CNBC Interview: Apple Executives Johny Srouji & John Ternus on Chips, AI & Innovation [Summary + Transcript]
CNBC reporter Katie Tarasov interviews Apple's head of silicon, Johny Srouji, and senior vice president of Hardware, John Ternus, to get insights into the company's silicon chips business.
Read the summary of this interview:
Apple Executives Johny Srouji and John Ternus Talk About Chips, AI, and Innovation: Summary powered by Fireflies.ai
Outline
- Chapter 1: Introduction and Roles (00:13 - 01:03)
- Chapter 2: Collaboration and Unique Approach at Apple (01:04 - 02:19)
- Chapter 3: Evolution and Focus of Chip Development (02:20 - 05:30)
- Chapter 4: Impact of Chip Shortage and Response (05:31 - 06:52)
- Chapter 5: Expansion and Importance of Chip Production (06:53 - 08:04)
- Chapter 6: Strategic Planning and Mitigating Risks (08:05 - 09:32)
- Chapter 7: Journey from A4 to M3 (09:33 - 12:42)
- Chapter 8: Importance of In-House Technologies (12:43 - 15:17)
- Chapter 9: Importance of Scalability and Focus in Chip Design (15:18 - 19:36)
- Chapter 10: Impact of AI and Machine Learning on Chip Design (19:37 - 22:56)
- Chapter 11: Future of AI and Machine Learning at Apple (22:57 - 25:29)
- Chapter 12: Addressing Critics and Embracing Competition (25:30 - 27:20)
- Chapter 13: Investment in and Future of Modem Technology (27:21 - 28:59)
- Chapter 14: Significance of the Three Nanometer Leap (29:00 - 30:20)
- Chapter 15: Risks and Rewards of Advanced Nodes (30:21 - 33:47)
- Johny Srouji manages the hardware technology center as vice president for Apple, joining in March 2008. John Ternus, a mechanical engineer by background, works closely with Srouji.
- Apple's unique approach is the integration of hardware, silicon, and software, working as one team to create the best chips specifically for their range of products.
- The team designs chips three to four years ahead of time, focusing on the feature set, customer experience, and optimization of software.
- The transition from using technologies from other companies to building in-house has been the most profound change at Apple over the last 20 years.
- Apple's focus is on creating the best products, utilizing the best technologies. This includes exploring options and having a diverse supply.
- In response to the chip shortage, Srouji spoke about Apple's ability to navigate the issue due to controlling their own chip design and working closely with foundry partners.
- Apple hires the best and works with universities to prepare the next generation of engineers. They also encourage their partners to scale manufacturing in the U.S.
- They focus on creating the best chips for the best products, not being constrained by anything else but the laws of physics. They believe they are not behind in AI, and have an advantage due to their ability to integrate hardware, software, and machine learning.
- The hardware and software teams worked together to develop an architecture for the Mac that was perfectly suited for Mac users, enabling them to create products they previously couldn't.
- The recent advancement to a 3nm chip allows for better performance and longer battery life. The company is aware of the risks that come with using an advanced node and takes calculated risks to achieve great results.
Notes
Read the full transcript of the interview below:
Apple Executives Johny Srouji and John Ternus Talk About Chips, AI, and Innovation | Podcast transcript by Fireflies.ai
00:13
Katie Tarasov
So let's start out with an introduction of each of you. I would love to hear who you each are and then when you started at Apple.
00:20
Johny Srouji
Johny Srouji and I manage the hardware technology center vice president for Apple and I joined. March of 2008. So almost 16 years getting there. I came to Apple with the purpose of building our own silicon for the iPhone. It was a very small team at the time about 40 to 50 engineers. And since then we have grown the team immensely. For the last 15 years we've been building a work-class team that is 100% laser focused on building the best chips for our products. And it's been a journey that I love.
00:52
Katie Tarasov
I want to hear more about that journey in just a second, But let's hear from you as well.
00:55
John Ternus
My name is John Ternus. I've been at Apple since 2001, So 22 years now. I'm the Senior Vice President of Hardware Engineering. And as you can imagine, Johnny and I partner together very closely. I started in product design, Which is a, I'm a mechanical engineer by background, And I've had the great fortune of working on pretty much every type of.
00:01
John Ternus
Thank you.
00:12
Johny Srouji
Thank you.
00:13
Katie Tarasov
So let's start out with an introduction of each of you. I would love to hear who you each are and then when you started at Apple.
00:20
Johny Srouji
Johny Srouji and I manage the hardware technology center vice president for Apple and I joined. March of 2008. So almost 16 years getting there. I came to Apple with the purpose of building our own silicon for the iPhone. It was a very small team at the time about 40 to 50 engineers. And since then we have grown the team immensely. For the last 15 years we've been building a work-class team that is 100% laser focused on building the best chips for our products. And it's been a journey that I love.
00:52
Katie Tarasov
I want to hear more about that journey in just a second, But let's hear from you as well.
00:55
John Ternus
My name is John Ternus. I've been at Apple since 2001, So 22 years now. I'm the Senior Vice President of Hardware Engineering. And as you can imagine, Johnny and I partner together very closely. I started in product design, Which is a, I'm a mechanical engineer by background, And I've had the great fortune of working on pretty much every type of.
01:15
Katie Tarasov
Let's talk about why you guys are here together. This is something that is unique to Apple,the sort of marriage between the hardware and the silicon. How do you guys work together on a daily basis? What is that like?
01:25
Johny Srouji
So the hardware,like John's team,and the technology,including the silicon,we work together as one team. And the software by the way. This is very unique to Apple that we have the hardware, Silicon, Technologies and software all in one place. We build integrated products that are fully optimized for the product because that delivers the best optimized software for the hardware. We get to design the chips ahead of time, Working with our partners from John's team and software and OS to exactly and precisely build chips that are going to be targeted for those products and only for those products. So we work together very closely as one team. We could define the chips three, Four years ahead of time together. And then my silicon team has the freedom to go and innovate for those products.
02:10
Johny Srouji
And we precisely target the feature set,the customer experience,and how the software is going to fully optimize. So no boundaries.We work together as one team.
02:19
Katie Tarasov
Anything you want to add?
02:20
John Ternus
Yeah, I mean, I think when I started, The way we tended to make products is were using technologies from other companies and were effectively building the product around that. We've always had an incredible design team and we made these beautiful products, But they were constrained by what was available. And I think one of the most, If not the most profound change at Apple, Certainly in our products over the last 20 years, Is how we now do so many of those technologies in-house. And, You know, Top of the list, Of course, Is our silicon, Which. As Johnny said from the very beginning,we're thinking about how are all these pieces going to come together to make the best possible product for our customers.
02:55
Katie Tarasov
Do you think that the regular Apple customer knows that the chips powering their devices come from Apple? And do you think they care?
03:03
Johny Srouji
I believe they know and I believe they truly care because here's why. We're not a chip company. But we have a very,I think,best in class,work class chip team. And the fact that we're working together and we're building that silicon exclusively for our products gives my designers the freedom to design for exactly those. The thermal envelope of the product,the shape of the product,the software how it's going to use the product. Without compromising design,without compromising focus. So by doing this,I believe we get best in class compute,whether it's the CPU or the graphics, Or multiple machine learning accelerators on the chip,and feature set with power efficiency. So users don't need to compromise if I want to get fast machine. That battery lifetime will be suffering. We deliver the best in class power efficiency chips with a rich set of features.
04:00
Johny Srouji
So I believe customers will care because the customer experience that we can deliver with predictable execution and making bets ahead of time, Only Apple can do because we are working as one team together. Again, We're not a chip company, We're a product company. So our North Star is to deliver the technologies that are gonna enable the best products. And that's how we operate.
04:21
Katie Tarasov
Do you think customers also care more now because chips have come onto the world stage in such a big way since the chip shortage?
04:28
Johny Srouji
I think maybe the chip shortage gave it more attention. It put it under the highlight. That our world today is digital.Everything at the end of the day runs on a chip. So that gave more attention,but it's been important for us since day one,which is why we started that effort. Back in. 2008,Apple decided that we wanted to create something. That we couldn't buy off the shelf. At the time we knew that we want to create a product,our vision. It required us owning and controlling the silicon,which is why we started that effort. And it wasn't common at the time. You had chip companies building chips,and you had product companies buying those chips for their devices. We decided that's not going to enable our vision and truly deliver the best customer experience. The recent chip shortage gave it a highlight, But we.
05:17
Johny Srouji
15,16 years ago,we figured we needed to own that. Before building our silicon,Apple always had an integrated hardware,software and operating system. But then when we decided to build the silicon,we took it one step further.
05:30
Katie Tarasov
One more thing about the chip shortage since we sort of ended up there right now and then we'll go back in history a little more. But do you think that then when the chip shortage did happen, It gave Apple some insulation against it because you were relying on yourself instead of as much on outsiders?
05:45
Johny Srouji
It depends. The chip shortage is based on what? The chip shortage could be based on manufacturing, Like you don't have enough scale. It could be based on some specific node that there is not,I mean process technology,whether it's 7 nanometer,5 nanometer or else. Or it could be guided by your own design. Capability given we own it. We own our design and we have a work class team. We were that was predictable. We work very closely with our foundry partners ahead of time. And by doing this,we can also navigate.
06:20
Katie Tarasov
Is there going to be enough 3 nanometer capacity now that the M3 is here and the A17 is here?
06:26
Johny Srouji
See, I can't talk about the capacity of technology or founders. That's for them to answer how much capacity they're building. But as it relates to our own chips, For example, If you take the latest M3 family of chips for the Mac, M3, M3 Pro and M3 Max, Obviously we're working with our partner, In this case, The A17 nanometer. And we believe they have the skill.
06:52
Katie Tarasov
How excited are you about the Arizona project?
06:55
Johny Srouji
Very much excited. I think,you know,Apple has always supported advanced manufacturing in the US. And we always supported having more silicon expertise in the US. We have thousands of engineers in the US. We also have thousands in Europe and Israel and other places. But thousands just in the US. We hire the best. We also work with universities on their curriculum so that their students, The next generation of engineers are prepared. And we also encourage our partners to scale their ability to manufacture also in the US. So I think it's exciting. It's a great move.
07:32
Katie Tarasov
Is there urgency to the risk of so much concentration of advanced chips being made in Taiwan?
07:39
Johny Srouji
So let me refer maybe to TSMC or Taiwan. We always want to have a diversified supply,Asia,Europe and the US, Which is why I think TSMC building Fabs in Arizona is great. And other founders are doing the same. There are other founders,Samsung and Intel building in the US. So I think that's great. We do rely on TSMC for a good part of our internal chips. And when I think about it, It's actually very complicated. Those transistor technologies are very advanced and complicated, But I still it down to a few principles. We always want to deliver and build the best chips on the planet for the best products. So that's the North Star. And that means you need to have access to the best tools, Including best transistor technology. So that's extremely important. You want to have a foundry that has that technology.
08:28
Johny Srouji
And you also want to have a partner. Where objectives and goals are aligned,they're reliable in terms of execution and delivery,and they can scale with your needs. Whether it's design needs and requirements,because you have multiple chips for multiple products,or the scale of volumes. We've been working with TSMC for over a decade,and they've been very reliable. Now having said that,we always explore options,we always want to get access to the best technology. And if some foundry has that and they meet our requirements,we always explore it,including outside of Taiwan. So,I think there is goodness to diversify,but you always want to go to the principle,where can you get the best technology and can they scale.
09:09
Katie Tarasov
Apple's Known For Planning Ahead Well And Insulating Itself Against. Issues And So Is There A Backup Plan Should Geopolitics. Spiral In The Taiwan China.
09:19
Johny Srouji
Area? I Can't Answer Future Plans. But What I Can Say Is I. Totally Support What You Said We Always Look Ahead. We have strategic bets and we are very careful with our planning.
09:32
Katie Tarasov
So let's go back in history a little bit. I want to hear about how quickly it ramped up. And I want to hear about the moment that the A4 came out. You joined, It sounds like there were 40 or 50 engineers. And then two years later,Apple's first silicon came out. Talk me through that ramp. How quickly did you have to ramp? And then how did it feel the day they before came out?
09:50
Johny Srouji
Felt great. Proud. Delivering something great and you know we built something from scratch. The iPhone had the first chip, A third party. But then Apple decided,we decided we're gonna control our destiny in silicon. So there were many good reasons that we wanted to build it. I had a very small team and then through the time we made a couple of really strategic great acquisitions that also helped us build that effort. A couple of years later we shipped A4. And it was a great moment, A moment of pride and sense of achievements. The team worked extremely hard and the team here, The collective team, Not only the silicon team, The silicon and the hardware engineering and the software engineering team truly delivering the best iPhone. And when we built it from the start, We knew that we want to build something that can be scalable.
10:39
Johny Srouji
Meaning you build the first iPhone and then you're going to build something even better later. And we build what we call the Unified Memory Architecture that is scalable across products. We built an architecture that you start with the iPhone,but then we scaled it to the iPad,and then to the watch,and eventually to the Mac. But the fundamental architecture,we build it in such a way that can scale. Which was important for the efficiency of the silicon team. And how the product team,John's team,and the software team,for them,it will look,the architecture is similar across different products.
11:12
Katie Tarasov
How many engineers did you have when it actually launched then?Like,I'm just trying to get a sense of how quickly you had to ramp.
11:16
Johny Srouji
By the time we ramped,probably we had a couple hundred engineers.Now we have thousands of engineers.The other thing also we did was...
11:22
Katie Tarasov
Sorry,thousands of engineers just on silicon?
11:24
Johny Srouji
On silicon. The other thing we did knowingly as we started building the iPhone is focus. And focus was on the product level. Because the phone,you know,there is a thermal envelope,right,that you have to build a chip to deliver that. There is a battery that you want to maximize battery lifetime. And you want to deliver the best performance with best power efficiency and features. So the phone,you can look at it as a constraint when you started,you know,because it has a certain capacity,which I think was our best friend. I always believed that constraints enable great engineering. So by doing this,we had to create an architecture that was so power efficient that when we started scaling to the iPad and the Mac,it played to our strength. Then the other focus was we're going to focus on each IP. IP means technology,
12:15
Johny Srouji
Whether it's a CPU or graphics or video imaging or camera processing,etc. And we knew day one we're not going to go build every IP on our own from day one. So were very focused and practical about where we can differentiate. And we build it one by one. Today,we own all of our technologies and we have best in class for all of these. But we started one,you know,owning one at a time on licensing where we can license.
12:42
Katie Tarasov
Does Apple have aspirations to be in charge of every single part of the chip,including,let's say,memory?
12:48
Johny Srouji
Our aspiration,again,we're not a chip company. And John will chime in maybe later. Our aspiration is the product. We want to build the best products on the planet. As a technology team, Which also includes the chips in this case, We want to build the best technology that would enable that vision. So if there is a technology that we can buy off the shelf and that delivers to our objective for the product, We'll do it. Because I want to focus the team on what really, Really matters, On the most differentiating aspects of the chips. So if there is a chip that we can buy and there is no differentiation in building internally, I would buy. So we can focus our efforts on the things that really matter and we can build the best technology that we can. We maximize our benefits.
13:29
Johny Srouji
Memory,there is NAND flash memory and there is DRAM memory. I'm not sure which one you were referring to, But you may know that actually we have our own team. That builds our storage controllers for memory for NAND. And we have those embedded in our system on a chip,which is the SOC,that chips in the iPhone and iPad and. Max and other products. The reason we did that is we care not only about capturing the best photos. We care about the endurance of the device,of the memory. Those photos are memories you want to keep forever. We want to deliver the best,most reliable storage and fast storage whether it's either right. When it gets to DRAM,we also like if you look at the. Mac chips,for example,M3 Macs,we enable an astonishing. 128GB of DRAM capacity with. 400GB per second of bandwidth. Which is amazing.
14:24
Johny Srouji
Like for example,if you want to run a machine learning,LLM machine learning on device,it's a perfect machine. It has great compute,it has huge DRAM capacity and great bandwidth. So we want to control the aspects that we believe are enabling the product. But we don't want to control something for the sake of building it internally. That is not our DNA.
14:43
John Ternus
Yeah, And I think that's right. I mean, There are thousands of pieces of silicon in our products. And like Johnny said, The goal is make the best product focus on the. The critical things that are going to allow us to do something uniquely great. And I think, You know, Johnny was talking about how starting with the phone and that constrained thing, And then constrained form factor, And then building up. I think the Mac offered this really interesting inflection point for us because it was the one product where we had for years been building with other people's silicon, And then we all of a sudden got to build with our own. And so when you're designing products, You're usually dealing in trade-offs, Right? You can add more battery life by having a bigger battery. The product will be a little bit thicker,
15:21
John Ternus
A little bit heavier. You can have more performance if you want to consume more power. That obviously also goes against battery life, And it means you need fans. And if you push really hard, It gets really loud. And I think as we started working with Apple Silicon on the Mac, For those of us who've been building computers for a really long time, It was almost like the laws of physics had changed. Like all of a sudden we could build a MacBook Air that's incredibly thin and light, Has no fan, 18 hours of battery life, And outperformed the MacBook Pro that we had just been shipping. And, You know, We just launched a MacBook Pro with M3 Max that is 11 times faster than the fastest Intel MacBook Pro were making. And were shipping that just two years ago.
16:00
John Ternus
And these new MacBook Pros have 22 hours of battery life, And the fan doesn't even turn on for most of what you're doing. So. I think, To your question before, Do customers notice that we're doing this? You know, Even the ones who don't necessarily know that we're using our own silicon, They fundamentally see a completely different. Computer experience and you can't get computers like this from anywhere else. And so that's because we're focusing on these areas where we can do something uniquely great.
16:27
Johny Srouji
The reason we actually decided to transition the Mac to our silicon is we wanted to build a better product. It enabled us to build the product we had in mind without compromises. When it gets to physics,I always believe that,you know,we shouldn't be gated by imagination or vision of what we want to build. If you want to be gated by something,you want to be gated by physics and time. That's how we drive the team combined. And we're pushing the envelope,which I think translates to a much better machine that the customer will enjoy.
16:53
Katie Tarasov
Is there a product that,you know,I asked you about your,whether the A4 was a very proud moment for you and for you,is there a hardware launch tied with silicon that is the proudest moment? Because you've been here for 22 years. Yeah. What do you look back as the most proud moment for the hardware combined with silicon?
17:11
John Ternus
It's very hard for me to pick one,right? It's like asking you which your,who's your favorite child? Like,and,I mean,I can name a few. AirPods was an amazing thing,right? That was enabled by technology that we built in-house. And I think,I mean, I don't think I know it fundamentally changed how people use earphones. And iPad,you know,as Johnny said,we started iPad using a phone chip and then we scaled up to something much more performant,taking advantage of that larger form factor. And I think that really changed,like,how people,you know,people realizing what they could do on their iPad. And then I mean the Apple silicon transition on the Mac was just it was so much fun, Right? It was it was so much fun to be able to work together and figure out.
17:56
John Ternus
Because you know we have the benefit of working You know for a long time with very stable customers who love the Mac and so we know what they do, Right? And so were able to work together to develop an architecture that was perfectly suited for the kinds of things our Mac users do. And, You know, Johnny talked about the unified memory architecture and the incredible memory bandwidth. And we also have these dedicated hardware blocks like the neural engine and video encoders. And so they let people do things they just couldn't do before. Right. A video editor can do a huge work on a huge project that previously would have taken a giant desktop. They can do it on their laptop, On set, On battery, Which is amazing. Right. And so being able to create that much of a leap forward,
18:40
John Ternus
That was a profound moment.
18:42
Johny Srouji
There's another difference,you know,when you look at the Mac transition and building the first iPhone. We started with an established platform. There were chips outside for Macs and PCs. The other thing, If you really wanted to build the best Mac, You probably pick different vendors for different technologies. Discrete CPUs from one vendor, Discrete graphics from one vendor,etc. Our approach was very different. We want to build an SOC,System-On-A-Chip, That actually has the best in class. Fully customized. CPU,Graphics,Machine Learning Engines,etc. Just for the Mac. So you make no compromises and you build it all in one chip just delivered and built for the Mac. And then we built multiple generations. If you look at how many generations since M1, Which we launched about three years ago.
19:28
Johny Srouji
It followed with the whole M1 family of chips,four of them,then M2 family of chips,four of them,and three M3 family of chips.Eleven chips in just three years.
19:36
Katie Tarasov
Is there a reason that some of the metrics that I've been looking at and the numbers that I'll probably report in my story compare the M3 to the M1 instead of the M2?
19:47
Johny Srouji
We wanted to compare since we started the transition. In our actually video,the charts have comparison for M1 and M2 and other systems. But we wanted to focus on M1,so we just focused one chip. But the advancements are great relative to M1 and M2. So it was a matter of focusing the message, But it's all there.
20:08
Katie Tarasov
And then you mentioned being constrained by,you know,the only thing you want to be constrained by are the laws of physics,which I think about a lot in my coverage of chips because it's wild. I mean,chips literally are like pushing the known boundaries of physics. And Moore's law is perhaps one of the constraints here. Some critics have said that Apple was advancing nodes every year for a long time and that has slowed. How do you respond?
20:31
Johny Srouji
When you look at Moore's law,ten years ago versus now. There is a slow down in terms of how many advancements year over year you're going to get. And the advancement you measure,like. I want to boil it down to three metrics. The die density,how many transistors can you pack in a given area. The performance and the power efficiency. So when you look generation over generation,two things are happening. Generations are taking longer because they are getting harder and harder. And the ability. To pack more and get power efficiency is also different than 10 years ago. But given again,I think this is actually one of the advantages that Apple has because we're not a chip company. So I don't need to worry about where do I sell my chips? How do I target a larger customer base versus I work with my friend here?
21:21
Johny Srouji
And customizing those chips for the products. So we get to wisely use those very precious transistors. Efficiently,how the product will benefit,how the software will use it. And that's how we can make advancement,I believe,faster than anyone else.
21:36
Katie Tarasov
Is there a brain drain that's causing some slowdown in advancement?
21:42
Johny Srouji
Good question. So we have,as I mentioned,we have thousands of engineers in the US,Europe,Israel,and these are just the main sites,but we have across the world. I believe we have the world class team building those chips. We have a deep,deep talent. Skills and leadership talent leading those teams. We are empowered. To keep hiring the best,whether from college or experienced. Many times,some people may choose to switch companies for various reasons. But I don't feel that we have an issue because we're still hiring the best. We're retaining the best people we have. And I'm extremely proud of the team we have. If you look at the advancements we made in this space, If anything,it's a reflection of the organization we have built. And it's strong.
22:32
Katie Tarasov
When Gerard Williams left for Nuvia,was that an example of what you're talking about? Of,you know,sometimes people,good people leave. And what led Apple to drop the lawsuit against him?
22:41
Johny Srouji
Let me start with the lawsuit. We have. Great engineers building amazing technologies and chips,very hard to build. So we truly care about IP protection. Beyond that point,I can't really discuss legal matters,but we truly care about IP protection. When certain people leave for certain reasons,that's their choice,that's fine. And again,as I mentioned,we have thousands of engineers and we have a deep bench of talent.
23:08
Katie Tarasov
Is there also room for plenty of competition in this space? I mean, I know one line of thinking, For instance, That Jensen at NVIDIA and I talked a lot about is the world just needs more compute. Period. And so the more the better. Is Apple sort of in line with that thinking or is it a more competitive environment that you want to stay? Stay on top of.
23:30
Johny Srouji
So here's again the main difference between Apple and others. We're not a cheap company,so we're not out there. Trying to compete with chip vendors. What we want to make sure is we are building the best chips to enable our vision for the products,the best products on the planet. There will be no other technology or chip that you can buy. That will fit our products. That's our goal. And it's not only speeds and feeds. And we actually win in speed and feed as well,but it's not that. We want to build the chips that you cannot buy elsewhere. And it's a combination of many,many IPs and technologies that are the best. So. I don't measure by how we can compete in the market because that's not our space.
24:14
Johny Srouji
We are a product company which gives us the freedom to really innovate for our needs and make those long term bets. We have great compute in our and in our chips. When you look at compute,it's a CPU. There is a machine learning acceleration in the CPU. Our GPUs, Which are really best in class for our products, Are highly optimized for machine learning. We have neural engines that we built, If you remember, In A11 Bionic in 2017, Which means we started even earlier that we did that. And we give DRM capacity and bandwidth for that compute. So compute is important for many applications, But we don't measure just by workloads or speeds and feasts because we're not out there sending our chips. We want to make the best products.
24:56
John Ternus
Yeah, We're focusing those designs on the specific products that we have in mind, Right? That M3 Max is perfectly suited for MacBook Pro, Right? I mean, It was designed that way from the get-go. I mean, To the extent that we have teams sitting together talking about the physical dimensions of the chip, Right, And the aspect ratio to make sure that it can nest perfectly within a really densely packed system. So that's our focus. But I mean, Competition is always good. It's always good that people are pushing the envelope and doing new things. But we're very, Very focused, As Johnny said. It's all about the product and how do we make the product better.
25:26
Katie Tarasov
Designing your own chips is incredibly expensive. Is this capex sustainable?
25:32
Johny Srouji
Let's answer the design and maybe John can answer the other aspects. So, We have thousands of engineers. But if you look at the portfolio of chips we do, Very lean actually. So very efficient again because we focus on the quality of engineers we hire. We focus how we run our teams and because we're not really selling chips outside, We focus on the product and that gives us freedom to optimize. And the scalable architecture lets us reuse. Pieces between different products.So if I step back and look at the expense. For running, You know, Building our chips relative to the products we have. I think it's a pretty,pretty good deal,including the amazing customer experience that you can never achieve. That's how I look at it.
26:18
John Ternus
Yeah, I think you're exactly right. I mean, If we build the right products with the right components within them, Right, Silicon, Other technologies, People love them and we do well and it supports itself. And so that's our focus. And yeah, I think as Johnny said, It's a great investment.
26:34
Katie Tarasov
Modems. Talk to me about what's going on in modems and the recent announcement to extend the deal with Qualcomm and why that decision was made.
26:42
Johny Srouji
So let me start with connectivity, Which includes cellular and modems, But it also includes Wi-Fi, Bluetooth, Et cetera. So connectivity in general is extremely important for our devices. Most of our devices are mobile, And that includes cellular. We have a strong team internally working on enabling technologies that would elevate our products. But I can't really give you much and comment on future technologies and products for obvious reasons. But we care about cellular, And we have teams enabling that. And we work with Qualcomm, And we worked with others in the past to enable our product roadmap.
27:16
Katie Tarasov
But modems are very difficult to do,correct?
27:19
Johny Srouji
Extremely difficult to do,it's the whole system.
27:22
Katie Tarasov
AI. Are you developing a chip like Microsoft is that we're going to see coming? What does Apple have in the works when it comes to AI chips?
27:31
Johny Srouji
So I'll give you the macro level. When you look at AI, Machine learning or generative AI running on device, For example, There are many things you need to have, But I'll boil it to three points. One is compute, DRAM capacity, DRAM bandwidth. Let's step back to compute. In A11 Bionic, Which we shipped in 2017, And given typically we started chip many years ahead, It tells you how long we started. We actually started, We built our machine learning engine, We call it A&E, App and Neural Engine, Which for some workloads, It is the most power efficient. Compute element to use and we use it. Our software uses it for the phone and we use it in other products including the Mac. So we have the neural engine, That's one compute element. We have embedded machine learning accelerators in our CPU.
28:18
Johny Srouji
We have our highly optimized GPU for machine learning. So we have those elements. And given our unified memory architecture where you don't have different chips. Interconnecting through PCIe or other lower speed interfaces,we have a DRAM that is. 128GB just for example for M3 Max and a very fast. 400GB per second. If you want more,M2 Ultra has. 800GB per second bandwidth. So I believe we have great technologies and chips that will enable these devices and we started early on actually. And I'm very excited about how this software is going to use this in the future. And they've been using it for many customer experiences.
28:59
John Ternus
Yeah, No, That's exactly right. I mean, I think this is an area that has been an area of investment for us for years, As Johnny said. And we continue to push and we continue to see more and more applications, More features in our operating system that take advantage of this. And that's a really exciting time.
29:13
Katie Tarasov
AI on device, Right? I mean, Is this the future that Apple sees for generative. AI? And is your silicon going to enable that in any particular way that you want to talk about?
29:22
Johny Srouji
Again, We can't talk about the future. We can't talk about the future. Okay. But if you look at the platform. We have,I think,a great platform. To utilize generative AI,but we can't really discuss future on that.
29:36
Katie Tarasov
Critics have been surprised that Apple appears to be falling behind when it comes to AI. How do you respond?
29:45
Johny Srouji
I don't believe we are.
29:46
John Ternus
Not to worry.
29:47
Johny Srouji
Not to worry. And actually we have an advantage. Again,go back to we own the silicon,the hardware,the software. The machine learning, All in one team. We're going to optimize for that.
29:58
Katie Tarasov
Are you relying on some NVIDIA at all? Are you guys buying NVIDIA?
30:00
Johny Srouji
We can discuss.
30:02
Katie Tarasov
Yeah. The three nanometer leap. My question is like, How big of a deal is that? I mean, It's the first three nanometer chip being used, Right? I mean, I want to hear...
30:11
John Ternus
It was the first phone and then it was the first computer, Yeah.
30:13
Katie Tarasov
And I want to hear like, I guess from a hardware perspective, From like the customer, From the actual like, You know, User perspective, How big of a deal that is.
30:21
John Ternus
You know,on the Mac,we talked about the huge performance leap with M3 Macs and what that means in terms of compute performance and graphics performance. And one of the areas that we're really interested in and we've been seeing some wins in terms of getting titles and stuff is gaming,right? And there's a handful of really exciting games that are coming out on the Mac and taking advantage of all this capability including. The Hardware Excel Ray Tracing,which we brought with the. M3 series. And it's just,I think, A really exciting time for the Mac. It's early days. We have a lot of work to do,but I think there's so many Macs now. Pretty much all Macs are capable of running AAA titles,which is not what it was like five years ago. And I think that's a really exciting time.
31:03
John Ternus
I think the other piece on Apple Silicon in general is just,we talked about how we can design these solutions for our customers. But, It's also enabled us to build some products that we previously just wouldn't have been able to do. We always kind of thought about and wanted to do. And, You know, The MacBook Air is a really good example of that. And, You know, This past summer we launched a. 15-inch MacBook Air. And there's no way we would have been able to build, You know, A 15-inch product that really felt like a MacBook Air without Apple Silicon, Without that efficiency in building something so thin and light and capable. And then, You know, We just brought M3 to the iMac. The iMac is this beautiful computer. It's so thin that it almost seems impossible there's a computer behind it,
31:44
John Ternus
Much less one that can edit and play back 12 streams of 4K video. It's this incredibly capable device. And so, I mean, You can probably tell I'm really excited about the products that we're making and even more excited about the things that we're going to be delivering in the future. There's just so much opportunity here for us.
32:01
Johny Srouji
And the reason we used the nanometer is it gives us the ability to pack more transistors in a given dimension. Chips AI. And much better power efficiency. Apple is the first actually to ship 3 nanometer in high volume phone and Macs. We're leading even though we're not a chip company. We are leading the industry. For a reason.
32:22
Katie Tarasov
Is there any risk to being sort of the first to market with a 3 nanometer chip? There were some rumors about overheating in the iPhone in particular,the iPhone. 15,that could have been caused by the 3 nanometer chip.
32:34
Johny Srouji
I'm not aware of that.
32:35
John Ternus
Oh, There wasn't an issue with overeating. There were rumors around that,
32:38
Johny Srouji
But actually we had nothing to do with the chip.
32:42
Katie Tarasov
And it's since been resolved? Yeah.
32:45
John Ternus
We had a software update to fix the bug and it's been fine.
32:48
Johny Srouji
But is there a risk of using an advanced node? Of course there is. These nodes are very complex. Our designs are very complex. So of course there are risks,that's why you need the best teams across both sides to work together,whether it's the founder and the design team. That's why I always say silicon design is unforgiving. It's extremely painful. When you hit a late found bug in silicon in terms of how can you recover. And what does it cost? In terms of time or other aspects,my team knows I don't like surprises,even on my own birthday. So,which is why we're very careful when we design those chips and we validate that we,everything has to be extremely checked. It goes back to the team. You can't deliver any of this if you don't have the best team. But of course, It's risky. But that's...
33:40
Johny Srouji
Good enough doesn't achieve great results.
33:43
Katie Tarasov
I mean,great risk and great reward,right?
33:45
Johny Srouji
Calculated risks. Yeah. With great rewards. Thank you.
Source: Apple Executives Johny Srouji And John Ternus Talk About Chips, AI And Innovation