TRANSCRIPTION 🔗
00:00:02
Speaker 1: Bloomberg Audio Studios.
00:00:04
Speaker 2: Podcasts. Radio.
00:00:06
Speaker 3: News.
00:00:17
Speaker 1: Welcome to Power On. We talk about Apple and the consumer tech industry. I'm Mark Gurman, and today we have a very interesting conversation and a very special guest, Gerard Williams III, joining us to talk about the AI data center boom, what he's building at his new startup, Nuvacore. what he built at Nuvia and Qualcomm, and of course, what he did at Apple as the chief architect on their CPU technology. He worked on chips starting from the A7, the first 64-bit Apple processor, all the way up to the Apple M1 chip. We get into his time there. We get into his thoughts on John Ternes and Johnny Sruji and what it was like when he first came up with the idea almost two decades ago to build a server-side CPU at Apple. And stay tuned until the end. We have another round of Apple Trivia. Just to give some background to those listening, Gerard has had an amazing career, and there's probably no one better to talk to right now about the AI data center boom, the silicon industry, and all of that coming together. Gerard, as you may be interested if you're listening to this podcast, he worked at Apple for a long time, Most recently, he was a senior director in charge of CPU work, platform work. And after that, he went to start a company called Nuvia, which made cloud silicon. It was acquired by Qualcomm. And most recently, he left Qualcomm, where he was a senior VP, to start a new company called Nuvicor. So Gerard, thanks for joining the Power On podcast. Thank you so much for being here. And good to have you, man.
00:01:57
Speaker 4: Oh, no, it's wonderful to finally get a chance to talk to you one-on-one.
00:02:01
Speaker 1: Let's get to your journey at Apple. So Gerard Williams III, you show up at Apple in what year? 2010. 2010. And so that silicon in the cloud was one of your first ideas. So who hired you? Who brought you on? What was the mission? Was this to develop the A7? Was that the first thing, the first 64-bit Apple chip? Tell me about getting to Apple, getting started, and what you did there from the beginning.
00:02:27
Speaker 4: Yeah, I joined Apple in, I think it was February of 2010. I joined a team called Platform Architecture. In the early days of Apple, that was a very small engineering team that was focused on literally looking at all silicon and hardware being designed in the company. And I was one of a handful of those engineers that was a part of that team. And we explored every design component. that would come across the desk of the senior execs and decide with them whether or not that that was a possible thing that could be developed and created in an Apple product. I had the opportunity to become Apple's chief architect of CPU development. The first product that saw the light of day that I worked on was the Apple A7. But yeah, the engineering team, we focus pretty heavily in silicon development. understanding
Load full transcript
that silicon, developing a roadmap for that silicon, and trying to make sure that the products year on year showed benefit.
00:03:33
Speaker 1: Apple really pushes this narrative about performance per watt on all its silicon. And when the M1 came out, which you worked on, I want to get into that, they were really pushing performance for watt versus Intel. It's not only about the speeds and feeds, but it's about the power efficiency. It's about battery life. Was that something that you were pushing from the beginning? Was that sort of one of your early innovations?
00:03:54
Speaker 2: I would say that even predates Apple.
00:03:58
Speaker 1: Tell me about that because you were at TI and ARM as well. I want to hear about the pioneer of this performance per watt idea because it's the biggest thing in on-device silicon right now.
00:04:08
Speaker 4: Even predating Apple, if you go all the way back to during the ARM days when I was there, we used to look at devices and the products that we were shipping. And in those days, back in the, I call it late 90s, early 2000s, You were talking about devices that were sipping 50 milliwatts, 100 milliwatts. So to you as an engineer, expending one milliwatt was huge.
00:04:35
Speaker 2: It was huge.
00:04:36
Speaker 4: It was a large percentage of the core itself. That grew over time, but that philosophy never changed. And in the world of computing, everyday life, us as users, I carry around laptops.
00:04:52
Speaker 2: We all do.
00:04:53
Speaker 4: We carry around cell phones, watches. They all have batteries in them. So burning microwatts, milliwatts is fundamental to those products because it basically means at the end of the day, can I make it through the working day with my product doing everything that I want to do and it still survives?
00:05:09
Speaker 2: And if that's the case, you've succeeded. The beauty of.
00:05:13
Speaker 4: That is once you understand those fundamental properties about how and where power goes and why, you can bring that to the data center.
00:05:23
Speaker 2: It's a property that people don't think about.
00:05:25
Speaker 4: You hear people throw out numbers like gigawatts and multi-gigawatts. I will tell you that milliwatts are still a part of that number. And if you can save milliwatts in every CPU that you construct, by the time you multiply that out by tens, hundreds, thousands, if not millions of cores, it's a material change that you can make. And so one of the fundamental things that we, I've always done, I did it at Nuvia.
00:05:54
Speaker 2: I've done it at every company I've been at.
00:05:55
Speaker 4: And I'm doing it again at Nuvacore on this clean sheet design approach.
00:06:00
Speaker 2: Power is fundamental, okay?
00:06:03
Speaker 4: Even though we don't think of it as a giant battery, it's a property that you want to lower as much as you can. Because at the end of the day, if the engines, as we're seeing, are driving up the core count demand and making CPUs what
I call front and center in the data center, if I can lower the power and I can suddenly squeeze two, three X, the number of cores into that data center building, I've won.
00:06:33
Speaker 1: So obviously you worked very closely with your product design, with your electrical engineering, with your hardware engineering teams when you were at Apple. Do you feel like You were kind of handing them a gift. It was sort of like a magic bullet. You were handing them with those chips, allowing them to design devices they wanted to design and implement features they wanted to implement they couldn't necessarily do with Intel. Was it seen that way internally that you guys were the wizards really making this happen but not getting the credit?
00:07:03
Speaker 2: I don't think about it that way.
00:07:05
Speaker 4: I mean, reality is, and what I will tell you fundamentally, a CPU is not designed by one person. It's not at some of the ideas come from a person, but it comes from the team. It's fundamentally a team and it's how you build that team, organize that team, um, and, and give them the freedom to explore. That's actually what at the end of the day makes products great. Um, and, and, and makes engineers want to come to work every day, wake up and, and, and do amazing things. Um, And that's one of the fundamental properties that I've tried to instill in every team that I've created. And we're doing it here again at NuvaCore. So even though I've experienced 35 years of this in my career, sometimes I sit back and I watch the engineers do their work and basically pass ideas to them and say, hey, here's an opportunity. But give them the freedom to explore. And I will tell you, if you empower a group, they can do some pretty amazing things.
00:08:05
Speaker 1: Obviously being in that Silicon group, I have to ask you about a couple of people. First, John Ternus, obviously the new CEO of Apple. What do you think coming from an Apple that was, you know, you started, it was run by a visionary. Then it was run by an operations guy for, you know, 95% of your career there. And now it's being run by a hardware engineer who rose up the ranks very quickly. I mean, by the time you left Apple in 2019, like really no one's ever heard of the guy. Even when I was talking about him, starting to tell people two, three years ago, starting to break the news that, hey, John Ternus is going to be the next CEO. This is happening. Everyone's like, who the hell is this guy? No one's ever heard of him, right? Okay. You having been there, did you work closely with him? Was this a shock to you? Does this make sense? What do you think of a more engineering-led Apple? Tell me if you have one, an interesting Ternus story, because obviously you were the guy making the chips. He was the guy, you know, pulling together the chips with the other components and the product design and all that.
00:09:03
Speaker 4: Yeah, I won't go into personal details about my time there, but putting names to who you mentioned, obviously, there was Steve Jobs, an obvious visionary. The world sees him that way. You've got Tim Cook, the financial individual, and then effectively now John Ternus, who you were just asking about. I believe they all have their different strengths, and they And in Ternus's case, time's going to answer that question really at the end of the day about the decisions he makes, how he utilizes the engineering team. But as to his strengths and weaknesses, I'm not going to elaborate about that. I did work with him while I was there. Very good person. I will say that he is a good person. I wish him the best in what he's doing. And I honestly do hope that he succeeds at everything that he does and what he what he can pull off for the company.
00:09:56
Speaker 1: Johnny Sruji, you know, he's known for running that team with a, with an, with an iron fist, anything, any thoughts on him and how he's doing and how he runs the team and his management style? Do you have any learnings or takeaways from working under him for so long?
00:10:10
Speaker 2: Yeah.
00:10:11
Speaker 4: I mean, I've, I've known Johnny for quite a long time. Um, probably about 20, 20 years of my career.
00:10:17
Speaker 2: Uh, I I've known him.
00:10:19
Speaker 4: Um, I have the utmost respect for him and what he's built at Apple. Amazing organization year on year. They're delivering and developing. And I wish him the best and hope they continue to do so, as I do with many of my colleagues and friends that are still there. So I wish him the best.
00:10:45
Speaker 1: And you left Apple in 2019 right on the cusp of the M1. Why did you decide to leave at the time? And what was that process like? And then how did you jump into Nuvia?
00:10:54
Speaker 4: The answer is yes. It was, I believe, February of 2019 that I stepped away.
00:11:01
Speaker 1: And you were a senior director at the time.
00:11:03
Speaker 2: I was.
00:11:04
Speaker 1: Just to explain to the audience, senior director, there's about 300 of them across the whole company. It's a very, very high-level position. There's the CEO, there's the senior VPs, there's the VPs, and then there's the senior directors. But then there's a large tier of senior directors, which are essentially in the VP tier. And I guess that's sort of the Venn diagram you were part of.
00:11:26
Speaker 4: Yeah, you've aptly described it, yes.
00:11:29
Speaker 3: Yes.
00:11:30
Speaker 4: Yeah, like departing any company where you've kind of enjoyed the experience of working on the products, developing the products. I mean, you end up, you want to give everything that you can to the company as you're progressing.
00:11:42
Speaker 1: Yeah.
00:11:43
Speaker 4: At the end of the day, when I did walk away, yeah, it was heartbreaking. I mean, it is for many people that walk away that invest a lot of time in helping products.
00:11:50
Speaker 1: And they were pretty heartbroken, too, I hear.
00:11:54
Speaker 2: They were.
00:11:54
Speaker 3: They
were.
00:11:57
Speaker 4: But ultimately, I walked away because of an opportunity that turned into Nuvia. And that opportunity opened doors for me to experience things in my career path that I wasn't able to experience. I'll say being an engineer, things like at the time I was the CEO of the company. And you are out raising money, funding, which is the money side of the equation. making sure EDA agreements are put in place, a variety of things, hiring people, building a team, making sure that we have direction set. So that's something that you can't experience as an engineer inside of an organization. And it was an opportunity for me to potentially, I'll call it spread my wings and kind of take flight, so to speak, in a startup world.
00:12:50
Speaker 2: Was it easy to step away from that Apple journey? No, it was not.
00:12:57
Speaker 1: I should note for the record that Apple did sue you and Nuvia after you started the company. They accused you of hiring engineers inappropriately and trade secrets and what have you. And then as we see in a lot of these lawsuits, courts determined to a large degree of being frivolous. And you and Nuvia obviously came out on top here and had a very successful launch at Qualcomm. And now you're on to your next thing.
00:13:25
Speaker 4: Yeah, I would– on the lawsuit side, what I will tell you, I can't go into that just.
00:13:34
Speaker 2: For the viewers. I can't dive into that for obvious reasons.
00:13:37
Speaker 1: That's why I didn't ask you to. I just had to– it would be weird. Someone watching this is like, how did he have an hour conversation with Gerard Williams III and not mention the lawsuit? So I had to throw it out there. Well, let's talk about NuvaCore. I want to dive in to exactly what are you building today and how does your learnings from Apple and Nuvia and TI and ARM, how did that all come together into what you're doing? And why is this something that you couldn't build at Apple? Why is this something you couldn't build at Qualcomm? Why did this require a new company?
00:14:15
Speaker 2: Yeah, so NuvaCore journey started actually... of this year.
00:14:22
Speaker 4: It was an opportunity to take advantage of the explosion of AI that we're seeing today and bring CPU to bear to the market. That's one of the interesting things that we are seeing in today's times, that the large language models themselves are utilizing GPUs, accelerators, AI engines themselves, and they've exposed what I've always been I believe, to be there since the beginning of my career. That CPU is an element that's front and center in the data center, has been for some time, and those high-performance CPUs are necessary in order to enable these large language models to do their work. They orchestrate memory, they orchestrate the security aspects, and do some of the fundamental work. And Pretty much single-threaded performance is a predominant element of that CPU that makes it great, and efficiency. And throughout my career, that's been something that I've tried to do every step of the way, is to bring high-performance compute to every platform that I was able to partake in. And NuvaCore is beginning that journey, and it's performing that journey as we speak today.
00:15:45
Speaker 1: But everyone is all in on GPUs. Obviously, NVIDIA has talked up GPUs. Everyone's talking about GPUs. And so what was the moment where you decided we're going to do CPUs instead and leave the GPUs to others?
00:15:59
Speaker 4: For me, actually, there wasn't a magical moment that I said I'm going to leave GPU behind and focus on CPU. CPU has been something that I have focused on almost since the beginning of my career, very early day one. I've always believed that that compute and the CPU that you build has the ability to scale. And that's one of NuvaCore's fundamental testaments is that we want to build a great core. We call it our core-first technology. It's basically focusing on the fundamental capabilities of a CPU, what it offers, how it explores the fundamentals of an OS. And we want to bring really solid performance to the end user and to the data center, which is what LLMs today are mandating.
00:16:49
Speaker 1: Can you get a little bit nerdy and technical on this exact first product that you're building, this exact chip? Like how many cores are we talking about? What are the technologies? What instruction set are you using? I know there's been some talk that you're trying to lock in on ARM versus RISC-V, right? So what exactly, from a technical standpoint, in terms of what you can share, is this first product going to look like?
00:17:10
Speaker 4: Yeah, so we are trying to focus on a clean sheet design.
00:17:17
Speaker 2: That's number one.
00:17:18
Speaker 1: And what does that mean exactly?
00:17:20
Speaker 4: It basically means starting at first principles of what makes a CPU great. That means understanding things like good prefetchers, good branch predictors, a fundamental sound pipeline, a.
00:17:38
Speaker 4: A pipeline that understands the fundamentals of computer science, things like graph theory, pointers, link lists, elements like that that make up everyday programming that we all get to experience when we are using a computer. Clean sheet design fundamentally says, I'm going to explore all these attributes. I'm going to look at the processes that are available in the industry, companies like TSMC, Samsung, Intel, all the different processes that we have the ability to partake in and marry these two philosophies together. High performance process with sound and reasonable CPU architecture. And then the last ingredient, which is what Core First is trying to do, is to basically say, I want to understand the available ISAs in the ecosystem, and you name two of them, there's basically three kind of two more prominent ones and one upcoming one, x86, ARM, and one that's being worked on is RISC-V. But understand the properties of those ISAs and make sure the CPU's
fundamental principles that we build are held so that they can be targeted at any one of the three at the time that we want to do our final I'll call it final optimization for a core that targets that ISA.
00:19:03
Speaker 1: I want to circle back to Apple because they did a very interesting thing, and you were obviously still very much in the thick of things there. So the iPhone X gets announced in 2017, and everyone's very focused on the new design, Face ID, the lack of the home button. But there was some special sauce in there that doesn't get much attention. That's the neural engine. Um, my understanding is that the origin story for the neural engine comes from the self-driving car project where they were trying to build a car and they need a gigantic computer system to do the live processing for autonomy. And they ended up with the realization that this block is going to need to be in the consumer devices and the phone at some point in order to power something. Can you talk me through the origin story of the neural engine at Apple? how it finally landed, and what the thinking behind it was at the time? Obviously, this has become a magical piece of technology that is enabling so many on-device workflows right now, but nobody was thinking about that a decade ago.
00:20:01
Speaker 4: I can't speak a lot about the internal workings of the projects as I no longer work there, and I'm It's just not an area I can go into. But what I will say at a high level, every company right now on the planet is actually focused on trying to build accelerators for a variety of different things, knowing that the AI boom is what it is.
00:20:29
Speaker 1: For you, though... What are your thoughts on, you know, on-device compute versus on-cloud compute? I mean, obviously, the big difference is how much performance you're able to get if you're putting it in the data center versus on the phone itself. But I would like your philosophy on that.
00:20:45
Speaker 4: I think there is a natural desire for every end user to be able to utilize the silicon that they have and the products that they have. I think there's an opportunity to bring large language models, scale them down. And I know many of the companies in the industry have been focused on scaling that to handheld devices or laptop devices or even desktop devices. And I think that is a huge opportunity for the engineering community, including the up and coming engineers, those students that are in elementary, high school and university level. to be able to understand and experiment with those technologies. And it allows you to scale to eventually one day exploiting, I'll call it the very large models that may live in the cloud.
00:21:39
Speaker 1: How does what you're building and this AI data center boom, how does it accelerate that? And as more people use this stuff, is this going to slow down over time? Because right now I feel like you and I, others may be listening to this, we're on the cutting edge, but at some point everyone is going to use this stuff. Is there going to be enough bandwidth?
00:21:57
Speaker 2: What I will tell you is I.
00:22:00
Speaker 4: Came from old school engineering where you sat down with pen and paper. Maybe it's modern times. It's a whiteboard or a marker board. And I was hesitant to step in a few I'd say a few years ago when when these models were ramping.
00:22:17
Speaker 2: Just because I didn't understand exactly. All right. What is it? Is it going to benefit? Is it not going to benefit?
00:22:23
Speaker 4: What I have found, and with the NuvaCore engineering team, what we have found is that these models, it does not replace your design team, but it's an extension of your design team. And what I mean by that is you still need experienced engineers brought into the company every day.
00:22:41
Speaker 2: We look for talented people.
00:22:46
Speaker 4: They are necessary in order to vet a lot of what these models expose or introduce as capabilities because you want to make sure that the product that you're building is robust. And I kind of equate it to a small child. It's in its early days. It has the capability of growing and maturing, but you need an adult to watch. We've been in the business now for about seven months. But we're seeing huge benefits for our engineering personnel.
00:23:18
Speaker 1: I look at it this way. It braces the bar for everyone. You know, Apple, you know, likes to say that it has no problem putting a competing product out because two products that are within the same company competing with each other, it just makes the products better. And so I think that AI, it's just forcing people to read up, to learn more, to get better at their jobs and push people. to sort of beat out this technology. And at the same time, you can't always trust these models. It can never be you. It can never do what I do. It can never do what you do. But it can help give us time back to focus on our strengths, right? But could you go a little bit deeper about how AI is actually helping with the nitty-gritty development process of programming, manufacturing, testing, making these chips more secure, and comparing it to even what you were doing a few years ago at Nuvia and farther behind that at Apple. How has the chip development process changed? And has it accelerated your roadmap and your ability to bring silicon out? And has it helped defeat this idea where you've seen CPU gains slow down, right? I mean, if you look at the initial charts when Apple released the first iPad, and then you look at the jump to the iPad 2, the GPU acceleration they had on that I think it was 8x or 9x. Now the year-over-year gains are 10%, 15%. But do you think with AI we're going to go back to the old days here?
00:24:40
Speaker 2: Yeah, many, many, many good points you brought up.
00:24:45
Speaker 4: Going back to, for instance, Nuvia, because I've got two startup experiences that I can relate to. Nuvia was pre the dawn of, I mean, AI was
there, but it was a minority. uh, in the ecosystem, um, not really used during development time. It was engineers, um, going to the, like I said, going to the whiteboard, doing design, exploring, um, today that has changed pretty dramatically. So comparing and contrasting, if you were to go through the design process, uh, developing a CPU and it's made up of many parts, um, and you want a There are phases to CPU development that we all like to understand. There is what I call the performance phase, where you're studying, researching, trying to understand what you go to develop. Sometimes that could take two months. Sometimes it could take seven months to a year. And extended research may take multi-year. Once you're done with and have created a view of what that performance wants to look like, There's a design process, a verification process, and a physical design process that you go through, which eventually turns this creative concept of a CPU, if I use that as the example, into a final piece of silicon that I try to ship into a product. Without that event of AI, that development cycle in its best typically was around two and a half to three years.
00:26:17
Speaker 2: Okay.
00:26:18
Speaker 4: That's basically clean sheet design going to an end product, two and a half to three years.
00:26:24
Speaker 2: In NuvaCore, what we have.
00:26:25
Speaker 4: Experienced actually in these early phases, and we're still developing the CPU as we speak, but I can tell you in these early phases of performance and performance modeling and characterization of a design, what would have taken seven months to 12 months to get to a point where you are ready to start the design process. We've actually collapsed that down to a two or three month window, which is very, I'd say, unexpected, but a positive. It's a positive because now that gives me time and the engineering team time to focus on potentially other ideas. And that gets to one of your fundamental things that you talked about, which is it looks like there's a slowdown in the industry. What I will tell you in my experience– That slowdown is really about your ability to explore and know what to explore and how to apply it.
00:27:24
Speaker 2: Where else can it be applied?
00:27:25
Speaker 4: It can easily be applied in the verification phase of designing, whether it be CPU or silicon. You can apply it directly there because it is what I call a very brute force technique where you can have it explore the state space of a CPU, understand what to test, how to test it, and then release it onto the design to basically find out, is it working?
00:27:52
Speaker 2: Yes or no?
00:27:53
Speaker 4: With the end goal being that you compress that entire two and a half to three year schedule into something that may look like a two year type of design point.
00:28:02
Speaker 2: That would be ideal.
00:28:03
Speaker 4: Year and a half would be beautiful if design teams can do that. And there have been a few examples out in the field recently where AI has been applied and accelerated the timeline. In high performance, very high performance CPU design, I think the answer is just time to actually show that the full clean sheet to endpoint can actually be compressed to something that is reasonable for the market.
00:28:30
Speaker 2: Does it compress to a year?
00:28:32
Speaker 4: Today, I can't answer that question, but I think given Gen 1, Gen 2 of our timeline, there's a possibility for that to become true.
00:28:41
Speaker 1: I still want to go a little bit deeper on why focusing specifically on the CPU versus the complete stack, because obviously one of the big, you know, inventions at Apple, one of the big innovations was sort of the system on a chip design. We were bringing together a full SOC with the CPU, with the GPU, with the neural engine, the memory and everything together. So can you go a little bit deeper on there and how this is one block and how that will fit in with other companies here? and how your strategy is really comparing to what you're seeing from Apple with its private cloud compute that they've been pushing for the security optimizations for AI, obviously NVIDIA is working on their own cloud CPU, and what you did at Qualcomm with Nuvia.
00:29:22
Speaker 4: Yeah, so at Nuvacore, our long-term ambition is to basically one day get to end product. Its performance has to be good. Its power efficiency has to be good. We're working with a variety of end customers to try to understand how they want to integrate that. And ultimately, that could turn into chiplet form of silicon with many CPUs on a die that's co-integrated with accelerators. It could turn into a full-scale end product like you see in classical data center and hyperscaler. which is a blade-based compute device.
00:30:03
Speaker 1: So I hear technical details. I hear AI features. I hear how my computing is going to improve. But investors, they hear dollar signs. And supposedly, you're about to raise a whole bunch of money at a valuation between $ 2 and $ 3 billion. Are you able to give me a little bit on that, of what your investors are putting into the company and how this thing is going to grow and really get off the ground here?
00:30:25
Speaker 3: Yeah.
00:30:25
Speaker 4: In terms of the exact financial investment in the company, we have not made any of that public.
00:30:30
Speaker 2: And I, I won't go into that right now, but what I.
00:30:32
Speaker 4: Will tell you is the investors that we have brought on board have been involved actively with Silicon. Um, the strategic players that we're bringing on board have actively been involved in Silicon and they know that this is a long-term journey. Um, and that basically means, um, Uh, money in, as you said, in, in, in the bees, uh, that have to be, uh, eventually expended in order to, to, to develop a technology like we're developing.
00:31:01
Speaker 2: Um, but in the exact dollar amounts, I won't go into today.
00:31:06
Speaker 1:
And how far are you off from actually going into production?
00:31:10
Speaker 4: As I said, we're, we're seven, seven months into the journey. Um, full silicon is somewhere between a two to three year journey. in order to create that silicon. So it's roughly around that timeframe.
00:31:33
Speaker 1: Now, circling back to where we started, obviously you leave Qualcomm, was it the beginning of this year?
00:31:39
Speaker 2: Yes, it was the end of January, actually.
00:31:41
Speaker 1: End of January. You said you left because you needed to finish painting up your new house. How did that go?
00:31:47
Speaker 2: I like that story. It's made its way around the industry.
00:31:50
Speaker 1: Is it finished?
00:31:53
Speaker 4: Well, if you want the honest truth, I'm actually still painting to this day. I recently have painted my daughter's bedroom, and I'm still doing it. This is the other side of me. I enjoy the relaxing. To me, that's relaxing.
00:32:09
Speaker 1: I was going to say, at this point, I feel like you could hire someone to paint it for you.
00:32:13
Speaker 4: I probably could, but honestly, that's me being an engineer at the end of the day. I like the details, and I enjoy it. You were asking about my early career. I had the opportunity, and many don't have this opportunity. As a small child, I got lucky. My father was into construction and a variety of things like that. So I learned, honestly— If I wasn't doing CPU as my day job, I could go into construction. And a lot of people don't know that. So, yeah, it's a nice, relaxing pastime.
00:32:48
Speaker 1: And last question for you. On this podcast, we focus a lot on the hardware itself, on the products, the physical devices people can buy. what does the industry shift look like in three to five years with a big offload of CPU into the cloud? I mean, I'm imagining things like AR glasses finally being able to be enabled, but what is this future that you think you're going to enable for hardware customers? And by customers, I'm talking about, you know, me and you and listeners, not like, you know, your NVIDIAs and open AIs of the world.
00:33:19
Speaker 2: Yeah.
00:33:19
Speaker 4: I think at the end of the day, um, Bringing CPU technology to bear in a market like this two to three years down the road or five years down the road, I think what we're going to discover is this utilization of LLMs and CPU and compute technology that every end user will get to experience.
00:33:38
Speaker 2: And you talked about glasses or devices. It's basically.
00:33:43
Speaker 4: Everyday, ordinary people picking up these devices and exploring these devices and utilizing them, being able to ask things that normally you'd have to go to a library and research and try to figure out. You can ask the language model these questions. You can even do it today. And my hope is that that actually is what happens with all the end users, not this fear that you see in the industry where, oh, AI is going to do this or do
that. No, it's about, can it make my working life that much better, my daily life? And if the answer is yes, we've done something pretty amazing.
00:34:15
Speaker 2: Cool.
00:34:15
Speaker 1: So we'll leave it at that. Gerard Williams III, thank you so much for your time. What an interesting conversation. I think it's really important given everything going on in the industry right now. And I'm glad we were able to tie it back to Apple, to Qualcomm, to ARM, all the hot buttons right now. So thank you very.
00:34:32
Speaker 2: Much for coming. Thank you. Thank you for the opportunity.
00:34:37
Speaker 1: Okay, it's Apple Trivia Time. Our producer, Alexis, is going to try to stump me. What do you have, Alexis?
00:34:43
Speaker 3: Okay, well, it's just you. So you're really on the spot with this one. But you know I like to go thematic, so I'm going to ask you a question about chips.
00:34:52
Speaker 1: If you're ready.
00:34:53
Speaker 3: Yes. It's a very chip-related episode. Okay. Okay, so the question is, how many of the Apple A-series chips carried the pro name? Ooh. Let me know if you want the multiple choice.
00:35:07
Speaker 1: Okay, I'm trying to think. When did they start giving the pro name? Well, there was definitely the Well, they did Fusion once. There was the A10 Fusion. I think there was the A11 Fusion. There was the A, or maybe it was the A11 Bionic. They went from Fusion to Bionic, and then they started doing the Pro chips. Was it the A15 Pro, A16 Pro? I would say, so A16 Pro, A17 Pro, A18 Pro, A19 Pro, A20 Pro. Okay, give me the multiple choices.
00:35:41
Speaker 3: Okay, so I'm going to give you these numbers correlate with the number of chips with the proname.
00:35:49
Speaker 1: Okay.
00:35:49
Speaker 3: So is it A, 5, B, 4, C, 3, or D, 10? It's not 10. It's not 3. It's 4 or 5. I'm going to bet 5. Oh, so close. No, it's 4?
00:35:57
Speaker 1: It's 4.
00:35:57
Speaker 3: Okay.
00:35:57
Speaker 2: Okay.
00:36:07
Speaker 1: What was the first Pro chip? Okay, so the first Pro chip was the A17 Pro. Yeah, A17 Pro, 17, 18, 19, 20. Yeah, exactly.
00:36:14
Speaker 3: I was off by one there. You're off by one. Do you remember what year they switched from Bionic to Pro?
00:36:19
Speaker 1: So A17 Pro, we're in 26, 25, 24, I guess it would be 2023?
00:36:25
Speaker 3: Yes, correct.
00:36:27
Speaker 1: That was when they split the chip into two variations, having the low end for the regular chip and having the high end for the Pro line, so... And they eventually put the Pro Chip in the iPad mini, and now they've expanded it. Obviously, the iPhone Duo and the 18 Pro and Pro Max have the iPhone's A20 Pro Chip, and then the iPhone 18 and iPhone 18e in the spring next year will have the regular A20.
00:36:53
Speaker 3: Well, I'll give you a half a point back because you got the follow-up question, but I'm going to count this as a win for myself.
00:36:59
Speaker 1: Okay, you got it.
00:37:00
Speaker 3: Okay. So the next question, I've noticed that you really like talking about different colors of products. So I had to throw a color question in here.
00:37:12
Speaker 2: Okay.
00:37:13
Speaker 3: So in 2023, Tim Cook was asked what color of the iPhone Pro Max he preferred and chose for himself.
00:37:20
Speaker 1: The natural titanium.
00:37:22
Speaker 3: Yes. Yes.
00:37:24
Speaker 1: That was the same color I got. It was the nicest color that year. A family member of mine has the natural titanium phone right now, like a couple generations old. I really don't like it at this point. But it was a good color back then.
00:37:39
Speaker 3: What is your favorite color of the iPhone of all time?
00:37:43
Speaker 1: Oh, I really liked the midnight green iPhone 11 Pro from 2019. And I really liked the Pacific blue of the iPhone 12 Pro in 2020. And this might be a little controversial, but I am really digging the Night Sky iPhone Duo. I think I went with the right color. A lot of people really wanted the white, but I like my Night Sky.
00:38:11
Speaker 3: That's not the first time you've said that on this podcast, but I like that you point that out. Okay, well, we got Tim Cook's favorite color. We also got yours. So there's another round of the Apple trivia in the bag.
00:38:23
Speaker 1: Thanks, Alexis. Okay, we have several more exciting episodes in the coming weeks, including a mailbag episode. So please send in all your questions to poweron at bloomberg.net or post them in the new podcast channel in our Discord, the Power On Discord. And we will try to get to all your questions in that mailbag episode. So stay tuned for that.
00:38:49
Speaker 3: For more from Mark, subscribe to the Power On newsletter at Bloomberg.com forward slash Power On and follow at Mark Gurman on all social platforms. Power On is a production of Bloomberg Podcasts. Our show is produced by Alexis Hott and Keshav Pandya. Our supervising producer is Magnus Henriksen. Audio engineering by Blake Maples. Original music is by Taka Yasuzawa and Alex Zagira. Our video editor is Bobby Kraus. Our executive producer is Amy Keene. And Cheryl Brumley is our head of podcasts. Special thanks to Nick Turner and Tom Giles. Thank you for listening.