Speaker 1: 0:22

Welcome to this new episode of Techzine Talks. On Tour, I'm at Mobile World Congress and I'm here with Parm Sandhu from NTT Data. You are the Group Vice President, enterprise Products and Services. I read here. So what exactly does that mean? What are you?

Speaker 2: 0:36

responsible for so I'm responsible Sandhu for all of our 5G and edge AI and edge computing product offering globally for NTT, and really that offering consists of several components, really, but all brought under one roof. So we're helping enterprises in key verticals like manufacturing, healthcare, smart cities, transportation, logistics, to basically help them digitize and transform their business operations, all the way from communications that they need, the right communication for the businesses they need, developing security solutions for them to help make a secure network edge computing infrastructure so they could run the next generation workloads, such as AI, as in XNOW, on their local premise. So really we're all very familiar with the cloud side of the business. Over the years we've got NTT, building data centers, ntt.

Speaker 1: 1:34

Well, and to be honest, NTT has changed over the years. A little bit right, so at first it was all different companies becoming entity limited and now it's actually the merging of entity data and entity limited, so now you're actually entity data as a whole. So I could imagine some of the listeners will say well, which entity are we talking about? But now it's all joined together, right?

Speaker 2: 1:53

Oh, I think you brought up a great point. So, entity data as you've seen, there's been a tremendous shift into the data center, into the cloud services, right that hardware. Now what we've done is we've also looked at the edge at the factory plant, at the plant on site, and the operational technology and bringing that ITOT convergence together. So my portfolio is really focused on growth services around this space and that's what we're doing. So how do we generate new revenue? How do we solve new problems that clients are having? How do we help clients take advantage of this tremendous innovation, the era?

Speaker 1: 2:27

of innovation that's coming. What are some of those problems or challenges that your customers or potential customers have?

Speaker 2: 2:33

at the moment? Yeah, so I'll give you an example. So enterprises are really looking to drive operational efficiency right. They want to improve their client experience and also make drive worker productivity, worker satisfaction. So how do we do that and what are those solutions that they're looking at? How do we bring the technology innovation underneath? So one of the areas that we started looking at was just providing good connectivity inside the factory.

Speaker 1: 2:59

And that's not as easy as it sounds. Right, it's not easy, if you look at the.

Speaker 2: 3:02

Yeah, if you look at millions and millions of factories out there, some are very old right and some are obviously new. Not everybody has the luxury of a brand new.

Speaker 1: 3:10

And a lot of those areas also have very sort of ancient ancient maybe is not the right word old legacy connectivity systems. You have the old SCADA systems. You have all that stuff that's going on. They all need to be connected as well and it's not easy because OT has a lot of different protocols and lots of stuff going on there. So I can imagine that's going to be one of the big issues that we have.

Speaker 2: 3:32

So I think IT has had tremendous investment to bring IT modernizing and the cloud has helped tremendously in that. So our view is edge is another area. Edge and cloud to edge, continuous transformation is critical. There's a ton of data that's sitting isolated inside these factories, like you said. Some of them are SCADA systems, some of them are PLCs. That you know. We introduced our Edge AI platform, which is a very ultra light, low compute, low, small scale platform.

Speaker 1: 4:00

That's an interesting name because those are two sort of hype buzzwords Edge and AI. Edge has been a bit on and off over the past couple of years. It used to be very, very hip and happening, and then it was a bit less so, and now again, especially with AI, it's coming up again.

Speaker 2: 4:15

Yeah so AI has become a tremendous use case for connectivity, for edge computing. But now it's a matter of bringing the right solution, the right size compute. I can't bring infrastructure that's typically designed for a data center into a factory.

Speaker 1: 4:33

So completely. Wi-fi may not necessarily work in a factory Exactly.

Speaker 2: 4:40

it may not work as well. It may not be continuous, may not support mobility use cases. It may also not have the capacity, because typically all this WiFi network's been around for a while and it's got a lot of other services just to run in a factory. So can we bring that low latency, guaranteed performance service, which is what 5G brings Edge computing? Now we need to make sure that the right size hardware is there for that application.

Speaker 1: 5:05

Are we at the point in factories and in the maker kind of industries? Are we at the point yet where private 5G is also commercially an interesting proposition? Because I remember a couple of years ago I think, we met first three years ago. We talked about private 5G back then as well, and back then it was still quite expensive, because it's not a cheap solution, right 5G, but how do you do the ROI?

Speaker 2: 5:30

Yeah, so I think you bring up a really good point. So I think where things were three years ago, ecosystem was still coming in, it was very telco-centric, it wasn't really. It was basically in the evolution stage. The number of devices with 5G was still limited. The types of devices that you had you had mostly smartphones, which was mostly consumer, but tablets and scanners and handheld devices, you know, was just not there. So now, completely, games change. 5g is now predominant in all devices. Price points have come down on the hardware side. The ease of deployment and installation on these networks has become very simple. We can get a 5G network running in a matter of weeks, whereas before some of the vendors that were taking telco technology which they still are today and trying to deploy it taking a very, very long time. The investment and cost is massive. In our case we can build a network in less six to eight weeks. As long as the equipment is installed, within two weeks we can have it up and running.

Speaker 1: 6:31

That lowers that whole cost. And what role does the rest of NTT play in this? Because obviously maybe primarily in Japan, you also have a telco business over there, Does that?

Speaker 2: 6:42

help in these things. Absolutely, oh, absolutely In these things, absolutely so. Ntt, for example. I'll give you an example. In Japan, ntt East and West have been deploying private networks for many, many years.

Speaker 1: 6:54

That's a misunderstanding a lot of people have now, because private LTE was already there. I mean, maybe not necessarily very successful all over the world, but it was there.

Speaker 2: 7:02

Oh, japan was leading the way and what they did. We have a whole lab set up. It's the size of a you know a small city and we have, like autonomous vehicles, level four drivers, self-driving buses. We have industrial factory, complete factory set up where we can do, you know, robot, robots that are moving from factory lines, self-driving, moving parts from one location to another, and that's that. R&d investment allows us to bring use cases right to our customers who are deploying 5G.

Speaker 1: 7:33

And because you already had the scale and before, you can also do it at a reasonable price point. I would say right.

Speaker 2: 7:40

An entity spends over almost four billion annually on R&D. A lot of that R&D goes into our IOWN network, which is our Integrated Optical Wireless Networking, and that means building use cases and applications. And that's really what the companies need to drive that ROI on the network. And that's why I think we're unique is that we're focused on not just building a network. Everybody can build. Look, there's enough engineers in the world. They can all build a network. That's not the issue. The issue is around how do we help those companies monetize that network, bring the right devices.

Speaker 2: 8:13

If you go over here, ntt Kunoku in our booth, here at the Docomo booth, they built complete custom AR glasses, working in partnership with Sharp. Kunoku is a NTT data-owned company NTT, sorry, docomo-owned company. So what we have done we've built the AR glasses and now we're building applications around it. We have both consumer version and we have industrial version. And imagine that in an AR VR setting somebody who's building complex machines, how do I train the worker when there's a maintenance task or how do I give them the next steps in the assembly that need to happen? These are very, very comfortable, lightweight air glasses, high quality lenses, built by Sharp and with NTT together and we put that together as a complete solution. So I encourage you to get a chance to look at it.

Speaker 1: 9:05

But it's an amazing, amazing, and it's lightweight, lightweight yeah let's return to the edge. Ai kind of yeah, and I think that we talked about right. So we talked about the well, the, the rise of AI nowadays, obviously, and the resurgence, I would, I would call it a full badge. Again, what makes it? What makes that a hard problem to solve?

Speaker 2: 9:30

for companies. So the combination of edge and AI. So if you think about AI today, right, you're running AI in the cloud, which is the big focus and that's where you're going to see the early successes. A lot of it's lots of hardware processing compute, so a lot of companies making money off of that today. But bringing that down to how do I make that economical? How do I run it inside of a factory where I have constraints on power consumption, I have constraints on capacity and performance, I have constraints on the backhaul network to the cloud? Most of these factories don't have 100 gig network links from the factory. To this cost, they have to run the machines in factories. That's where they need to put the money and connectivity needs to be there. So, putting all of those conditions and limitations, together that's quite a hard brownfield environment that you're dealing with.

Speaker 1: 10:18

You got it you got it.

Speaker 2: 10:19

But there's another issue is the data. The data is all over the place. It's in sensors. You might have ABB sensors, you might have Siemens sensors, you have hundreds of different. A factory contains thousands and thousands of pieces of equipment. Well, they're all talking different protocols. They're all talking different languages. They were never meant to work in an AI environment. Ai, you need to take all of that data. Take all that data, contextualize it time, synchronize it and then feed it into a machine.

Speaker 1: 10:49

So you see we solve that problem, you take on touch because then then it actually means that you that needs to be an intermediate step. That's exactly what. You can't take all day, all the 1500 protocol that you have in your, in your environment, and just feed it into somewhere yeah, you have to. Somebody has to do it into it, into one MQTT or whatever. It is a kind of protocol.

Speaker 2: 11:08

Yeah, exactly so we started with that fundamental problem. We created a very small box called our Edge AI and we just stick that into a machine right away. As soon as you plug it into a network on a factory, it'll start learning, okay, what protocols you have, what PLC controllers you have, and learns from it which manufacturer.

Speaker 1: 11:26

And that part in itself isn't new. I've been to many of industrial events, industry focused events where we have lots of I've seen lots of servers and small kind of things that can collect lots of data and can show you how everything is working and if something's going to break or all that stuff. So the harvesting of the data that's not the problem.

Speaker 2: 11:49

No, that's not, but it is still a problem. So now we've solved that problem in a way, but on a machine that's completely remotely monitored and remotely managed, has to be done in a very low cost way, has to be done in a sustainable and operationally effective and cost efficient way, and has to offer the SLAs that they need. So then, on top of that, we built predefined edge AI models. So smaller models, you know less than 10 billion parameter type models, which can then self-learn, self-train, yeah, and that's possible because you're working with a very defined use case right.

Speaker 2: 12:24

Use case or defined constraints right. We can say you know, we can pick a small production line of a factory. There might be 20 machines in that factory. We can look at monitoring of operational productivity of each of those machines Any time. We sense quality reduction in the part, maybe the stamping machine has to be, the dye is worn out. We're starting to see defects coming in. How does that affect downstream? If one machine has a problem, maybe it slows down the production. I can start using that data to plan my workforce better, automate all of that task. Worker safety can automatically, before you had to train computer vision models, draw lines on the floor. If a person stepped over the line, the computer recognized.

Speaker 1: 13:04

I saw a demo three years ago. It was actually a private 5G demo by some other company that exactly, did this right. So continuous video stream via private 5G and then decide whether this thing that has been built is according to spec or not. Right yeah?

Speaker 2: 13:19

exactly. So there's the quality, but I'm telling you, what's changing now with AI is that you don't need to build predefined trained model. I mean, the model itself learns. It automatically changes. If any condition changes on that floor, the supply chain, anything changes. All of a sudden, ai says hang on, I'm seeing a lower reduction in volume of this machine. Okay, I need to look at downstream what to do. I can look at workers all working together in a collective with forklifts driving around. I don't need to draw lines on the floor anymore. The AI is able to interpret the visual data, the video it's seeing, and see what it's happening and make decisions and make responses and acts, and then obviously the next step is going to be automation To do something with this insight, With that data, exactly, and that's what NTT does.

Speaker 2: 14:08

I think the unique thing that I always say to NTT and a lot of our customers say is we don't just do one part, we don't just do the network, we don't just do edge computing, we do it all under our house. And the benefit of that is if you want, as an organization, want to deliver a program effectively and fast, low cost and reliably on time, you need one party who can bring it all together. Otherwise, if you're trying to stitch together four or five different vendors and partners, you're going to fail.

Speaker 1: 14:37

But it still means that you have an extra or new control plane from entity data in this instance. Right, so you do add some complexity to your. Maybe you also remove complexity because you're only with one vendor. But you also add complexity with the extra layer right.

Speaker 2: 14:54

Well, what we're doing is we're really focusing on open standards. That's the one thing we really stick to with NTT is that we want to make solutions that are not locking in our vendors okay, our customers because our customers fear lock-in right and, especially in the post-COVID world, everybody wants to have multi-vendor solutions. So we bring multi-vendor solutions. We give them two choices on equipment and infrastructure, multiple choices if they want, and we make sure that anything we're doing is all standards-based.

Speaker 1: 15:23

So it doesn't really matter what you have underneath.

Speaker 2: 15:25

It doesn't matter what I have underneath, so that gives our clients the faith that hey, okay, you know what NTT, maybe we have more flexibility. We want to be able to swap vendors. We want to be able to swap SIs. You make it sound quite easy, but it probably wasn't to actually build this. No, because it's built on years and years of you know we talk about. I think you mentioned the concept around AI-driven networks as an example as well. You know we're starting to introduce AI into our operations environment. We've been doing that for a long time. We can predict outages of potential networks well before the customer even sees it and we remedy those by just analyzing all the logs and analyzing that data and creating signature of events that are going to happen.

Speaker 1: 16:08

Right, and that also actually gives you opportunities in the security space as well. Yeah, right, because OT security is a bit of a thing, yeah, and I say that with sort of a sense of understatement, because it is a big deal. It is extremely hard to secure. But now that if you have everything centrally available, all the telemetry, all the behaviors, all the anomaly detection, maybe even you can actually do lots of stuff around security you bring up a great point.

Speaker 2: 16:40

Everybody's been talking about ITOT convergence for many years and it hasn't happened because it's been so complex to do. We just announced this last week our partnership with Palo Alto where we took a 5G network technology, which was you know? We know that 80% of talking to CIOs, 80% of the network investments, are driven to improve security, cybersecurity.

Speaker 1: 17:02

They want to ensure that their facilities, their operational factories, are cyber secure and to be fair, the network is a great place to start security, exactly so if you're going to add sensors and these devices on the end of them, you're exposing potentially your factories and your plants.

Speaker 2: 17:20

right, because there are companies who will disable all the communication in devices and devices and sensors, except with the wired. You know, and they think you know we will not provide security. We, we don't have, it's not secure unless it's all turned off, and that's intentional, right, they're worried about it. So with 5g, and what we've done with palo alto and their xor and cortex platform is we've integrated that right into our 5g network where we can detect, from an ot security point of view, the behavior of each device, signature of each device, what it should be doing, if any time it changes. We can feed that data into an AI and detect there's a problem and compartmentalize that. And we've also designed our 5G network to fit the OT model which, like you said, it is something they use a Purdue model type architecture and they segment the traffic and we do that segmentation innately into our 5G solution, and so the integration with Palo Alto does that also, so that works two ways.

Speaker 1: 18:20

So you feed your data into their platform, yes, but can you also tell that platform?

Speaker 2: 18:26

to do something. Yeah, yeah, yeah. So then, what we've done is we work with our 5G vendor and we've exposed APIs between the 5G core so that platform then learns from what's happening with the behavior of those devices and it can then tell the network what to do with that device Maybe restrict the traffic, maybe just say, hey, we wanted to. Or we can shut it down altogether, because sometimes you don't want to shut down, because you want to analyze what's going on, but you want to reduce the risk to the rest of the factory or plant. We can say, hey, this device now can only speak to this one location. So we can control all of that. And that's the power of what we're doing here.

Speaker 1: 19:05

It all sounds very good and very well thought out. You didn't rush into this, but what about the adoption? Do you see? Can you tell us anything about that?

Speaker 2: 19:15

Yeah, we're seeing amazing growth in our 5G deployments. Now We've got clients all over the world. There's still some challenges, like Spectrum globally it's not available everywhere, it's not at the faint of heart. Okay, let me just tell you, building 5G network business is difficult because there's been a lot of technology challenges, the ecosystem that is solved, I think now the next big thing. And then there was Spectrum. We're getting more and more countries coming in. But, more important, we're building up partnerships with governments, regulators. But we're also building partnerships with M&Os. We're saying, look, we see this 5G enterprise but we're not really able to meet the needs of that enterprise, so let's partner with SI.

Speaker 1: 20:00

So the mood has shifted from we don't trust this we want to do it ourselves Because the implication was always, or the assumption was always, that private 5G especially was very niche, and why should we focus on this? But apparently the other parts of the equation have also seen the value of it now and they are actually willing to go for it.

Speaker 2: 20:23

They're willing to go for it, we're willing to. And look, no carrier wants to say my only value is spectrum, right, but they bring a lot of technology expertise together. They bring a lot of local business relationships. So we want to make sure it's a win-win for both of us, right, both for us as an SI and a global SI and a managed network service provider, but also for our partners. So I think we've done enough of these that we understand the business value, and that's really helped us to do that.

Speaker 1: 20:53

So just finally, what's the next step? Right For the.

Speaker 2: 20:58

You know, I think the growth, I mean I kind of look at it as we definitely want to continue to grow, expand the country coverage globally. Okay, that's number one. And now next is so that's more countries, more deployment.

Speaker 1: 21:11

But that's something that you have to also do with governments, right, so you have to lobby and Well governments and carriers and operators.

Speaker 2: 21:16

We're continuing to do that. You know we've had some really good success. Now we want to with large clients that are multinationals. How do we roll it out across them? All over the place, right?

Speaker 1: 21:27

That's number one. But that is actually quite hard to do, right, because what always struck me about the telco space is that, especially inside Europe, for example, there are all sorts of rules and regulations. What can happen inside of countries, right, right, and they all apply to all the countries individually Right, but you still have this thing where you cross a border, you will still drop your connectivity for about 15 seconds. Yeah, and so getting actually a completely covered, fully covering, spanning network, that's not as easy as it seems.

Speaker 2: 21:57

Well, you know, I'll give you an example. So the industry is starting to think right, so coverage, there's always a balance. Right, there's a coverage, there's a cost balance, there's coverage, and then there's quality, is it? You know, in some cases best effort is good enough, but in some cases you need seamless connectivity. So I think the next big thing we're starting to see is like, how do all of these multimodal communication technologies come together? So NTT has a, you know, our transit L business is our global MB&O, and we're looking at how do we make that seamless connectivity between, you know, between private networks, public network, terrestrial networks and non terrestrial networks.

Speaker 1: 22:37

how does that whole ecosystem for example, if you have a train crossing borders with something with containers on it, with a 5G sensor or whatever, or a ship, so you need NTN networks, right, so we're putting it all together.

Speaker 2: 22:52

Actually had a very interesting industry roundtable with a whole cross industry of aircraft manufacturers to automotive manufacturers to, especially in countries like Europe where there's small countries where you might cross two or three of them. So how do you do that? So this is why NTT, with our edge, private 5G edge, ai, iot solutions and our global macro coverage, come put it all together. I think you know, as we get more and more use cases are coming in that we're noticing and that sort of seamless connectivity is really happening. I think we're about to launch on that that journey.

Speaker 1: 23:31

Well, that's actually quite good to hear. Yeah, because I remember the early days of 5g without the private in front of it. Yeah, and we actually talked about 5g for about 10 years before, and now actually private 5G has a faster adoption cycle.

Speaker 2: 23:48

So that's good to hear. All right, all right, so yeah.

Speaker 1: 23:52

Thanks a lot for joining. I thought it was an interesting conversation.

Speaker 2: 23:55

Well, I hope to see you again in due course. Thank you for having me and having NTT Data part of your podcast. All right, thank you.