 from San Francisco, it's theCUBE. Covering Micron Insight 2018, brought to you by Micron. Welcome back to San Francisco Bay, everybody. This is theCUBE, the leader in live tech coverage. My name is Dave Vellante, I'm here with my co-host, David Floyer, we're covering Micron Insight 2018. Really bringing together memory, storage, and artificial intelligence, talking about AI for good, talking about changing the way in which we work, new workloads. Raj Tallory is here, he's the Senior Vice President and General Manager of Micron's booming mobile business unit. Raj, I think it's the fastest growing business unit at Micron, at least of size. And Keith Crescent is here, he's a Senior Vice President at Qualcomm, Keith, welcome. Thank you. Gentlemen, thanks for coming on theCUBE. My pleasure. Thank you. So Raj, let's start with you. What are the big trends that you see that are driving that at 60% growth rate in your business? Yeah, I mean, we are finding now that mobile phones and the use of memory in mobile phones, and this is DRAM, and also use of storage in mobile phones. This is where we actually flash. Both are growing inside a mobile phone. We've seen people launch four and six gigabyte phones, and now we have our customers talking about eight and 10 and 12 gigabyte phones in the future. And one of the big reasons is the applications. A lot of machine learning and AI applications in the phones, and those are driving the need for a lot of increased, both storage and memory. Keith, so talk about what's changing in your business over the last several years. I mean, mobile obviously exploded onto the scene, but now people talk about AI and mobile, and just increased use cases and applications. So what's going on from your perspective? Yeah, I think from a handset perspective, there's been some visual changes, right? The screen's got a little bigger, the bezels get smaller, the phones get a little thinner, and so you see some of the visual cues, but really the excitement of what's going on underneath the visual cues. So the amount of AI processing is accelerating at a rapid pace. A lot of advanced cameras, moving from two, three, four, now there's phones with five cameras. Cameras to recognize depth perception. There's a voice recognition. So there's a lot of AI processing and a lot of capability getting into the phone for a variety of applications. So okay, so voice recognition, that makes sense. You'd use AI for that, but you're talking about AI and visual in the handset. Talk more about that, explain how that all works. Yeah, so actually I would argue that today, imaging is probably the primary application of AI. So cameras used to be just capturing pixels, but now it's about much more than capturing pixels. It's about understanding what those pixels are. So are they recognizing objects or food or something else or is it facial recognition for things like payments and biometrics? And so the cameras now are much more intelligent. And with multiple cameras, you're adding depth information. So then you start to get to much higher levels of realism for things like avatar and gaming and other areas where the camera is capturing and also perceiving. So not a hot dog for you fans of the show, Silicon Valley. So Raj, what does this all mean for the memory and storage requirements? Yeah, I mean, as Keith mentioned, I think the mobile phone with the process from Qualcomm and other processor vendors has really gotten to have a lot more camera applications and a lot more AI applications. Now there's a difference when you actually drive applications that need AI and machine learning, versus other applications. And the difference is that the compute paradigm in AI and ML is different. In the sense that these are like complex neural networks that need a lot of data very close to the processor to achieve this result. So as more AI applications come in, we are actually finding that the customers, which are Qualcomm's customers and our customers, are asking for more processing from their side, but they're also asking for more higher speed, higher density memory to couple tightly with the processor so they can realize these AI applications for the consumers. And also storage because sometimes you want to store a lot of the images and videos on the card. So both storage and memory are increasing as these applications come in. Well, I'd like to ask about the gaming side of things and the AR. That seems to me that it's starting to improve to a level which is frightening in the reality. What do you have to do to get it to that stage where you can have true VR and have games, for example, which exploit that? Sure. First I'll talk about gaming specifically. Gaming obviously continues to grow, a lot of money in gaming, gaming tournaments and so forth. The gaming certainly is getting more realistic with better graphics and so forth, better displays. Multiplayer gaming, very hot, right? Multiplayer gaming requires very fast connections, very low latency connections to another source so multiple players can play at the same time. Also many times for games, you'll have a play with partners, maybe you're on a team of five, but now and coming soon, maybe that team of five, you don't need to find that fifth person or that fourth person. Maybe there's an AI engine that's running similar to the human capability and you're actually playing with a simulated player, right, from teams. So there's, and that requires, opens up a whole new area of processing for the games. And then I think for AR and VR, AR is a little further out than VR. AR requires some more advancements with respect to optics and so forth. VR is taking high-end displays and AI and graphics kind of packaging it all into one to really change the paradigm of how you interact with the computer. How does 5G change things? Is it as much of a game changer as people think or is there just so much data that it sort of allows us to keep pace? I wonder if you could talk about 5G. Yeah, so every 10 years or so there's a G transition. 3G in 99, 4G in 2009, now 5G in 2019. So it's not a 10-year cadence and every time when you have a G transition, you couple that with a transition in computing and it changes the paradigm. So what's going to happen is 5G is going to bring a lot more capacity, a lot lower latency, at the same time AI is coming in. And that together is going to create a pretty powerful platform for applications for the future. And then of course there's so much more data now. How do you guys keep up? Because you mentioned, you've been talking to the street and you mentioned this morning that the rate of bit density is moderating. So how do you guys keep up? Where are your investments that allow you to keep pace? Yeah, so just maybe a little bit of comment on 5G. I think as the bandwidth to the device gets faster and faster, there's more and more data that comes in. You can imagine for example, one of the things a lot of people like to do is to download content. If you look at Netflix, if you look at Amazon Prime, if you look at even Direct TV and I have all of those, you can actually download the content now and watch them offline. So as it happens and that content gets to be 4K and even higher frame rates, the amount of storage that's needed to download that is getting more and more. Like my wife upgraded her phone the other day and the first thing she said is, I want the entire Netflix season on my phone when I'm in the gym. So the simple things like that have changed a lot. That's one of the reasons why the storage is getting so much. Now when you go to 5G and the download speed gets higher, you can download a 4K video really quickly. Now you got to put it somewhere after you download it. So that's actually driving the need for this. So before people wouldn't download a 4K video because where would you put it? So as we increase storage, that kind of stuff comes really fast because you couldn't take a long time to download before. So as the bandwidth gets higher, the storage requirements and the memory requirements are getting higher. What we are doing on that front is we are investing a lot as Sanjay and Scott talked earlier, both in our fab capacities, both in our technology transitions. We have a lot of new interesting technologies like new emerging memories that actually like 3D crosspoint we talked about, that kind of blur the line between storage and memory. So there's a lot of new interesting technologies that will actually take advantage of that. Super exciting time. So going back to the image processing side of that, one of the trends, it seems to me, is that the processing is going further and further to the edge itself and going inside the camera itself. Can you talk a little bit about that and what it's going to take to put that your memory technology, your bandwidth right inside the camera itself? Sure, so there's no doubt that you want to maximize the capability on the edge as a first step and then you want to reduce the latency as much as possible to the cloud as a second step. So on the edge, if you had something, for example, if I'm taking a picture of you and I want face recognition, I don't want to take every frame and send it up to the cloud, right? Because I'm going to waste bandwidth. So I want that capability on device and that's true for a variety of different applications. You want to maximize the capability on device and then focus on the fast connection so the cloud and the device from a latency and bandwidth perspective are much more tightly coupled. You know, when you think about the evolution of computing, you know, obviously everything was centralized and then PCs, the world was about PCs back then. It was kind of, the centralization was a blip on the screen compared to the PCs was everything, you remember that? And then of course, mobile drove cloud through the roof and now with the edge and cloud and mobile, you're seeing just this ubiquitous capability that senses, now you bring in AI, you bring in machine intelligence. What do you guys envision for the next 10 years in terms of what the world looks like, centralized, decentralized, distributed, intelligent? I mean, it's just mind boggling. What's your vision? Well, I think if you look at client devices, client devices certainly generate a lot of data. Maybe we get a little bit of data from a sensor and a bridge, but maybe we get a lot of data from the car traveling across that bridge. And what you need to do is you need to make sense of that data locally and then transmit it back to the cloud. So you want the cloud to have the most useful data or sorted data, right? Data that can then improve, you know, automated driving or reduce traffic accidents and so forth, but you don't want all the data sent there. So what's going to happen is on the edge, there's literally, you know, devices going from smartphones where there's about one and a half billion a year to billions and even trillions of IoT connected devices. Any device that has a computing element also is going to have a connectivity element, also is going to have an AI element. So it's going to be a much more connected world as opposed to just connected people. Yeah, I mean, I think Keith explained it very, very well. You know, if you step back a little bit, I think the history of technology and evolution has been very similar, right? You know, having been in the industry for a while, we all remember the times when we said we just need one mainframe and everyone needs dumb terminals, right? Then we went on to say, hey, you know what? I think distributed computing with everyone having a PC is the right thing to do. Now we are back to maybe we should have everything in the cloud and the edge devices. So I actually think, you know, world goes cyclical and the more you do at the edge, the more it drives the need for the cloud and we call it the watcher cycle. And I think the best way to think about it is you want the edge devices to send information, not data, which means data needs to be processed with memory and compute to become information and then you send the information to the edge into the cloud. Yeah, and I guess my point was that, you know, I've been around a while too and you can see the pendulum swing. I feel like the pendulum is not swinging anymore. It's just exploding. Both sides, that's right. It's just, it's really an exciting time in our industry. Guys, thanks very much for coming. Thank you. I really appreciate it. Thank you. All right, keep it right there, with a crown in sight from the Amarcadero. You're watching theCUBE, right back.