An Introduction to Full Stack Innovation

0:04
Welcome, this is circuit talk. My name is Nitin Shah from Mitre Engenuity. And today we're going to talk about full stack innovation, especially in the context of the chipset. Our guest today is Todd Holmdahl. So Todd, welcome. Please, may I ask you just to introduce yourself some of your background, please? Sure. Yeah.

0:26
Thanks. My name is Todd Holmdahl. I've been working at Microsoft, for almost 30 years now. I'm an electrical engineer at a software company Ss that's got its good points and it's bad points. But over that time, I've worked on everything from Xbox, I started the Xbox hardware group to HoloLens to quantum computing, and kind of just finished off with running the hardware for the Azure Systems Group A few years ago. And now I'm really engaged on looking at strategic silicon and ensuring that we have a good continuity of Supply here in the US, as well as a secure product no matter where it's built. So Todd, that's the

1:15
very wide ranging background, but very relevant to our conversation today, in the context of the chips Act, and the challenges that we face, but also the opportunities here in the US. So as you know, semiconductors are core to any of the products and services and entertainment, including Xbox, which you worked on, as well. So could we just start with what is your thought about the impact of semiconductors on the nation as a whole, it's been talked about a great deal, but your personal perspective on semiconductors and their relevance to our, our country and our industry?

1:57
Yeah, I mean, the, for me, the semiconductor is kind of both the brains and the heartbeat of both our economy as well as our national security. If you look at all the products that are out there, whether it's a car, of course, your computers, your phones, or TV sets, your now your appliances, they all have semiconductors in them. They all have high performance semiconductors that do a lot of processing and connect you into the internet, which allows you to have a better customer experience. So allows you to do more things and be more efficient. And then of course, for national security, you know, more and more you're, you're finding not only our semiconductors and the products that we use in national security, but a lot of them are tied to the cloud to process all the information that they get. And when you get something in the cloud, you just need an incredible amount of processing capability. And then of course, that's where semiconductors and silicon come into play. So I think it's vital for the country, it's an existential threat, if we lose access to him at any node, and that, you know, we need to do all that we possibly can to ensure we have a continuous supply. And to ensure that we have a secure supply. I mean, it is one of those vectors that hackers will go after if we're not careful with security.

3:33
No thanks for those observations. So regarding semiconductors, one of the topics that we want to cover today at the core of semiconductors and innovation is this notion of full stack innovation. So my understanding is that full stack innovation is perhaps not well understood by everybody, which is why we're talking today. And it seems to be an approach that results in the highest impact and the most innovative results. In other words, in order to achieve sustainable competitive advantage. We shouldn't only focus on one layer of the technology stack for computing or memory or other semiconductor products. But what we've described as the full stack, so just wondering, the full stack, what is it?

4:22
Yeah, that's a great, great question for me, you know, when you look at a product, let's take the iPhone, it's it's made up of a computing chip, typically CPU and memory that's made out of materials like silicon and other different substrate materials. And that's where a lot of the processing is. There's a you know, there's a motherboard that it goes on that's made out of a different material and it connects to all these pieces of semiconductor and other components together. There's an LCD display which is made out of different materials that this plays this information, you know, then there's a an interface, you know, that you have to touch in use in order to input information into the device that that, like when the first iPhone came out now is a unique concept that was developed, particularly for the iPhone. And then there's the case, you know, its iPhones are well known for the amount of energy that they put into designing their case, and the materials they use to make it slim and sturdy, and reliable. And then when you then you move up, what I would call the next level was stack, and that's the hardware or the software in something like an iPhone, they completely did a new operating system for a different form factor of device, they did a new operating system for a different input paradigm, you know, when you have touched and things like that, and then the last part of it is the applications, you know, the application construct is completely different. So, full stack to me means you touch all the different points on a on a product stack, where you start with materials, you go to the compute devices, like the semiconductors, to the materials that you make it to the, to the operating system to the applications and, and you can see in that particular case, that was a revolutionary product, right, because of the fact I mean, had no one expected that type of device, we were all prior to that had these big PCs, or we had clunky devices that kind of look like a tiny PC, but you couldn't input into them, and you couldn't look at anything because they were too small. And what Apple had done at that time, was they were really able to create a revolutionary product. And my thesis is that, that we're running to the end of a lot of these different, like, just focus on one layer of it. And that in order for us to have revolutionary products, think of a moonshot or the Manhattan Project that you really need to touch every level of the stack. And that's what I'm calling a full stack innovation or, and a full stack solution that you you have the ability to manipulate all these different variables in order to change the paradigm.

7:26
So I think one of the things you mentioned several times is revolutionary. And I think you're tying the ability to do something truly innovative and revolutionary, with full stack. So what's the counter example? In other words, if I limit it a new design or a new innovation to just one or two layers of the stack, so to speak, then I'm surmising what you're saying is you limit yourself or your possibilities by not looking at the full stack? Is that the right way to look

7:59
at now? That's right. I mean, when you may, I think the canonical example would be a CPU, right? They just they make faster and faster CPUs. And so you can, you can go ahead and put a faster CPU into an into an apple type device or previous data, a Blackberry type device, but unless you actually changed everything, like the form factor and the input paradigm, and the types of materials, you're using the operating system, you're going to end up only with a Blackberry, which isn't really good at, like taking a look at the internet. Nor is it grayed out as an input device for things that are non typing. And so that to me, was the revolutionary part of the iPhone, they were able they touch the full stack. And if you're only just if you're only just increasing your clock speed on your CPU, you're not going to have you're going to get your applications to run a little faster, which is nice. You might be able to do a few different things, but you're not necessarily going to be able to do something revolutionary. And that really changes the paradigm.

9:14
So there's a kind of a historical perspective here, I recall a lecture by professors Hennessy and Patterson, in aterian lecture a few years ago, and they reflected very much what you said, which is that in order to have big leaps in the area of computing and computer architectures, which is what those two professors worked on, they really had to look at every element of the stack, including, like you said, applications and software, and not just the hardware piece in order to achieve the extraordinary achievements that they had. So this is very resonant and maybe reminiscent of that discussion as well which is, if you want to make big big breakthroughs, big challenges, then in order to achieve that the full stack approach is the right approach. Yeah,

10:07
I mean, I think there's a good parallel with quantum computing. I mean, it's a, it's a different computing paradigm than a Turing machine, for example. And if you look at how most people are building a quantum computer, you know, they don't use transistors anymore, they use some sort of qubit, some flavor of qubit, they don't typically run it at room temperature, they're running below four Kelvin in order to, like not get noise involved with the they, the their operating system is completely different. Because the computing paradigm is completely different. All the applications have to be rewritten in a way that takes advantage of this the superposition of your qubits. And so that is a nice parallel to Dr. Hennessy. And said about, you know, these full stack solutions, and it's only by changing the complete stack, you know, different materials is not going to be silicon, that's going to be some semiconductor superconductor. But changing the qubit not it's not a transistor, by changing the system, you're not in a room temperature at four Kelvin, you're not running Windows OS on it, you're running some crazy new operating system, and your applications are completely different, you have to take advantage of these, this ability to do superposition. So I think that to me, that quantum computer is a revolutionary device, I mean, it's gonna do things that the Turing machine, you mean, it's gonna take it a billion years to crack some sort of encryption, where a quantum computer can do it in a few days. So that those are the types of things that you know, I want to aspire to. And I want to inspire other people to get into this field that they see the types of things that we're trying to do.

12:03
So before talking more about the future, I also did want to talk a little bit about your past regarding the Xbox, which is very personal. So I think all of us have been entertained and amused by the Xbox. And just wondering, you know, the Xbox had many extraordinary capabilities. So is that also another example of where you personally pioneered sort of full stack? Because you have sensors, the input device? And obviously all the computing and gaming capability as well? Yeah, no,

12:35
I mean, that that was a full stack solution as well. And you look at the paradigm kind of before game machines, when a PS One PS two were out then and Nintendo was out, but the paradigm before that was really about using a general purpose PC, right, with, with, you know, something that's done for doing your Excel spreadsheet, as well as playing video games. And so you had a mouse and a keyboard as your input devices. And what what we were able to do with the x box is basically focus just exclusively on gaming solutions. So we didn't have any extra transistors to do an Excel computation, you know, we use the game controller instead of a mouse or a keyboard, the operating system, again, is not windows, it's a completely rewritten operating system that allows you to process games faster. You know, it's, it's a heavy, it's a heavy machine on on GPU performance versus, you know, a standard one, we spent a lot of time I remember in early 2000s, debating whether or not to put Ethernet on it. And we did and that that was really the creation of Xbox Live, that was the first time you had kind of the start of this ability to play gains across the internet and online and the these you would not have that if you're only changing one aspect of the stack. And so I think the x box is a great example of that. And you know, it's a it's a product, we launched first in 2001. We started at 1999. It's it's been 21 years strong. It's a good story, but in a fun to be part of. And yeah, it's it's revolutionary in its own sense. And it really you can see, the real important thing is the customers liked it, and we were able to generate something that the customers like by touching all these aspects of it.

14:54
Got it. So actually, let me bring this back to the chips act. semiconductors and the future. So one of the things that we've discussed is this notion of breakthrough challenges, meaning that rather than focus future r&d In the United States on incremental work, we are at a point in time where we are looking at the chips Act legislation, which allows us to look at things in a different perspective. So this time between full stack innovation, breakthrough challenges, and looking a little bit to the future of semiconductors, everywhere in different markets, different places, and so on, just your observations on how we can really inspire the industry, and academia perhaps, to take on Research Development prototyping projects, which really are going to turn into massive commercial successes. So your thoughts just in general,

16:00
I, I'm a huge believer in these in these this idea of breakthrough challenge challenges, I think it really forces you to think outside of the, the, the normal paradigms that are out there. And whether it be you know, operating computers at a 10th of the energy they use today or, you know, looking at quantum computing, or going after climate change, or any of a number of interesting solutions out there. I I like the idea because it again, it forces you to think differently about the product to really stretch yourself. It gives you some guidelines to your you're not just doing business as usual, putting like research and into small, different silos. And I think the beauty of these breakthrough challenges is it forces a bunch of us to collaborate really well together. And when you get all that collaboration in that horsepower, you can start really building off of one another's ideas. And you can, you can touch all these different parts of the stack. And if there's one area where you can't solve something, maybe you can actually go out there and solve that. Another area of the stack. So I I'm a huge believer in this because it kind of gives us some guardrails to go forward with it gives us it's aspirational, it will inspire people to come and work with us because they see this helps us in the decision making prod process. We're just not spreading the peanut butter across the number of different research groups. You know, we're we have a goal in mind, it's easy to talk about somebody what we're trying to do. And I just and I do think that we are running out of runway, on today's technology with transistors we're not Moore's law is slowing down. And so we're gonna have to look at other parts of the stack and other materials in order to continue to be the leader in the world. In making products. I mean, there's no place close to the US and coming up with with new innovative products. And this is an opportunity to continue to build on that.

18:16
So your comments more on the on a human or personal level is this notion of collaboration, and many people do talk about our competitive strength be mean, we being collaboration, in other words, collaboration leads to very, very good forms of competition. And it's something that I think, stems from not only the technology aspects, but really using these breakthrough challenges to allow different companies or different universities or different parts of our industry to really work together towards these goals. So you mentioned also moonshots earlier. So just a few thoughts more about people and maybe the next generation of innovators and students listening in on this. Because we've talked a lot about technology and products. But how about people?

19:14
Yeah, no, I, I am totally biased. But I think that our engineers and our scientist are existential to the country that they drive a lot of the commerce, they drive a lot of what allows us to protect ourselves, these are very important things to consider going going forward. And that, you know, we want to be able to show them what they can do with this with math and science and engineering. We want them to be able to work with other like minded people in order to create these new these innovative product products. And I think, you know, we don't do a very good job along the whole educational stack. In letting people know what's possible and inspiring them, by giving them the tools to see what they can do with this information going forward, and I worry that we're really losing our competitive advantage as well. But other countries are much further invested in creating these next set of engineers and scientists and mathematicians, and that we need to continue to push that. And one of our strengths has been and always will be with the way we run our country is this this collaboration and this opportunity to work on new ideas. And we have, you know, 1000s of new ideas popping up all over the place and, and people vote with their feet to go and work on these things. And that, that is a great concept. And so I just think we need more of that. We need to enable that and we need more people that are capable of doing it.

20:58
So Todd, that's very inspiring. And thank you so much for being with us today here on circuit talk. And, again, Todd home data from Microsoft was deep thanks. Thank you very much for joining us today.

21:11
Thank you. This is a great effort and thank you for and MITRE for for driving.