 Live from San Francisco, it's theCUBE. Covering Google Cloud Next 2018. Brought to you by Google Cloud and its ecosystem partners. Welcome back everyone. This is theCUBE live coverage of Google Cloud Next 18 in San Francisco. I'm John Furrier, Jeff Frick. We're at day three of three days of wall-to-wall coverage. Go to siliconangle.com and the cube.net to check out the on-demand videos and the cloud series special journalism report that we have out there. Tons of articles, tons of coverage of Google Next with the news analysis and opinion, of course, SiliconANGLE. Our next guest is Ingrid Hill, Chuck Roborty, Product Manager for IoT, Google Cloud. Certainly IoT, part of the network part of the cloud. One of the hottest areas in cloud is IoT. We've been seeing that. Welcome to theCUBE. Thank you. Thanks for joining us. IoT is certainly the intersection of a lot of things. Cloud, data center, AI, soon-to-be, cryptocurrency and blockchain coming down, not for you guys, but in general, those are the big hottest areas. IoT is not like, you can't say it's an IoT category. So IoT has to kind of sit in the intersection of a lot of different markets that are kind of pure playing. So I first want you to explain to the folks out there watching what is the Google IoT philosophy? What is the product's trying to do? And what are you guys announcing here? Absolutely. Thanks for having me here. It's really great to be here. And if you think about IoT and if you think about what we have on Google Cloud, we already have a great set of service for data storage, processing and machine intelligence. So we have cloud machine learning engine. We have announced AutoML. So most of those data processing and intelligent services is already there. What we announced last year was Cloud IoT Core, which is our fully managed service for customers and partners who easily and securely connect their IoT devices to Google Cloud so they can start transmitting data and then ingest and store and use our downstream services for analysis and machine intelligence. I mean, IoT is a great use case of cloud because one, cloud shows that you can be incented to collect data. Because now you have lower cost storage and you have machine learning, all these things that are going on, it's great. But IoT is now the edge of the network. You've got sensors, you've got cars like Teslas where people can relate to. And so everything's coming online that has not just an IP connection, anything that's a sensor, the IoT's been just evolving. What is the edge to you guys? What does that mean when I say IoT edge? What is Google view of the edge? Yeah, absolutely, it's a great question. And we identified early on the emergent trend of moving compute and intelligence to the edge and close to the device itself. So this week, as you already know, we've announced two products for Edge. One is Cloud IoT Edge, which is a software stack which can run on your gateway device, cameras or any connected device that has some compute capabilities, which extends that powerful AI and machine learning capabilities of Google Cloud to your Edge device. And we also announced Edge TPU, which is a Google designed, high performing chip for to run machine learning inference on the Edge device itself. And so with the combination of Cloud IoT Edge as a software stack and with our Edge TPU, we think we have an integrated machine learning solution for on Google Cloud Cloud. How does that get rolled out? So the chip, I'm assuming you're doing OEM or deals with manufacturers and same with the software stack as a software stack portable. Explain how you roll those out. Yeah, you know, we are big into working with our ecosystem and we really want to build a robust partner ecosystem. So we're working with semiconductor companies such as NXP and ARM, who will build a system on module using our Google Edge TPU, which can then by used by gateway device makers. So we have partnership with Harding, Nokia, Nexcom who can take those some, add it to their gateway device makers, gateway devices to take it to the market. We're also working with a lot of computing companies such as Adelang, Acton and a couple of others, Olia, so they can build analytic solution using our Cloud IoT Edge software and Edge TPU to combine with the rest of Cloud IoT platform. So we're pretty excited about the problem. But every coin has two sides, right? So the kind of the knock on the edge is now your attack surface on the security side is growing exponentially. So clearly security is an important part of what you guys do. And now this is kind of a different challenge when you're now, your points of presence, not like hard points of presence, but are going to expand exponentially to all these connected autonomous devices. That's a great point. And we take security very seriously. In fact, last year when we announced Cloud IoT Decor, we only reject connection, any connection that doesn't use TLS, number one. And number two, we individually authenticate each and every device using an asymmetric key pair. And in addition to that, we've also announced partnership with Microchip. So Microchip has built this 8BU microcontroller crypto which can have the private key inside the crypto and we use JotToken that can be signed by inside the chip itself. So your private key never lives a chip at all. And so that's one additional reinforcement for security. So we have end to end security. We make sure that your devices are connecting over TLS, but we also have hardware root of trust on the Edge device as well. The token model is interesting. Talk about blockchain, because you know, David Floyer and our analyst team, he and I are constantly riffing on that. IoT actually is an interesting use case for blockchain and potentially token economics. How do you guys view that? I know that's, you just mentioned that this is kind of a thing there. Does it fit in your vision at all? What's your position on how that would work out? You know, we're closely looking at the blockchain technology. As of today, we don't have anything specific to announce in terms of a product perspective. We do use JSON WebToken, which is standard on the web, to use to sign those, using our private keys. So that works beautifully. We're closely monitoring and looking at it. We don't have anything to announce today. Not yet, but I'm going to ensure that the researchers are working on it. Interesting scenario. Okay, so in general, benefits to customers when we're working with IoT, your team, because you have the core, you have the chip, you have the software stack. And there's always an architectural discussion, depending upon the environment. Do you move compute to the data? Do you move the data to the cloud? What's the role of data in all this? Because certainly you've got the processing power. Wow, what's the architectural framework and benefits the customers of working with Google? Yeah, so let's take a specific example. You know, LG CNS, they want to improve their productivity in the factory. And what they've done is they've built a machine learning model to detect defects on their assembly line using cloud machine learning engine. So they've used this one engineer a couple of weeks and they could train the model on cloud. Now, with cloud IoT Edge and the HTPU, they can run that trained model locally on the camera itself so they can do real-time defect analysis at a pretty fast moving assembly line, right? So that's the model which we're working on, where you use cloud for high compute for training, but you use the HTPU and cloud IoT Edge for local inference for real-time detection as well. How do you guys look at the IoT market? Because I mean, depending on how you're looking at, you can look at smart cities, you can look at self-driving cars. There's a huge aperture of different use cases. It could be humans with devices. Obviously, you guys have Android. So I mean, it's kind of a broad scope. You guys got to kind of have that core tech which sounds like you're kind of putting in the center of all this. How do you guys look at that? How do you guys organize around that? Diane Greene mentioned verticals, for instance. Is there different verticals? How do you guys go out that market with a product? Yeah, IoT is a nation market and what we offer as Google Cloud is a horizontal platform. What we call it is a cloud IoT platform, which has got cloud IoT core on the cloud side, cloud Edge, the HTPU. And we really want to work with our partners, our solution integrators and ISVs to help build those vertical applications. And so we're working with partners on the healthcare side, manufacturing. We have Odeon Technologies, one of the partners, to really build those vertical applications. You guys are not going to be dogmatic. Here's how our IoT is going to be led by the Thousand Flowers Bloom kind of philosophy. Put it out there, connect, and then let the innovation happen with the ecosystem. Yeah, we really believe in sort of driving, moving the, having robust ecosystem. So we want to provide a horizontal platform which really makes it easy for partners and customers to build vertical solutions. Another kind of unique IoT challenges, which you didn't have in the past, we've all seen great pictures of the inside of Google data centers. They're beautiful and tight and lots of pretty pictures. Very different than out in a mind field or a lot of these kind of challenging IoT environments where power could be a challenge, the weather could be a challenge, connectivity to the internet can be a challenge. Obviously, and then you need the power when you talk about how much store do you have locally, how much compute do you have locally. So as you look at that landscape, how has that kind of shaped your guys' views? What are some of the unique challenges that you guys have faced and how are you overcoming some of those? Yeah, that's a great question. And this is one of the primary reason why we announced CloudID Edge, which is a software stack and the HGPU, so that for use cases where you have limited connectivity, oil wells or farm fields, windmills, connectivity is limited and you cannot rely on your connectivity for reliable operations. But you can use CloudID Edge with our partner device ecosystem to run some of the compute locally. You can store data locally, you can analyze locally and then push some of the incremental data to the cloud to further update your model in the cloud. And so that's how we were thinking about this, that we have to have some compute locally for the way things are going. Release the hard coupling, if you will. So it's really got to be a dynamic coupling based on the situation, based on the timing, maybe have scheduled updates and these types of things. So it's not just connected. Exactly, it doesn't need to be continuously connected. I mean, as long as there's enough connectivity to download some of the updated model, to download the latest firmware and the software, you can run local compute and local machine learning inference on the Edge itself. That's the model we're looking at. So you can train in cloud, push down the updates to the Edge device, and you can run local compute and intelligence on the device itself. A lot of consciousness we've been having lately has been about how do you manage the Edge, been an area of discussion, why I want to have a multi-threaded computer basically on a device that could be attacked with malware. So putting bounds around certain things, you need the IPK, you want to have as much compute. Obviously, we'd agree, but there's going to be policies you're starting to think about. This is where I think it gets interesting when you look at what's going on at the abstractions up the stack that you guys are doing. How does that kind of thinking impact some rollouts of IoT? Because I'm also going to imagine that you want to have policies. Some might trickle data back, it might not be data intensive, some might want more security, containers, all this kind of tying in. Is that right, am I getting that right? How do you see that happening? Yeah, so when you think about Edge, there are different layers, there are different tiers. They are the gateway class devices which has high compute and all the way to really sensors. Our focus really is on the Edge devices which has some decent compute capabilities and you can scale up to high end devices as well. And when you think about policies, on the cloud side we have IM policies. So you can define roles and you can define policies based on which you can decide which devices should get what software or which users should get access to particular data types as well. So we have the infrastructure already and we're leveraging that for the IoT platform. And automate a lot of those kind of activities as well. Exactly. All right, so I got to ask you about the show. What's some of the cool things you're seeing for the folks that couldn't make it that are watching this video live and on demand? What's happening here at Google? What's the phenomenon on Google Cloud? What are some of the hot stories? What's the vibe? What are the cool things that, cool things you're seeing? Absolutely, so I'm biased. I'm going to start with IoT. We have an IoT showcase where we are, we have a pedestal where we're showing the HTPU and the HTPU board as well. And there is a lot of work which is happening there. There are predictive maintenance demo as well. So I would highly encourage attendees to go and check it out. What are people saying about that, the demos and the sessions? What are some of the feedback? Share some color commentary around reaction. Yeah, we've been getting a lot of positive reaction. In fact, we just had a couple of breakout sessions and a lot of interest from partners across the board to engage with us. So we're pretty excited with our announcement on the edge side. And the whole orchestration of training model in the cloud and the pushing it out and then setting up is, that's what really sort of makes it easy for a lot of the partners. So they're excited about it as well. They're going to make some good money with it too. You guys are making the market, making it, not trying to go too far and play the foundational work, the horizontal scale. Yeah, exactly. And we really focus for the HTPU really focused on performance per dollar and performance per watt. And so that has been, which we're striving to really have high performance for lower cost. That's what we're targeting. In a couple of other things you, the whole serverless capabilities and the fact that cloud functions have become GA is pretty exciting. And Cloud IoT Core is also a fully managed serverless architecture. In a machine, the AI and auto ML, which we announced with NLP and text and speech is pretty exciting as well. And that works very well with some of our IoT use cases as well. So I think those are a couple of announcements which I'm pretty excited about. I think the automation theme too is really resonates well and all that. Because what comes out of that is, humans still got to be more proficient in doing the new stuff, but also they got to run this, right? And then you got developers who have to build apps that drives value. So you got the value development with the applications. And then also the operational side, which is like kind of, I won't say becoming generic, but you know, it's not specialized. This used to be network operator. This guy does this, this gal does that. I mean, it used to be very stove piped. Now it's much more of a, how do you run the environment? Exactly. And to your point, even on the IoT space, it's also very relevant, right? I mean, there are a lot of overlaps between what used to be just DevOps and OT and IT. There are a lot of overlaps there. And so we're looking at it closely as well to make sure that we can really simplify and the overall requirement and the tooling which is needed for building an IoT solution. For the people that are not following Google as closely as say we are, for instance, that aren't inside the ropes, inside the baseball, if you will, in the industry, let's see Google Cloud. They know Google as Gmail, search, et cetera. They look at a couple of years ago, Google Cloud, had App Engine, the OG of Google Cloud, it's been called. What would you say to the folks now that are watching? What's different about Google Cloud now? And what should they know about Google Cloud that they may not know about? What would you say to that person? Absolutely. And the first thing is, we are very serious about enterprise. You can see here the number of attendees who have come here and the how we have multiple buildings where we organize the conference. So very serious about enterprise. Second, I think in back in the days, two years or three years back, we were really focused on building products which works for specific use cases. We didn't think about sort of end-to-end solution. But now the focus has changed and we're really thinking about, we always had the technology, we packaged it into products and now we're thinking about providing end-to-end solution, the framework where for a business user, an enterprise user, they can just take the solution and they know it will work. So there's been a lot of focus on that. And you know, our key differentiator is about machine intelligence and AI, right? I mean, that's where Google thrives. We've been spending a lot of time on it and now we are focused on democratizing AI, not just on the cloud, but also on the edge with the announcement of HTPU. And I really think you guys done a good job with the mindset of making it consumable. In an end-to-end framework with the option, we've got Kubernetes and Container's been around for a while, but to work on multiple environments, I think that is a real mindset shift. Exactly. So congratulations. Thank you. Thanks for coming on. Appreciate it. It was great having you back. Google IoT, just plug in the Google Cloud, it'll suck all your data in, give you some compute at the edge, open it up to partners, really focusing on the ecosystem and enabling new types of functionalites to keep bringing you the data. Here on day three at Google Cloud, next 18, we'll be right back with more coverage. Stay with us after this short break.