Stanford Seminar - The Future of Edge Computing from an International Perspective

well good afternoon everybody and welcome to the Thursday September the 26 2019 session in our series on international technology management this year our theme is about edge computing I'm Richard Dyer I direct the US Asia Technology Management Center and if you believe that I started this series when I was only three years old I can tell you that we've been doing this series of lectures for 27 years now we pick a different theme every year and we examine the newest kind of information that we can get not only about work that's being done here in some technical area but also work that's being done in Asia in some technical area so first of all I'll talk a little bit about the series itself and then kind of get into edge computing and what that's all about talk about the drivers of edge computing adoption and then look a little bit at upcoming sessions in this series so the first biggest message is welcome to everybody we're really happy that you're here I know that this is not required for anybody if you're a student and it's wonderful to have a mixed audience of people from inside Stanford and people from outside Stanford one of the things that I've always wanted as we've done these series over the years is to have more dialogue between these two groups so that you get to know the industry people and why they come to these sessions and what things they're interested in and conversely let's hear from the student side how things look to you and what kind of fresh ideas you might have in regard to the general topic of this of the of the series so we'll be going every Thursday from now until December the fifth we'll skip Thanksgiving Thursday of course we are an industry affiliates program that was an electrical engineering for 25 years and for the last two years has been understand for global studies and the School of Humanities Sciences so we are interdisciplinary our continuing mission is to introduce trends and current developments in the intersection of a particular area of technology in business so these are not serious technical presentations but there is a lot of technology involved so if you feel uncomfortable either because there's too much technology or too little technology you're in exactly the right place we think that having a discussion from both perspectives will help everybody to understand what's going on and find useful things for yourselves so we do make special reference to the US and Asia we have some people coming in from Europe will be speaking in the seminars but mostly this is about us and Asia because of kind of traditionally there's not as much knowledge exchanged across the Pacific as you would think and we have the speaker slides and also videos for the last at least six or seven years of these programs up on the web so if you're interested in a particular topic run a search and you'll likely find us for it oh yeah you can see how these things go let's look at edge computing which is really the next major architectural stage in the evolution of networked information processing thanks to Microsoft for having these cute you know arrow shaped templates that you can use really there are four stages that I would say have occurred or are occurring in this history of networked information processing mainframe terminal client-server cloud computing and finally edge computing so to look at these in a little bit more detail the era the mainframe terminal goes way back am I the only person in the who remembers Wilbur anybody else remember Wilbur that was Stanford's mainframe and a lot of things were done on Wilbur it was time access and so if you really had a program that required some number crunching you kind of had to do it in the middle of the night so the clients were really dumb terminals the first computer I ever had even a PC you couldn't display graphics on the PC you had to buy a special graphics card the connectivity was by a private custom network on timeshare and that meant that the software for these systems were not compatible with each other I worked for a Japanese company in the early 1990s and with a company with only 50 people in it we had our own mainframe computer because nobody had a PC the mainframe did financial processing it did personnel records it did you know the basic forms and kind of stuff but there was absolutely no compatibility between our computer and our customers computer so you know that was the way things started out notice how you had people who are doing data processing right the second stage was really when the engineering workstation came along and engineering workstations were very powerful individual desktop units that could do a lot of things and if you could network them together you could communicate you could do group projects you could do all kinds of things I remember when I first got to Stanford in McCullough building I helped run Ethernet cable under the ceiling or on top of the ceiling to link the different computers in our own little network so at that point in the early part of this era which was really around 19 it was mostly workstations and a few pcs and later you had PCs workstations and also peripherals printers and so forth on the client side and during this era internet protocol took over so that pretty soon the definition of a network was not what kind of protocol you were using it was basically some sort of you know firewall or something else the standardization of the application software could happen at this point this was the era of the wind hell monopoly right Windows and Intel and things like whatever could run on the client especially in the later stages Microsoft Word Microsoft Outlook Microsoft whatever and the intensive processing intensive applications would still run on the server somewhere the server would also access to things like a central database and so forth near the end of that era so from dumb terminals with all the processing in the mainframe suddenly have a lot of processing happening out on the client-side and some processing being done inside the server that's kind of unifying the group along about the time the smartphone came along this evolved into the era of cloud computing and what happens in cloud computing is you have all sorts of different client types with various levels of intelligence probably smartphones and PCs are still the dominant ones but you've got tablets you've got mp3 players you've got sensors you've got smart appliances coffee cups movie tickets anything can be connected to this cloud now and really Internet Protocol is by far the dominant thing that is being used to connect the clients into the internet and really a local area network is nothing more and some sort of a secure position on the Internet most of the processing has gone into the cloud what happens is that a copy of the devices operating system and the application it's trying to run will go into some server in some data center through a process called virtualization and so there will be a virtual machine meaning a copy of the client inside the data center that will be where all of the actual processing in the application happens this is efficient because you can have three million cell phones all connected up and and you know however many different operating systems you can have 99 percent capacity use on your servers instead of running every single one of those 3 million devices and dependently so it's a much more efficient way to handle the processing power efficient and so forth some interesting things the legal system still hasn't figured out how to deal with some aspects of this if you are infringing somebody else's copyright where do you sue the person do you see the person where the client is or do you sue the person where the server is because the server may be in Finland and if you're trying to engage in real severe political kind of whatever legal issues like that in some countries you can remove the physical server from the data center if you can identify it in other countries you cannot so the legal system still hasn't caught up with cloud computing but this is where we are now we have hit the era of cloud computing full-on - back in 2016 so three years ago there was a survey of almost a thousand IT professionals that were conducted yearly and in the 2016 survey about 72% of them said that their company was using at least one cloud application by 2018 that number had gone up to 96% so it's ubiquitous what kind of things are being done in the cloud email of course Gmail is a cloud application all you're doing is the reading and inputting on your device all of the processing of the emails are happening in some data center somewhere Stanford's email system is the same way you have other things that are very you know useful thing customer relationship management voice over Internet Protocol human resource management helpdesk kind of thing the interesting thing about these particular areas this shows how many what percentage of companies in their sample are actually using the cloud four of these different business functions and the middle sized companies companies between 500 and a thousand one hundred and five hundred employees actually have a higher use of the cloud than do the big companies partly because the big companies have their own private set ups and they don't need these standard applications that are provided by some third-party vendor now we get to the edge so you've still got this kind of cloud in the middle and what's happening is that there's a lot of intelligence in the devices at the edge in the clients at the edge you've got in addition to smartphones which continue to become more and more like miniature computers you've got connected cars you've got airplanes doing a lot of information processing being connected you have drones you have networked robots and for the things that are less intelligent sensors in the smart factory you will have some sort of an edge server or a gateway server or a micro data center out near the edge and in regard to connectivity what you're seeing is integration of really high-speed mobile access as well as fixed network access so our our presentation next week is by Professor Paul Raj on 5g networks and really from 5g you cannot tell the difference between mobile and not mobile so you it's interesting how this is basically an optimization problem where some applications you really want to run them at the edge because it takes too much time too many nanoseconds for the electric signal to get to the data center and be processed in the data center and get back now that sounds kind of hard to imagine but if you're running a self-driving car you really don't want the data to go all the way to the data center and back before you in order to avoid some obstacle so really it's a continuing optimization one of the things that's happened in IT is with the in the spread of artificial intelligence there's this need for huge amounts of data to process some problems especially in business analytics so business analytics tend to be closed done closer to the cloud whereas the real-time applications navigating your drone tend to happen more out at the edge you get rid of all the latency involved in going back and forth to some data center some some you know thousands of miles away it's important to remember that edge computing does not replace cloud computing so unlike the client-server era in the mainframe era the mainframe era kind of pretty much declined although you can still buy mainframe IBM computers I think but really edge computing is sort of an add-on to the cloud computing structure you'll see this processing at the edge for some things I just mentioned and processing that requires a lot of data or heterogeneous data from various sources will tend to be done more in the cloud you'll also hear the term fog computing and fog is typically this sort of micro datacenter range in between the edge devices and the cloud so if you think of a rack of servers typically they will define a micro data center is anything with less than five racks of servers this is in the fog and again it's an optimization problem how do you get the processing done in the most efficient way possible so why is this happening first of all the amount of data generated is just exploding I've got a couple of slides on that and a lot of the data that's being created like this real-time data as your self-driving car is running down the road is so transient it doesn't make sense to store it in a data center at all keep it at the edge deal with it there there's this huge demand for inform or real bail date real-time data processing concerns about privacy and security actually kind of drive edge computing it's really hard to share medical data between one hospital and another and if you're doing an analysis of data using the cloud what happens is the data that you have on your edge device whether that's a computer or a little mainframe or a little data center inside your company at the edge is going to be cop into some database a new file right into some database in the data center where it will be put along with a lot of other data for the analysis that process of copying increases the security risk and so the HIPAA regulations in the u.s. make it really difficult to do that so this argues for doing more analysis of the data at the edge and we've got a company coming in in a few weeks that has a really interesting solution to how to combine data from lots of different hospitals in an analysis but have the data analyzed inside each of the hospitals not inside the center in the cloud another thing that's really driving the growth of edge computing is a new life in the semiconductor industry so most of the computer chips in the data centers in the big racks and the data centers are either CPUs or GPUs graphic processing units so NVIDIA has made tons of money selling GPUs into the data center and this is fine there are multi-purpose chips though and so if you can design a single purpose or an application-specific chip right an ASIC you can usually make it run even faster and with less power consumption which means it's great for edge devices so this kind of new wave of computer chips is also driving the growth of edge computing we will have a presentation in two weeks from the director of AI marketing from Intel who is a good friend and so I'm glad that we can have this coming up - yeah this explosion of data look what is going into the central cloud back in the old days it was all form based what the the technical term is structured data and then you have blogs and social media click-through data and so forth that the technical term is unstructured data along about 2015 there was more sound files and video files floating around the internet than there were text files that was a major kind of turning point and it's not all Netflix a lot of it is thanks to Alexa and Syrian and this group but we're really right at the early ages of you know a lot of Internet of Things data sensor data connected smart devices robots everywhere each time you have a new type of data coming in you have to reconfigure your analytics to be able to deal with it there's a lot of really interesting company start up companies out there now that are bringing in new types of data to very old problems so there was a company what is the name of this company zest financial I think that has the ability to look at lots of different kinds of data including how long it takes you to fill out the form that they put in front of you and your if you're applying for credit and so they have a huge joint partnership with the Chinese e-commerce company JD comm now that allows extension of credit for e-commerce to people who are unbanked they've never had a bank account they don't have a credit card and so this is kind of the new thing that you're getting by using these new data types but as this new data comes online one of these graphs right for the exponential increase in the amount of data that is created now just to give you an idea of what they're really talking about if a gigabyte was the size of one brick a zettabyte would be the size of the Great Wall of China so around 2010 there was a little over one zettabyte of data that was created during the entire year by the year 2025 the prediction is that there will be more than 160 zettabytes of data created and during that particular year some of it will be stored long-term a lot of it won't one of the interesting things about this is that the real-time data is going up like crazy about 40 or 45 zettabytes of real-time transient data is predicted to be created in the year 2025 so one of the problems with this kind of new IOT data is that it's unpredictable if you have a fire or an explosion or something unusual happening all of the sensors in the factory may go haywire and fire off all at once so suddenly you give this huge spike in the amount of data that you've got that's a challenge for you know processing of the data and so the real-time data the closer you can keep it to the edge the better it will be how much money are we talking about follow the money right so I looked at five different market research reports on edge computing and just for comparison I looked at three other market research reports on cloud computing all of these were published this year none of them are older than 2019 but they would look at you know four of the four of the five edge computing ones started off in the year 2018 one of them started off in 2019 and you see a wide range of predictions for how big the edge computing market is going to be in the year twenty four twenty twenty four twenty twenty five you see quite a wide range in the cumulative annual growth rates that are predicted by these numbers but if you'll notice some things about this notice how much bigger first of all the cloud computing market is than the the edge computing market so edge computing is still almost a subset of cloud computing they're not directly contiguous be careful they're there they're pulling their data in different ways but if you have a cloud computing market of somewhere around 700 billion dollars a year that's twice the size of the semiconductor industry that's really about half the size of the entire retail industry in the world and so it's you know this is huge huge amount of money and quite respectable growth rates that are interesting to venture capitalists right when you get over 25% of the V so you start to pay attention go ahead these were independent market research reports so each one of these market research reports and these have nothing to do with these ok they're separate reports yeah and one reason that that's an important point is because there are so many differences in how they calculate and forecast the size most of those are definition differences are they looking in the narrow sense at just edge computing services as opposed to other cloud computing services or are they looking at a broader definition that would include some of the growth of related technology businesses so for example 5g networks which we're going to talk about some more are predicted to be a huge market what's the overlap between the market research predicted for 5g and the market research predicted for edge it's probably quite a big overlap the two are really going to grow in tandem with each other and really some of the cost of a connected car is going to be related to the data transfer problem the edge computing problem yeah most cars have a lot more information processing power than my old PC did back in the year 1982 so some of the value of the cars that are going to be sold of the energy systems that are going to be delivered will come from the computing power should that be included in the definition of what they're measuring it varies from company to company but consistently across you see how big cloud computing is and you also see that the growth rates predicted for edge computing are higher than the growth rates predicted for cloud computing that's another kind of important indication that edge computing is becoming more and more important to the world as we know it so if you turn this into practical use cases what are we talking about first thing smart buildings if you do a search on Internet of Things IOT it seems to me that 80% of what I'm seeing coming through the news now is all about devices for the home you know smart thermostats or smart speakers or whatever but there's also not only homes but factories warehouses I could have said hospitals smart manufacturing process control is also an edge computing function you've got the self-driving cars that people talk about all the time when they talk about edge computing drones airplane systems connected car services these are the services besides actually running the car but how you're going to have location-based advertising and so forth mobile augmented reality networked gaming is going to be a driver of e-commerce and things like Amazon go register 'less stores that information is being calculated somewhere some of it's being calculated on your edge device your cell phone and other is going into some sort of a local server healthcare monitoring wearable medical devices part of this is the security and privacy issue and part of it is the fact that that's real-time data if your heart is starting to do strange things you don't want to hear about it three hours from now physical security and surveillance systems we'll see some examples of these especially when we look at what's going on in Asia energy systems telemedicine agricultural systems control twenty years ago NASA and Stanford had a really interesting joint project on telesurgery the networks just weren't ready for it then now you're really getting to the point where you could probably have telesurgery where the actual data transfer time to go thirty thousand kilometers to a synchronous orbit satellite is too long to have the surgery being managed up in the cloud somewhere by the satellite right so you want to do it at the edge and maybe virtual assistants and I put this up because I'm really not sure just how much of your conversation with Alexa is happening locally and how much I was happening in the data center certainly she's getting her information in the data center right she's doing her analysis back in the data center but the actual processing of you know text-to-speech the actual understanding of what you're saying to her I'm not sure how much of that stun at the edge and how much Ivan's done in the clouds anybody no okay we'll leave that one find out see if we can find out go ahead with all the assistance that privacy consider right great thank you that's thanks very much great to have that so the second half of what I wanted to talk about today was the kind of regional perspectives what's going on why are we doing this series with the question mark different directions for the US and Asia first of all edge computing has definitely hit the asian IT world some of these market research reports are predicting Asia to have the highest growth rate of edge computing of any region in the world and a lot of people in Asia say that edge will be important there was a survey done a couple of years ago by this company of almost 10,000 people about 25 percent of them were executives 25% were IT professionals 25% were other business people in Asia and I forget I think the other 25% might have been independent consultants but in Southeast Asia and North Asia and Australian New Zealand 62% had plans to start edge computing projects within the next three years and 97% of them had plans or said that edge computing would be a relevant part of their business or IT strategy beyond the three-year limit so does that kind of look a little bit like this cloud computing graph that we this one we're in 2016 you had 72% of the companies had a cloud application and now you've got 96% so it's very much a few years behind cloud but definitely going in the same direction Asia has been deploying a lot of really interesting edge computing projects some of them in China involve a test bit or are things like a testbed in Ken Jun for video optimization and security monitoring China Telecom has this smart parking project in Yunnan and in Greedo and there's actually a similar project in Taiwan that's being sponsored by a different telecom company one interesting thing about those two examples next week we're going to talk about 5g networks and regulation in China divides up the responsibilities of mobile versus fixed for communications so China Telecom is the the fixed telephone company right the old wireline group and China Unicom was the mobile communications company suddenly they're in each other's turf and so it'll be interesting to see how the Chinese government angles that I would say that Baidu's Apolo platform which is a technology development and also operation platform for self-driving cars is a good example of an edge computing project taiwan has got this smart streetlight system there's a nice photograph of this you can find on the internet so this is actually in operation now where to save power they've got edge computing going on in each of the streetlights this spring South Korea announced the world's first 5g enabled autonomous vehicle test bed which is actually a district regular streets so normal people are driving through this area - it's not a closed track and this was they claimed to be the first one in the world they have said that they're going to build out their nationwide 5g network by next year so I'm hearing not only government investments involved in this the this this project is actually run out of hon Young University although it has government money behind it and a lot of company money behind it - but I'm hearing that the Korean big companies like Samsung and and Eun Dae and so forth SK KT are planning on investing upwards of 20 billion dollars in edge computing in the next few years maybe that's because the Japanese had said that they're going to put 14.4 billion into it Japan was a little bit behind and so they came to the table late but they're definitely trying to move forward with these major investments and DoCoMo also has one of these projects that's looking for video analytics to be done on surveillance cameras yeah we'll see how that goes with the data privacy issues - now what about the different directions the reason that I really started to put this series together was over the last couple of years I'm watching this process of decoupling between the US and China and so what's going to happen right a lot of the things are in the news there are restrictions on US firms from buying equipment or selling hardware software to a certain list of companies that the government maintains that are considered national security concerns Huawei and ZTE were put on the entity list but then they have kind of temporarily been removed off of it and so what happened as a result while we developed their own operating system and the reports about this Huawei operating system for cell phones that I see is that it really is very different from Android which is what they used to use Huawei and Alibaba have both announced their own AI processing chips the interesting thing about alibaba's chip is they don't plan to sell it it will only be used for company internal applications the US administration is putting pressure on firms to move their manufacturing back to the US so that's happening some but it's also moving manufacturing out of China and into Vietnam so this is one of the things where the global supply chains are definitely realigning and there's a lot tighter monitoring and control of foreign investment inside the United States - Lee wrote AI superpowers has his investment firms in ovation and Sinha vation appears to have pulled out of American investments now it's kind of questionable as to whether this is serious and permanent or whether the person who is here in Silicon Valley has just kind of gone home for a while but clearly their investment portfolio has moved much more toward China than and out of here they do have investments in about 40 companies here and so somebody is still taking care of them a couple of some acquisitions by Chinese companies have been blocked some Chinese investors have been ordered to divest from their US investments and so not surprisingly look what happened from Chinese foreign direct investment in the u.s. from 46 billion dollars in the year 2016 we are now down to one tenth of that 4.8 billion dollars in 2018 so as the money doesn't come the systems may get more and more different the ecosystems may get more and more different and it's not only the US administration China is also focusing more domestically things like the indigenous innovation policy things like the made in China 2025 which has kind of been backpedaled by the Chinese government are definitely aimed at doing more things independently and not having connections to the way the technologies are developing here there's a new set of cybersecurity policies coming online in China and that November I think and they appear to have kind of special targeting of foreign firms okay well what goes around comes around if we do that with investors that's going to happen on the other side of the Pacific and Chinese regional influence is increasing I was very shocked to find that trade between China and the rest of the indo-pacific region is twice as much as the trade between China between the u.s. and the indo-pak region so China accounts for 24% of exports from indo-pak and the u.s. accounts for 12% of exports from indo-pak now a lot of this these two bullets had to do with components being made in one Asian country and being shipped to China for final assembly and then they would go to some other market well what's happening is the some other market is no longer the US and Europe its domestic China so you're really seeing a major difference in the way business is being done the regional cooperation economic partnership I think that's what that stands for is still going on the negotiations are still continuing China this is China's answer to the TPP the trans-pacific partnership which the u.s. pulled out off as soon as our current administration started so Japan is Japan and Australia have tried to keep the TPP going it does not include China but in some ways I think the artis RCEP that does include China is more active at the moment you're getting a lot of investments from China into Southeast Asia so this made a lot of news when dd2 shing and Softbank put two billion dollars behind the ride-hailing company in singapore grab but i was fascinated to find out that chinese venture capital investments in asean startup companies increased four times four hundred percent in the first half of this year to an astounding amount of money six hundred and sixty seven million dollars u.s. that probably is because of a few mega deals big deals that are more than 100 million each but it's still an amazing increase and the belton Road is still alive and well they've gone up in there Southeast Asia investments by twice as much in the first half of this year over the second half of last year so yeah China's regional influence is increasing although the people who do business in Southeast Asia point out that Japan still has a lot more influence there than China does and a lot of the big Japanese companies are trying to do corporate venture capital and trying to become more actively involved with the growth of the middle class in Southeast Asia so this is what's it going to do to edge computing right this is the question yeah go ahead in Thailand and Vietnam and Malaysia and in that regard it's a little different from world patterns of investment into Southeast Asia where most of that is going into Singapore and the second to Singapore is Indonesia on a worldwide level go ahead thank you by the US government of the long term medication that would connect Asia to Africa involving many many many nations prevention of Chinese domination of technology most bringing through this I think one belt one road did shake up a lot of people and made a lot of people concerned whether this is purely emotional you know we've got to contain China if you're Chinese that's what it looks like they're they're just trying to keep our economy from growing they're trying to keep us from having the influence that would naturally come with the world's biggest economy if you're in the u.s. though you worry about the national security aspects of someone who is not necessarily an ally having more influence around the world than you do so I don't know I think an American would argue back that it's not aimed at containment but that it is aimed at meeting our national security needs okay but yeah this is this is one of the big problems with this is it is it looks very different in the two different places and I think that what you can say for sure is that unless we have some things to do together then the two systems are just going to go further and further apart now my take on that no matter which side you want to follow in this is that the ultimate losers are probably consumers because when you have a small number of extremely large companies controlling the market and over there it's you know b80 Baidu Alibaba $0.10 maybe put in huawei - right and over here it's either Fang or fog or whatever you call it Facebook Amazon Google and and whoever Netflix people have less choices so if you really do have world competition normal consumers should have more choices and that should be better for everybody that's the that's the way it looks right yeah it is that's exactly right now again people have very impassioned views for both sides and I don't really need to get into that in this series just recognize that these are the consequences you're going to face right so how's this gonna play out it now you know in in edge computing as I stand here in front of you I've got to say I don't know that's why we're doing this series and so I can bring in people who I think will bring us information from different aspects of the edge computing world that might enlighten us as to what kind of emerging trends are beginning so next week I'm really happy that Professor Paul Raj can come and speak to us about five g 5g fifth generation mobile fifth generation communications networks are really a huge enabling factor for edge computing but the amount of infrastructure and investment is staggering because of differences in frequency you have to put the towers for antennas for a 5g system a lot closer together than you do for what we have now for G five times as many base stations space station towers for a 5g network so this is a huge amount of investment that's going to be required to bring in 5g it's one of these key technologies that the US has been concerned about and one of the reasons that it's it's definitely being treated treated as one of these national kind of security implication technologies and professor Paul Raj is not only an expert who has been on the faculty at Stanford for a long time and won the Marconi prize he's an active consultant consultant the industry and adviser to foreign governments and I promised him I wouldn't say who but he's talked to a lot of these countries about adopting 5g technology so I'm hoping he'll address this next week two weeks from now we've got Gary from Intel who's going to talk about these chips for processing AI functions gary was with a startup company named mo videos that had an AI acceleration device and they were bought by Intel they are going to definitely Intel Intel is is is really at the front of a lot of these new initiatives that are related to edge computing and Gary can tell tell us about them he has worked in Japan for many years he has also a good you know kind of grasp of Japanese language in you in addition and so it'll be interesting to get his international perspectives on the 17th we've got dr. biker who used to run the cars lab at Stanford he was the executive director of cars lab and dr. Martin C R Hui who is the head of Alliance innovation lab Nissan right used to be the Nissan Research Center but because Nissan and Renault are having such good relations with each other they changed the name anyway they'll talk about edge computing and autonomous vehicles then on the 24th we have our speaker about medical data federated learning is this approach where instead of copying the data into the data center the data center has what they call the global state of some algorithm the processing engine and it will choose which of the edge devices typically computers in this case it will send for a sample right you've got to pick your sample anytime you're going to do an analysis for AI and it sends the algorithm to the edge devices where the processing is done and then once those edge devices do the processing they send the updated algorithm back to the center where it creates a new global state algorithm it's a really clever approach instead of moving the data around you move the algorithm around this is a term that you will hear used a lot from Google Google I think may be responsible for the term but in any case we've got a real use case for it this is a startup company and they are actually using this for medical data analysis on the thirty-first we've got two people coming to talk to us from the tiny ml ml machine learning consortium and I'm delighted that they can talk to us because this is also how do you get the operating system and the overhead on the edge devices down as small as possible so that you can do artificial intelligence processing at the edge so that'll be an interesting session on the seventh we've got the co-founder and CEO of Greene roster which is basically a platform company platform technology company for the mobile gaming and mobile interactive other noble interactive application industries yeah and then on the 14th of November we'll have a session live from Tokyo I will be in Tokyo Japan and interviewing dr. Ken Amato who's the chief scientist for a little startup company called leap mind that was strongly recommended to me by a large Japanese company our and so they have done a proof-of-concept with this big Japanese company on how to use artificial intelligence to control doors on subway trains and if you've ever been on the subway in Japan you know how hard it is and people are always squeezing in at the last second if you can save a second in closing the door more efficiently than before the whole system just explodes it's the equivalent of taking one tomato out of the salad on every airplane meal and saving five million dollars as a result so anyway he'll he'll speak I am talking with a speaker about the 21st but I'm not going to go any further on that we don't have class on Thanksgiving and then for the last session in the series I'm really happy that dr. yoky Matsuoka has agreed to come into a fireside chat and talk about how she sees edge computing developing she has a wonderful perspective in this she was the CTO of nest the home energy management company she is a robotic specialist she really developed the robotic arm and yeah so it'll be great to have her come up and wrap up the series so that's really what I wanted to tell you about today this is what we have coming down the pike it's like a few minutes if you have questions or comments go ahead yeah you see terms like Cold War and a lot of the press reporting about this kind of technology divergence I actually saw I think it was Hank Paulson who is a really well-known kind of economist journalist who said that you know the result of this decoupling and that's I think what the administration calls it could be the equivalent of an economic Iron Curtain we're just you know people on that side have their economy and on our side we have ours that's true and I think that both sides have a point in some ways right because we do have to look at national security I will say my own personal opinion China is going through economic development very rapidly and a lot of what I see now kind of reminds me of Japan in the 1980s and Japan had reached the stage of being an advanced economy but it kept some of the old developing economy policies and customs going on and so the country needed to change and Japan did a lot of the change by itself the big difference is that Japan was totally dependent on America for security and recognized itself as an American whereas no one can argue that the economy of China will eventually be four-year four times as big as the American economy there's that many people if their income levels approach those in America the economy will be much bigger and so I do think that you have to consider your national interests but is it a zero-sum game I mean again on personal interest would be to go back to more rules based kind of discussions to have international organizations like the World Trade Organization handle a lot of the trouble but that's you know here we are we're moving in this direction what's going to happen I saw one back in the back maybe that the less developed infrastructure in developing countries would encourage edge solutions to be found first even before you come up with some cloud-based solution and certainly you know you've seen that leap frogging over and over again mobile communications jumped in developing economies because they didn't have to string copper wire throughout the country and then the next thing is that you're seeing mobile applications like mobile banking much further ahead in developing countries where people didn't have an alternative this is satisfying a basic need that before they didn't have a way to deal with so yes that's that's a very good point it's a very good possibility who's going to do it so that too I'm not sure I want to go into Kai fuli's model of the US and China or the two superpowers when it comes to artificial intelligence or technology I think it will be a lot more interesting than that at least I hope it will be and I hope we can have more cross-pollination thank you good good question go ahead so what you're talking about are things like requiring data on citizens to be stored inside the country and Indonesia went down that path it this is one of the big things that were discussed in the TPP negotiations I think that edge computing will probably take a little of the pressure off of that the thing is the clouds not going to go away though and so if you're doing something like business analytics you really want to have as big a body of data to analyze as as possible and so I really kind of wonder if if it's the solution of the problem I don't think it's the solution of the problem but it might take some of the pressure off for a little bit thank you other questions comments so one of the features that we have in this series is that we always have refreshments at least I hope they're out there after the session and we want to encourage everybody to hang around and get to know each other and and yeah so thanks for coming today and look forward to seeing you again next week [Applause]

Channel: stanfordonline

Views: 18,237

Rating: 4.9453926 out of 5

Keywords: ee402a, Stanford Online, seminar, Richard Dasher, Edge Computing

Id: Hhobq4fs87w

Length: 61min 53sec (3713 seconds)

Published: Thu Oct 03 2019