 Thank you. So thank you all for coming here to listen to me My goal over the next 20 minutes is going to be to make sure this is not the least productive session for you at this amazing conference So with the bar set pretty low for myself. Let me just tell you a little bit about My background I think rich just in a little bit of that But I may say essentially I'm an early-stage investor at a venture capital firm called lightspeed venture partners So we are located in Menlo Park, California with offices in Israel India and China So I actually started my career as an engineer and an entrepreneur when I was still in school at Georgia Tech The company was a spectacular Flame-out complete failure, but I learned a ton about what not to do which I think qualifies me as a venture investor quite a bit I did start a second company after that when I was in business school. That's going much better So I did learn a few lessons from my earlier gig But then fast-forward a few years I joined Anderson Horowitz as a partner where you know, you heard some of the companies where I was involved in and then I moved on About 18 months back as an investing partner at lightspeed So in the last 18 months or so I have invested in companies like serverless.com stream Leo And then a security company called XA beam and another seed company we have in network management space Now few words about lightspeed very quickly the firm started about 17 years back and We managed close to about five billion dollars today Our most recent fund raised last year is a billion two fund that we are just in the early stages of deploying some of our recent Some of our portfolio companies that have been in the news recently are companies like App Dynamics We were cities investors new techniques cities a investor Snapchat we were seed investors mule soft cities be investors So we actually have at least we claim to have deep expertise and experience in building Enterprise and consumer technology companies. So if any of you in the audience is thinking of Starting one do look me up offline, please Now I was asked to come here and share with you Our perspective as an investor and my perspective as an investor in investing in the IOD space So for the time we have today I'm going to talk about the areas within the IOT ecosystem where we are spending time as investors Trying to identify, you know, the businesses and the teams we should partner with The only thing I would urge To you is that take this as one person's viewpoint So and nothing more so if you have if you have other areas that we we are not looking at and we should be You know, I'd love to have that dialogue, please Now as an investor whenever we are looking at a space to find opportunities to invest in we first try to Figure out if there are macro trends or driving forces that we should pay attention to So we are not super tied to any particular thesis because then you're you tend to miss the outliers But what we do we tend to do is we keep this blueprint in our mind as we are filtering through business plans and Teams and companies in our day job So in the IOD space here are the trends that we feel that we are seeing and I would quickly walk you through To this particular few ones I have there So first is the number of endpoints are increasing exponentially So I'm going to spare you the eye chart of how the number of devices are moving up and to the right But I think you know, we all can agree that it's really huge And then, you know, if you talk to people like Gartner, they estimate that by 2020 We should see 20 billion devices Cisco says it's going to be 50 billion I don't know who is right, but I know that it's going to be pretty big The other thing is if you really look at consumer and enterprise applications of IOT It's probably split 60 and 40 60% in consumer 40% in enterprise Now because of the increasing number of devices what what is naturally happening is the volume of data that these devices are collecting and sending back is also growing exponentially and This is even getting worse because for every use case that you can imagine we are collecting more data and Hopefully reach our data as well The other thing that we are seeing happening quite a bit And I'll talk about in some of the projects that we are tracking later on is that the IOT devices the endpoints are actually getting Smarter and they're of course getting cheaper But they're actually getting smarter to the point that you can now run a stripped-down version of Linux or Android on The endpoint and actually can run some lighted applications The other thing that's happening here is not only are they getting smarter But the price points are falling to the level and I would argue I would argue that it has already happening Where a lot of use cases are actually finally making sense, which wasn't the case two three years back And the last point I'm going to make on the trend is the use cases are evolving quite a bit So well the majority of the use cases even today tend to be data collection for visibility Control alerting things of that nature. We have started to see See use cases where instead of just alerting a human being the IOT System is supposed to take an action automatically Now this evolving use case is having a profound impact on the infrastructure stack the infrastructure IOT stack We're going to touch on some of those things But as a result of that real-time Processing streaming technologies even driven computing are becoming really important when you think about investing in IOT and IOT vendors are already exploiting some of those to create to skater to some of those use cases Now I'm going to put up a very simple framework in terms of how we think about the IOT ecosystem So we tend to think about the technologies and the businesses we see in the IOT in four buckets So there is the build bucket which is where we group all the technologies that go behind Creating the device building the device or the hardware the actual hardware Then we have the manage bucket where we put all the Technologies that are required for connecting the networking then the managing the controlling Securing all of that or IOT devices then we have the infrastructure bucket at the bottom Which is where we put all the infrastructure technologies you need to cater to the emerging use cases that I've been talking about I was talking about just recently and then finally you have the DEF category where we tend to think about all the developer resources that you need The paths the developer tools the open source projects You need to kind of cater to the developers who are going to build out IOT applications So let's talk briefly about the cat technology in the build ecosystem so just to be clear, I'm actually talking about real hardware the stuff that goes into the devices and You know we investors tend to be fairly allergic to anything that smells or feels like hardware a lot of scar tissues We have when our competing with large hardware manufacturers But the question is like why am I even talking about that? Well, you know as the devices are getting smarter guess what they need fancier brands IE in a better processors and custom processors and not only they have to have custom functionality they need to be cheap they need to be part efficient and You know when it comes to building custom devices that have embedded processors today arm is probably the only game in town Probably and there are some other solutions as well But given that you are trying to build devices where the price points are in low to double-digit dollars Arm is not the probably a great misfit there So, you know, there are some interesting projects out there and one of them I'll call out is the risk five project out of UC Berkeley, which is the open instruction set processor architecture and The way in a in our naive way We kind of almost think about it as a sort of an open-source version of arm where you could actually use that technology To go and build custom processors in a self-serve way where you can meet the price point You need to meet to get to the market now As we move, you know once we have the devices the next task is to hook them up and connect them to a centralized system running in the cloud could be public could be private and What I would argue is that's at this point It feels like it's a relatively more mature Ecosystem as there are a number of companies already targeting this particular aspect of the ecosystem and I use the word Relatively because it's still far from being solved. It's just that there are more companies and teams out there some are already shipping end-to-end systems consisting of Sensors that talk to a gateway over BLE or Zigbee the gateway talks to a cloud Backend over Wi-Fi or cellular and the use cases they are handling is mostly around visibility analytics alerting running on the cloud and then some of the other approaches that we have seen is Companies taking their software is mostly SDKs trying to partner with device manufacturers The device manager manufacturers use those SDKs to embed in their devices and the startups build the rest of the back end Which is the gateway and the cloud back end as well Now in our view We tend to think that you know as devices get smarter and they come built with network connections and as Some of the standardization happens on the connection side the protocol side The opportunity for some of the startups might shrink in the long run just just in our view The opportunity the part of the ecosystem where we think there's a lot of opportunity And we are very bullish about is once you have this device is connected. How do you manage them? Now managing is a loaded term. So, you know it for us it covers a bunch of things first is monitoring So once you have the IOT solution in place, how do you know what's going wrong and what is going wrong? A device might be offline because it has run off or run out of power or it might be sending data in the wrong format Now a corollary to monitoring is debugging So as devices get smarter you are running applications on the endpoint So there's a need now to debug when an application is not running properly on the endpoint Now the question is like, you know has this problem not been solved already in the IT world. Yes But the scale at which this problem has been solved for the IT world is very different from the scale that you are going to see in the IOT world which is like several orders of magnitude more Devices and data volume. So we definitely think there are a lot of opportunities there the other Area that we are spending a lot of time in this particular Part of the ecosystem is security. So if you look at traditional IT security approaches They have been designed to secure hundreds of servers Maybe low thousands of endpoints in a mobile phones tablets and laptops at best but The other point about the traditional IT techniques is the servers and the endpoints have powerful compute available So you can run agents on the endpoints to gather data to detect a security breach and once you detect it You can actually apply some kind of policy or enforcement policy to stop that breach None of that exists in an IOT world where your endpoints do not have such sophisticated compute It's getting better But it's far from there where you can run agents to collect and apply security policies. So the approaches we are seeing in securing IOT devices has a lot to do with machine learning and kind of profiling the devices from Observing how they operate in a in a in an ideal condition and then trying to come up with the policies automatically and then trying to Secure those devices using those policies at the network level. It's a very different It's a wholly different class of technique than that we have seen in the past in the IT world Now the final task within the management layer where we think there are still in a lot of opportunities is the patch delivery So, you know this patches could be far more upgrades could be ways of grades could be application in a different versions of the applications But you need to deliver those to the endpoints over a network connection that could be patchy at times So how do you manage all of that? How do you build a stack where you are not? dependent on the network as much or at least you build some fault tolerance in that and we are seeing some companies that are Trying to solve that part of the problem now the bottom line that I'm trying to make here is Techniques the problems exist today in the IT world But they are not going to translate very easily to the IoT world because of two things one is device count and the other is data volume So as a result, we are seeing a lot of very interesting ideas and teams going after this particular part of the ecosystem now I'd have fundamental At a fundamental level I strongly believe There are a lot of opportunities in the infrastructure IT stack that you need to build Below all this monitoring or securing part of the ecosystem and you know, that's the part where that's a part of the ecosystem Where we are probably spending the most time and we are probably most bullish about now We strongly believe that the infrastructure we have today is not going to sustain in a world Where you have billions of devices sending petabytes of data? So I want to touch on a few different aspects or two few different needs of this infrastructure and in a way We are spending time. So first is real time So as more and more devices get connected the transmit a lot of data that need to be managed and analyzed continuously across large data sets, so Now these specific capabilities are new and I would argue that you know They are very specific to the IOT use cases that are emerging today with all the autonomous cars and the in a OT Kind of use cases that we are seeing so we are seeing in a lot of smart entrepreneurs building the next generation of data processing Technologies that is moving away from batch processing getting into real time and they're like a bunch of in a open-source Projects out there. We actually have invested behind a couple of those that are trying to go after that particular Problem and trying to solve it with real-time data processing systems The second is event-driven infrastructure. So in the traditional world We are used to standing up servers physical or virtual We're used to standing up those servers that cater to clients the clients can be mobile mobile app or a web app and You know the server you run the server pretty much non-stop in a trying to cater to the client request In an IOT world where you have billions of devices and not most of them are sending probably sparse data You don't want to stand this kind of server infrastructure You normally would to cater to that kind of use cases because the cost is just unsustainable You cannot do that So people are turning the problem on its head and they are trying going after the event-driven paradigm where you only come up with The server side application when there is a request from the device So we actually invested in a company called serverless.com which is based on this serverless or event-driven paradigm Where you only stand up a serverless serverless server side application when there is a request from the device side The final piece here in terms of what you need from this next generation infrastructure is edge computing So as both in a device count and data volume increase It's kind of imperative that the data processing Needs to move away from a central model to a more distributed model So more needs to happen on the edge because you don't want to send all the data back to the central Whatever you are running your applications. So, you know, it saves time and money It's better efficiency and it prevents the need to overwhelm your databases that are running centrally now you know as Devices IOT devices are getting smarter the point I was making earlier Beyond that in a beyond not overwhelming your central data processing system The other need is you can actually run applications and not just analytics on the edge So it might sound a bit like science fiction today, but we strongly believe that in a few years out your Data processing system will probably be not confined within a few servers in a light lights our data center more More importantly, it will be the distributed systems will be truly distributed some processing happening on the edge Some in the gateway some maybe in a data center in a closet and then some That absolutely needs to happen in the cloud will go to the cloud, but What now you know, you might imagine if that happens you that throws up a whole set of challenges How do you monitor? How do you deploy? How do you debug? How do you secure this truly distributed systems? So you're spending a lot of time over there trying to Sort through some really interesting companies. We have seen lately the final point. I would make here is Gittering to the developers. So no matter what you do what you build Eventually you need to get developers adopt that framework so that they can build applications that can eventually cater to the use cases So open source is a big part of our focus as an as an investment firm Obviously Apache has done a great job there and then we are looking at a lot of deaf tools A lot of past kind of solutions that help developers build applications that go on and run on this new Infrastructure and the new IoT systems that that are coming to the market now no Talk on IoT can end without a caution in note So you know as investors we have been looking at the IoT space for I would say four or five years now and every year It feels like the market is going to take off this year, right? It hasn't happened yet So the reason question is why why why has the market under from underperformed over the years? The first problem is monetization. So we can talk about all the interesting infrastructure needs and all the different kinds of monitoring and securing solutions for IoT the problem is it's very difficult to charge a Respectable amount of money to secure or monitor an endpoint which only costs a few dollars or even less And we didn't have that problem on the traditional IT side because the servers and the mobile phones and tablets They cost several orders more magnitude more in the IoT world that business model needs to evolve. We are seeing some Initial work on that, but it's far from to a point where we can say well, you know This is the business model that's going to create the next billion dollar company in this space That also brings me to the next point which is fragmented use cases. So if you look at the most of the IoT use cases and specifically on the enterprise side, which is where I spend more of my time Use cases are fairly fragmented. So some of the bigger use cases like temperature monitoring things like that are probably a few hundred million dollar market And most other use cases are smaller than that. So the question or the challenge is how do you build an IoT? Not only from a technology standpoint I would system from a technology standpoint We're also from a business model standpoint that can cut across multiple use cases so that you can capture a much wider market across all of those and then You know the other point I would make here is interoperability so I was actually looking at a McKinsey study sometime back and you know some of they mentioned that 40% of value from an IoT solution could be lost if we do not take care of interoperability And you know we might divide about the exact number But I truly believe that is the case that is going to be the case because unless we have standards and open protocols Lot of the solutions will be custom solutions and a lot of the and a lot of the startups will struggle to break into this Ecosystem so that needs to happen so in summary I would say in a summer and I would kind of Categories our invest our sentiment in this market as cautiously optimistic And there are two things that's happening in this market, which makes us more optimistic than before So first is we are finally finally seeing in the emergence of real use cases with real ROI So it's happening on the consumer side with autonomous cars and smart homes It's definitely happening probably happening more on the enterprise side with OT kind of use cases as well as in a smart energy And service management kind of use cases so that use cases are finally emerging where somebody is saying okay I have a budget if you can improve my efficiency or if you can you know help me diagnose my in a Failures much better than I could do before the second part second thing that's happening is the underlying Technology stack has matured a lot in the last five seven years because of all the stuff that's happening on the big data side So techniques such as machine learning streaming and all the big data processing We have seen are finally mature enough where they can cater to those use cases So the convergence of those two trends make us very optimistic than we were probably two three years back Still very early in an awfully large market. So at light speed we continue to be super bullish about this space Spending a lot of time here So any of you planning to build a billion dollar company in this space do look me up. Thank you Thank you for listening and for your time