People Behind the Packets

So, welcome everyone to our second episode of People Behind the Packets. I'm joined here with my guest Stein Lundby. So thank you so much for joining us Stein. People Behind the Packets, this is only our second episode. And so I'm very excited. My name is Nitin Rao. I lead the teams that build Cloudflare's global infrastructure. And in doing that, you realize that the, you know, it really takes a village to help build a better Internet and, and, and through this show, I'm hoping to have really a series of conversations with, with leaders I admire around the world who are, you know, building the under, building and shaping the underlying infrastructure that the Internet relies on, you know, be it hardware, connectivity, you know, data center technology. And, and one of the companies that's really at the intersection of all of those is Qualcomm. And today we're joined by Stein Lundby, who's the head of technology strategy at Qualcomm. So, so thanks. Thanks for joining us, Stein. Thank you. So I figured, I think we have about 28 minutes. So that should be enough time to fully understand what 5G is, right? Hopefully, yes, or some aspect, get an idea of what it could be. So the, so you clearly have a really fun job. But because Qualcomm is such an exciting company, and because of your role, how does, how does, so what is, what is, what is technology strategy mean that a company as expansive as Qualcomm? And how did you get into, get into the job you're doing right now? So the first thing is technology strategies is about the intersection between technology and business, right? You can ask yourself, why are cell phones so popular, right? From, we're told that there are more people with cell phones in the world than people with toothbrushes. We see these villages where they don't have electricity, and people in the evening hike up the hill to get to another village to plug in their cell phones, charge it overnight, and then go back next morning, hike back up to go and pick it up, right? You see people around you who've got no means to have a car, little means for other things, yet they have a smartphone. So there's something very, very important that humans find in cell phones. And at the same time, there's a significant business opportunity in that. And so that's the understanding exactly how this technology translates into great business and human value is about technology strategy. The converse is also true, meaning you see a lot of very advanced technologies that come out to research centers, out of universities, out of government centers, and they don't really make much impact, right? They're able to manufacture them, develop them, they're able to launch them, but the impact is really small. And what we've seen consistently over the years is that over the decades, actually, is that our ability to understand how technology translates into business value is pretty poor, even for experts. I mean, you can go back as far as, well, not very far, but back to the Segway, right? The Segway, we had some really smart, smart people who said that this is going to change the world, right? Steve Jobs was saying new cities are going to be developed, developed around it. There were investors that were rushing to put money into it. And eventually, it's just a novelty, right? I mean, it's used a little bit here and there. I think even the Mall Cops now have stopped using them. Maybe novelty tours in Washington, DC, and things like that. And then you've got, for example, more recent ones like 3D TV. The launch of 3D TV was huge a few years ago. It was supposed to really change the world. And now nobody wants a 3D TV, right? Curve TVs, or we had drones, right? A few years ago, your Starbucks coffee is going to be delivered by a drone. And now, where are we? Well, drones are still maybe for some niche applications, professional applications. But otherwise, they're more for entertainment. And people buy them just as a toy, right? Which they often don't even, they use for a month or two, and then they just leave it in their garage. And so can you only tell after the fact? I'd imagine in your role, you must be sort of seeing technology that's three years, four years, five years, and much more out. How do you know in advance, or what has a higher or lower chance? Or, or can you just not know? So that's the thing. That's the area of study that's very interesting with technology strategy is to try and formalize methodology to actually understand. It's not binary, but you can bring in more understanding, right, through frameworks. So you can start looking at, for example, I'll give you, well, 3D TV, let's take 3D TV, right? You can look at it, you can analyze it from a technology perspective, and say, what's the likelihood? And what are the challenges to develop the technology and make it work? Right? And that's fairly straightforward. I assume you've got the right engineering skillset, right, to do that. We can, you can estimate how likely or how much time it's going to make me to get it to mature. How much time and what's the manufacturability challenges? What's the cost? And can you actually stay within the correct range, and so on? The second aspect is the business aspect. Can you actually make a viable business? Assuming that the technology is good, and people want it, can you make actually a viable business out of it? Meaning, can you actually charge for it? Can you make sure people stay on it, and so on? And that's also fairly well known and understood, in particular, in MBA schools, and it's pretty well studied, right? And then the third one is the human aspect. And the human aspect, and linking these three is less well understood. That's where you need to start understanding, how does the technology and how did the business actually translate into human impact? And how, in this particular case for 3D TV is, how does it enhance the experience? So you can do a little test, let's, for TVs, just to give an example, right? Sure. What are TVs used for, right? So they use, what do you actually, Nitin, what do you use your TV for? I'm curious. Well, I don't actually own a TV. So I used it, I used my laptop to think, but for, if I had one, I would use it just just for entertainment. Yeah. So, so that's right. So people use their, they, they use their TVs, you could think of generally two experiences. One is, one experience is that they use it to entertain themselves. And what does it mean to entertain? It means to take you out of your environment and into something else, right? You're watching a movie about Mars, right? And you think you're actually part of that guy who's going to Mars and left behind, right? The other purpose, you'll see also some background, a little bit like a fountain, people have got a little fountain in their home, where they've got like a fish pond, fish going around, just as background noise generators, right? So let's just take the entertainment, the story part, right? And being really taking on words. If you look at the evolution of entertainment and storytelling, it started with the radio, which was, you had only spoken words, but no images. Then there was a significant step forward when movies came out, it was a significant step forward when you actually could watch the movie at home and watch the show at home, right? When color actually appeared, it was a reasonable step forward. It was in terms of the experience. High definition TVs had to be mandated by the government for them to be successful. So they're already there, you say to yourself, well, it's not adding that much to the experience, right? That it would have just sold by itself. Now, when you add on top of that 3D TV, then you realize it's more of an incremental improvement in the human experience. So for example, have you ever seen a movie that you thought was great in 3D, but you did not like at all in 2D? Sure. Yeah, it's a very different, it's a very different experience. Right. So the 3D actually is just incremental. And if the movie is a good movie, it'll be a good movie and just 2D and color, right? With a good sound system. Adding the 3D on top of that is just incremental. So when you've got an incremental experience like that, that adds to the already the basic experience and doesn't redefine it, the customers are going to be very, very demanding on the quality, on the cost, and it better work perfectly. And if you have to wear a pair of glasses, if it gives you a headache, or if there's any problem whatsoever, they're going to say no, and they're going to reject it, right? And that's what happened with 3D TV. It could come back, but it has to be perfect. It has to solve all these other problems, because it's not enabling the experience. And if anything, I think now would be a particularly good test, because we're not going to theaters anytime soon. And miss being in one. So if you're not going to buy a 3D TV now, I don't know when. That's right. Exactly. So that's what technology strategy is about. It's understanding that engineering, and how three links together engineering, business, and then human. And that's, it's a way to evaluate the potential for a particular technology or particular business scenario, particular strategy, in a more predictable way. And how did you know, this is, I'd imagine you're sort of the, you know, you're, how is how is this different? You know, it feels like you're, you're sort of moved for into a, into a different type of role. And so how is this different from, you know, the, the, the, then, then, then sort of being an engineer, like, what's the, what is the, what is the transition like along the way? So my, I'm an engineer at heart. And I started up as a computer science engineer, as a telecommunication engineer, electrical engineering, and actually designing Qualcomm's early chips when it was called the chip department. And it was a really a side job of Qualcomm, a little thing at the time. And then eventually, what I did is I, I did a variety of engineering roles. But eventually, by chance, I had a friend who worked in the legal department, and they dealt with intellectual property. And what I realized is that we have people who are dealing with patents, in particular, something that's very technical, but it's being left to lawyers, and people who are not only lawyers, but paralegals and all the experts in that particular area, right. And that's the first place where I realized there's an intersection between two worlds here that typically don't talk to each other. And so I purposely moved into the legal department saying, Hey, I never want to become a lawyer, I'm moving here, even though they're all pushing me to go to law school, nothing against law school, it's just purposely, I wanted to set up the challenge and try and see if we could bring the engineering rigor to the legal world. And we did that, we did that successfully. Qualcomm had a lot of challenges at the time, it was when it was, it was, it was being challenged by Broadcom, it was in battle with Nokia, and it was just ugly. And we were able to slowly unwind, understand how the legal system work, understand how the patent system work, and start formalizing a an engineering approach to intellectual property strategy. And so we developed a whole team eventually and staffed up about 50 people or so and developed new tools, new methodology, and a completely new approach of how to do that. And, and that's still a still a core function of Qualcomm. Now, other companies have started mimicking that has become a very important thing. So the next step off that I figured, well, okay, where does actually value come from? Right? Well, these patents, this intellectual property is generated from ideas, right? And its value depends completely on its business value, essentially, right? A patent can be worth not even the paper it's worth written on, or your entire company, it could save your entire company against an attack, or it could be a important foundation for a new product to grow your company and move it forward. And so it's evaluating these very, very binary assets to a large scale, when you've got 100 patents or 1000 patents, that made me realize, okay, we've developed a methodology to do it with patents. But let's try and see, can we move upstream and start moving, doing it with the technology itself, and the strategy associated. And so that's what interested me. And I started working with academics, with professionals, with strategists, with people throughout the world, and studying that. And I created this function at Qualcomm, which we now we've now established, which is technology strategy. So similarly, technology, you can spend billions, I mean, Qualcomm, other companies, and also have spent billions of dollars on one idea, right? And no money comes out of it. And nobody wants it, right? Google Glass, how many billion dollars did they spend on Google Glass, right? Then the question is, was it not successful? Because it was off by a little bit? Or is it just fundamentally a bad idea? And how do you redirect it? Is it that it doesn't need to be? So you hear the engineers and say, Oh, it needs to be higher resolution, or it needs to be a greater field of view. They all go for the technical terms, right? Then the business people are saying, ah, it needs to be sold directly to the customer. Oh, we need to sell it through Disney Plus, or something like that associated with a streaming service. But they're all talking their own language, and they can't actually come together to characterize opportunities like that and understand what is the real potential? And how do you redirect them and reshape to go in the right direction? So it's, it's, it's hugely important from a financial perspective. But in my mind, also from an economic perspective, and a human perspective, because we have so many smart, talented people in the world in the US, who work on some ideas that don't go anywhere. And if only we could improve that efficiency, even by just a few percent, I think we can do much more. But even by just a few percent, it can, it has the potential to advance the world tremendously. Now, now there, I think, like leaders, like you really at the, at the, at the intersection of sort of an interest in, in different fields. So I'd imagine play a role being being the glue at a company like Cloudflare, which is obviously fewer employees than Qualcomm. One of the things I really appreciate is you look at the chat rooms, and there are engineers who are, who are some of the most active participants in public policy chat rooms, and, and sort of lawyers who are in, who are in, in the sort of in the rooms about, about marketing and marketing and product and so on. Because like, really, some of the best ideas are at the intersection and, and it, it, and it also like forces you to, to sort of describe your work in, in simpler terms, then, then, then, then, then, then, then you, then you otherwise might, how do you build those bridges between, like, you know, if you're, if you're working with folks in, in, in different groups, and as you said, they're speaking their own language, how do you, how do you, how do you build those bridges? How do you do it sort of organically? Any insights on that? Yes, frames of reference, frameworks. Human beings, they understand concepts, and they communicate things most efficiently through frames of reference. So what do I mean? For example, the time, right, the watch, right? That's a frame of reference that allows you to communicate and operate much, much more efficiently among individuals. You and I knew when we're going to talk together, and we didn't have to communicate much to be able to arrange this call, because we had a frame of reference, and everybody, and that frame of reference is perfectly synchronized, right? Even if we were speaking different languages, even if we were from different backgrounds, completely different professions, we could have still met at the same time. So frames of reference are very, very powerful for human beings. And that's why, for example, streets, people spend so much effort naming every street, labeling every, every house with a number, every apartment building. Anytime there's a frame of reference, a GPS is extremely successful. It's a frame of reference. So human beings can come to an understanding through frame of references. The alternative is just talk. When you have talk, the problem is that, in particular, if you don't have a shared context, meaning a shared brain, you have a lot of difficulty understanding the talk. And there's a lot of things for misunderstandings and going the wrong directions. So in any field that you've got, where you've got multiple people from multiple backgrounds, professional backgrounds or other backgrounds, it always helps a lot to bring them onto the same frame of reference. If they buy into it, then you might be able to work together to communicate to understand. What it helps also is that it moves the discussion away from the individuals and actually to the topic. And you've probably seen that in meeting, right? You get into a meeting, and each person is trying to speak and say something, right? And eventually, in the end, which route, which path is taken, there's a decision to be made. It's often based on the most charismatic or the person who monopolize the conversation the most or others, right? And that's a very inefficient system, right? And that's because in those type of meetings, the outcome is really determined more on the individuals than on the topic itself. Frames of reference allow you to lay out the thing on a, on the table for everybody to look at, and everybody to discuss, and also everybody to audit, right? And say, well, I disagree, or you've done it wrong. And so I'm guessing you do that, like before everyone gets in, gets in the room. So you just like sort of get it right at the start. That's right. You don't even need to get in the room. Yeah, yeah, that's right. Sorry, I'm still adapting to the new word. Yeah. So many people are like zoomed out by meetings. As human beings, our normal way to resolve problems is to huddle together, right? And it works fairly okay, when you're in person, because you all get together, and you kind of, there's a lot of... I'm going to just walk up to this person. You what? I'm going to just walk up to this person like you. Yeah, so it works fairly well. But in an environment like we're in right now with the pandemic, you actually have to formalize things and set up a time and set up a bridge, a zoom or whatever bridge and discuss. And so it doesn't work that well. And that's where I think these, yeah, actually, by the way, you've seen all these tools that allow you to get more organized and are being so successful nowadays. No, we met at an event, I think, I think a year or two ago, we're around, around mobile edge deployments. And I think we're like everyone in the, in that case, room was trying to figure out just like, what exactly that meant. And whether we'd be sort of essentially deploying code incredibly close to cell towers. And to make that happen, you need, you have these, these different cost of characters, you've got, you know, different cloud computing companies, technology companies, hardware companies, and, and sort of applying the concept you just apply, you just spoke about, they have different frames of reference. And they're each sort of speaking their own language and coming, coming from different places. How do you, how do you, how do you, like, how do you, how do you, how do you get them to like, what's, what's the shared frame of reference for something like that? And how do you, how do you, how do you get folks making progress? And, and actually, if you could take a second, like to even just to just explain, like, what is, what is the idea around, around running code at cell towers? And why does it matter? And, and why do you need to get different people together? Yeah, so that's a good question. So cell towers and cell phones, right, were developed, I mean, this is a more modern cell phone, but they were developed, essentially, with the idea that it would be a closed system. So you've got the, your, your cell tower here, and the equipment that's behind and actually processes your signal. It's connected to the Internet, it's connected to the old telephone, wired telephone system. And its sole purpose was to receive the radio wave, convert it back into data packets, and then route it either through the Internet or route it to the phone system, if it's just a voice call. And vice versa, it receives the data packets, it converts them into electromagnetic, magnetic waves, and you receive them in the phone, right. So, so far, that's been the focus. With the smartphone, what people realize is that they actually, it becomes very interesting if the system can become programmable. But the way it's programmed is there's essentially the OS here that's opened up so that you can actually play some apps inside it. And that works pretty well, right, the app model, right, so now you can actually program this, which incidentally, from what I'm told, was even a surprise. When, when Apple and the iPhone started that, I think that the idea was that Apple would write all the apps. And then they realized, oh, actually, people like to write apps. And oh, it could be interesting. Maybe people have better ideas than us. And that became just huge. But if you think on the infrastructure side today, there actually is kind of a loophole in some sense, the way it's done. The app talks to the infrastructure, and then from the infrastructure, it then pipes over to the Internet, and it's only on the servers, like cloud player servers, back on the Internet, that actually the corresponding function for the app is run, right, the server, the service, whatever it is, right. So this whole portion here is actually unused, which allows you to do quite a bit. And we've done quite a lot with apps in this model, and the server model. But if this whole portion here, beyond just trying to make it high capacity and low latency, and who knows what else, right, which we are doing in 5G, if now, if you start embedding also, so you have in here, you've got the capability of communication, right? Now, if you start embedding the capabilities of computing, right, so processing, communication, processing, and then you've got storage. Now you have the three elements to start making a programmable networks. So now it becomes very interesting to start looking at what could you do with that. Because those are the three necessary components of making a system that can be adaptable, programmable. So communicate, process, and store. It has all three. And so how do each of the, like, how does everyone in the chain, I guess, I guess, I guess, think about it? Who's running towards it? Who is not running towards it? Like, what is, like, how does each person in the chain think about it? And what frame of reference are they bringing up? Yeah, so that's why it's a little bit difficult. It's a little bit difficult, because the wireless systems have evolved separately from the rest of communications in some sense. Meaning that the, they started off, it was extremely difficult, it still is, but it was extremely difficult to make a cellular phone system work, right, cellular system work. And so everything was done with purpose-built components. They even have their own, they used to have their own software language. They had their own communication, wired communication protocols. They certainly have their own protocols internally. Slowly over time, they've tried to add more Internet protocols, IP, and so on and so forth. But it's still stayed one large monolithic system. And that's been fairly excluded of the other world, which is the Internet world, with all the variety of players. And now there's a transition. If you want to be able to do these new functions, there's a transition to opening up these networks and having these more based on standard pieces of equipment from the IT world. Servers, high-performance routers, and so on and so forth. So that transition is really a difficult one. They're generally speaking, two groups of people there, or companies. There's a group of companies who actually built this system on our benefit from this, the old, the system, right, and attached to it. So they believe that that's the way to continue with the monolithic type of system integrated for reliability, for performance. And there's the other group of more, the challenges, which are startups, but also more the Internet world, where we're saying, no, no, no, no, we can do it differently. And let's change the way things are going. And honestly, it's a challenge between the two to try and find a good transition. The transition is very similar to the transition we had with, you have in any industry, most industries start off in the completely closed integrated world, and then they eventually move into a desegregated world. If you think of, for example, the IT industry, right, computers used to buy them, the processor was made by IBM, Wang, or DEC, you know, the operating system, the screen, the keyboard, the power cord, everything was made from the same vendor, and you couldn't interoperate or switch one vendor to the other. And then there was this transition to the model where it opened up, now you've got multiple vendors for the various components. Yeah, you can, and like now you can, you can, you can at one extreme, even like, you know, use platforms like ARM or, and sort of really like assemble your own machine. Like one startup I find very interesting is a company called Oxide. And they're, again, they're trying to, they're trying to make sort of servers ground up more trustworthy and more programmable. And having the hardware talk to the software more tightly. So, so it's really exciting to see, see, see innovation in this space. Well, we're at a super exciting time. Really, we're living in a very, very exciting time for this space. Yeah. Well, you have a, you have a uniquely interesting vantage point. And, and so I appreciate your, your, your sharing that with us. If I can give you just one, one last question. So like what, what, what, what keeps you most excited? You know, What keeps you most excited about, about, about what you get to do? 5G and this revolution that's going to happen in the wireless world. The one I just described, meaning that the networks are going to go beyond just becoming, being bit pipes and they're going to start doing compute and storage and have all the capabilities to be programmable systems. That's super exciting. Well, well, thanks. Thanks so much for joining me, Stein. This has been a lot of fun. Appreciate it. You're very welcome. Take care.