, have Steve tell you, you don't get to hire people that make phones. Ah, isn't that a great little twist? How awesome is that? Well, it turns out it's pretty shitty, but it forces you to invest in building the IP for yourself instead of being reliant on somebody else for such a core part of something like that.

Yeah.

I was the guy that had to pitch visual voicemail to AT&T. So visual voicemail, when you're a software person, is a really simple thing, right? I mean, it's IMAP, right? It's a different face on top of IMAP. And, you know, the message is effectively the attachment. the metadata kind of surrounding it. But when you get in front, I remember pitching it. It was just like a room like this, but a lot more people in suits. And it was like an awful wedding. There was like Apple people over here and AT&T people over there. And you get up there and you're like, ah, here's this, that, and the other thing. And this is why visual voicemail is so great. And we've got to do this. And then I remember the woman who was the CTO of AT&T at the time, this woman named Chris, very nice. But basically, let me tell you why everything you said is bunk. And here's this guy to basically shoot down everything you just said. We can't do it because of this, that, or the other. And we argued about it, and at the end of the day, we were at a stalemate. So I went and I cried to Steve. I was like, we couldn't get it. Like, we don't have an agreement on visual voicemail. So Steve solved it at the CEO level. Turns out it wasn't that hard. It wasn't the whole big complicated thing.

He took credit for it.

He did, exactly. But then, you know, it was interesting. That got me thinking because When you kind of roll back to the very first day when everyone's sitting around thinking about a new way to do voicemail. Remember, back in the day, voicemail was done with like, hit 4 to go back, hit 7 to delete, or whatever bullshit you had to do. You didn't have like a nice fun thing. You had to like have all these scripted commands. And if you go back, let's say I'd hired somebody from Nokia, like to be on my team building the phone. Now we're having the brainstorming session about, hey, let's like, let's make this new way to create voicemail. Let's do this, that, and the other.

He may not know how to do it.

Yeah, this person inside of the team would be like, guys, it's really complicated. Like, we can't do it because, you know, sphincter valve and Newton. Like, it's complicated. And we would have shot it down internally. So Steve was actually smart. I don't think he saw that particular thing, that instance happening, but I think he knew we had to make our own kind of mistakes.

More sort of philosophically, you almost want to— you want people who don't know what they don't know. In fact, one of the reasons why this school is 18 to 28 is we I have invested in a lot of things that didn't work. I've seen a lot of things that didn't work. So when you come to me with an issue, I first have to sort of fight off this temptation. Oh, it didn't work 5 years ago.

Right. Why would it work now?

Right. But things have changed. And in fact, I remember the first time I saw the visual voicemail, I was like, this makes so much sense. Why didn't they do it before? I did not realize it's that painful to do.

Well, we didn't do it before because there's like 6 people in Des Moines, Iowa that need to have this particular feature. And once that becomes ingrained, like it becomes, you know, the sacred cow. Oh, we don't get to touch that because of those 6 people in Iowa.

Yeah.

And then it becomes this morass, especially if you spend time— and I would urge, by the way, all of you guys to spend time at a big stupid company to see this firsthand, because sacred cows exist, and you find people who are petrified of touching them or getting anywhere near them. You guys have the luxury— and I call it a luxury, because it is— of having those fresh eyes to go and be like, yeah, fuck it. I don't care about that. I've got no vested interest in seeing 6 people in Iowa? Like, no, no. And so I think that we bring a naivete, but that's necessary to innovate. Now, I would argue against this if we're building bridges and planes and all sorts of other shit. If someone's life is on the line, let's have those sacred cows. I mean, if you're building a new way to buy groceries, fuck it, try it.