MWC25 | Sean Lawrence, NTT

Clips
More from MWC25

Sean Lawrence

Vice President, Head of IOWN Development Office

NTT

play_circle_outline Introduction to NTT and its global IT services outside Japan

play_circle_outline IOWN concept: transitioning from electronic to optical data transfer

play_circle_outline Collaboration with IOWN Global Forum for industry shift to optical data transmission

play_circle_outline The Crucial Role of All-Photonics Networks in Accelerating Computing Growth and Supporting AI Boom

play_circle_outline Optical Technology Experiments Enhancing Data Center Redundancy and Resilience Through Low-Latency Database Replication

Info
Transcript

Sean Lawrence, NTT

Sean Lawrence

Vice President, Head of IOWN Development Office NTT

Sean Lawrence, vice president and head of the Innovative Optical and Wireless Network (IOWN) Development Office at NTT, joins theCUBE at MWC 2025 in Barcelona. In this engaging discussion with Dave Vellante, Lawrence elaborates on the evolution of NTT from a leading Japanese telecom to a global IT service powerhouse. He shares insights into IOWN's ambitious vision of replacing electronic data transfer with optical solutions to address power consumption and performance limits. Analysts from theCUBE Research contribute to the dialogue, underscoring the impact o... Read more

explore Keep Exploring

What services does NTT offer globally and how has the company diversified over time? add

What are the advantages of switching from electronic data transfer to optical data transfer in computer technology? add

What is the purpose of the IOWN Global Forum and who are its founding members? add

What is the importance of connectivity and the network in supporting the AI boom? add

What experiments were conducted by the IOWN Global Forum members regarding redundancy and backup, specifically involving the movement of a running VM from one data center to another data center with less than one second of downtime? add

bolt Powered by CUBE AI

Sean Lawrence, NTT

search

Dave Vellante

>> Hi everybody, welcome back to Barcelona. We're here at the Fira, it's an amazing venue, there's seven halls, eight halls, a hundred thousand people. My name is Dave Vellante. I'm here with Bob Laliberte. Savannah Peterson is nursing her voice, but she's going to be back in action any minute now. You're watching theCUBE's continuous coverage. This is day four of MWC 2024. This is our fourth year in a row here. We started in 2021 in Cloud City. We've been at this new location ever since. We're in Congress Square, come by and see us. Stand CS-60. Really excited to have Sean Lawrence here. He's the vice president and head of IOWN Development Office at NTT. We're going to talk optical. But before we do, first of all, Sean, welcome. Thanks for making some time this morning.

>> Thank you very much, Dave. Really a pleasure to be here.

Dave Vellante

>> Yeah, and all the way from Tokyo, which is pretty awesome,

Sean Lawrence

>> where the headquarters are of NTT. Let's start there. Tell us about NTT. We've had NTT Data on, so we kind of started bottoms up or middle in. Tell us about NTT.

>> Okay, so NTT, it's our Japan-based company.

Sean Lawrence

>> Pretty big, about 330,000 people globally, and almost half of those people are outside Japan now. Our roots, legacy is we were the incumbent PST and telco in Japan. That's where we started. But we've diversified a lot. And now in Japan where we are still the incumbent telco, Docomo was one of our subsidiaries. Outside Japan NTT Data is our biggest go-to-market. They're in Japan and overseas. But now we offer pretty full- stack services everywhere from subsea cables, data centers, network integration, system integration, application development and management, up to consulting. So pretty much the full-stack of the IT services that we're offering outside Japan.

>> So enabling critical infrastructure, essentially,

Dave Vellante

>> core operations. I mean, some serious stuff that you guys are involved in.

Sean Lawrence

>> Yeah, yeah. We've done a lot of acquisitions, and actually NTT Data, we've consolidated all now under the NTT Data brand. But yeah, so it's a pretty exciting place. And my function, I work at the NTT, the headquarters, sitting between our core R&D. So our core R&D function sits in the headquarters doing a lot of fundamental research, and I sit between that and the subsidiaries, so trying to bring information from the markets back to R&D, help drive the agendas and the priorities, and then take what we're making in R&D, how can we take it to market outside Japan. So that's the function I sit in, and particularly I'm in the IOWN area.

Dave Vellante

>> So explain IOWN.

>> So IOWN is a concept, it was born out of our R&D labs

Sean Lawrence

>> where we do a lot of fundamental optical research, and we have a long history in optics as an incumbent telco. But also we were looking around in the world around us and see these trends in the IT industry of, first, power consumption is just increasing dramatically, dramatically, dramatically. AI is just compounding that problem. But also performance limits being hit where the computer's running into the limits of what it can handle with electronic wiring. So the IOWN concept is basically to switch from electronic data transfer to optical data transfer, not just across the WAN where we've traditionally been doing that, but deep down into the computer, into the circuit boards, all the way into the silicon packages themselves, switch the data transfer to optical instead of electronic. And optical has some pretty incredible properties. I only came to optical networking about a year ago. I was always from the packet switch world in my previous roles. But optics, you can transfer data, obviously very high speed, but also very low power consumption and long distances. So compared to electronics, it's just a better way to get data from here to there. It's just a bit harder to do it, especially at the small scales.

Dave Vellante

>> And electronic is copper? - Electronic is copper, yeah.

Dave Vellante

>> And I don't know, you may compete with Broadcom,

Sean Lawrence

>> I don't know, they might be a partner. I have no idea. But when I hear them talk about this, they talk about the short distances, they actually want to use copper because of the economics. Long distances, they have no choice. Is that... Am I getting that right? Is that the case or are the economics flipping?

>> We're doing a lot of R&D on that short...

Sean Lawrence

>> Making it shorter, more economical, lower power consumption. We believe that ultimately it's the better way to increase performance, reduce power consumption and heat. So yeah, I mean the economics are part of that and we're doing a lot of research and development to try and make that work at a small scale.

>> So at scale down the road potentially optical will be the

Dave Vellante

>> more economical offering for both short and long distances. I remember the switch from emitter couple logic to CMOS. At first CMOS was more expensive, but now, I mean of course you're using CMOS.

>> But actually we're working with Broadcom as well.

Sean Lawrence

>> still in some of the stuff we're doing we're using Broadcom switching capabilities, electronic, but getting the data to the switch using optics.

Dave Vellante

>> Okay, so that's where you add the value. Go ahead.

Bob Laliberte

>> Think about it from when you first got fiber to your home versus using the copper and the improved performance, the better latency, all that kind of stuff. And what they're trying to do is instead of that transition of fiber to the home, they're trying to do fiber into the data center. So as performance, especially when you think about GPU clusters, things like that that are really intensive for throughput, latency-sensitive, et cetera, and also especially with those environments, the power. So not having to power the copper and having to go over the optical networks is going to be a huge change. Like you said, the price will come down with scale, but moving to optical, that's always been... Even when you talk about... We're here at Mobile World Congress, so you've got 5G. Before 5G went into effect, there was lots of fiber laid and lots of optical networks built to handle the backhaul of what the additional capacity would be, so they're starting to take a look at it from how do we drive this down into the data center, down closer to the applications, closer to the compute, which is what we're seeing across the board. Even if you even look at security, it used to be perimeter, now it's how can we get it closer? They're looking at how can we take this performance and drive it closer to those GPU clusters that compute inside?

Dave Vellante

>> How's he doing understanding your business?

>> Fantastic. You can see he got a five minute brief

Sean Lawrence

>> before this and he's got it down. So an important thing about this, we're talking about transitioning all the way down into the silicon package design. So that's not something any one company's going to be able to shift in the whole industry by themself. It was born out of our labs and our view of the market and the trends, but we early on decided we needed to take an open ecosystem approach to try and realize this and really shift the market to a more sustainable approach for it. So we set up the IOWN Global Forum in 2020 with... Intel and Sony were the founding members. Now we have over 155 members, including everything from the silicon manufacturers, server manufacturers, most of the networking vendors are part of the IOWN Global Forum, hyperscalers. So we're working with all these different companies to try... And it's incredible for me to see the collaboration. We've got people who are competing head to head in the market day to day, but they're all inspired by this vision of moving almost all data transmission inside computers, et cetera, from electronic to optical, all working together on the standards, the road map, use cases, POCs, et cetera. So incredible experience for me to watch these companies come together and forget what's happening in the competitive market and say let's just do the right thing and move IT to optical data transmission for both performance, lower power, lower heating and cooling, and a lot more flexibility in how you design things.

>> And ultimately lower cost. - Ultimately.

Dave Vellante

>> That is part of the North Star.

Sean Lawrence

>> And do you design your own silicon?

Dave Vellante

>> I'm sorry for not knowing that, but...

Sean Lawrence

>> Yeah, no. So in our R&D labs, we have expertise. We don't do actually silicon manufacturing ourselves. We do have some device manufacturing, so we manufacture some of the optical transmission components, but we don't do the silicon ourselves.

>> And you don't design your own silicon, is

Dave Vellante

>> that correct? Or do you?

Sean Lawrence

>> No, we don't do the actual silicon design ourselves.

Dave Vellante

>> Okay, all right. Interesting. Because the reason I ask is with the fabless model, so many firms are now thinking about designing their own silicon, you see it with Amazon, we just had them on, Apple obviously designs its own silicon. Is it something that is a topic of conversation in the industry or is the value add that you could bring from designing your own silicon not worth it?

Sean Lawrence

>> Yeah, I don't think I'm the right person to give you a good answer on that one. Sorry about that, Dave.

>> No, that's okay. So talking about the ecosystem,

Dave Vellante

>> it's a pretty rich ecosystem that you talked about in order to get... So the data center is a big target market for you guys. talk about your served market.

Sean Lawrence

>> Right. So the early stages of IOWN, so we have the WAN side, the long distance is ready for production. We've actually launched it. NTT has launched a carrier service in Japan. We call it All-Photonics Network. And overseas this year I think we're going to try and get to market. We have some ideas how we're going to do that. The later stages, getting to the circuit board to circuit board connections via optics and eventually on the circuit board and into the silicon, those are different stages of R&D. So the circuit board to circuit board connections, the first, what we call IOWN 2.0, we're going to actually have a prototype running at the Osaka Expo coming up starting next month. So that's not too far from reality. We're hoping next year maybe we can commercialize that second stage. So we get first stage long distance, second stage between the circuit boards, getting close, and it's probably out to the 2030s before we get it really down into the silicon die.

Dave Vellante

>> And the data center is the target market, correct? So the data center super cycle that we're in now, some people don't like that term, but too bad, we're in a super cycle, that's a tailwind obviously for you. I want to share, Sean, some numbers that we just came out with in theCUBE Research and get your reaction. The data center market was... When I think about everything, all in, power, cooling, compute, storage, networking, optics, everything, was about a $200 billion market forever. Up a little down, down a little, COVID it dropped down a little. And the vast majority of that was general purpose computing dominated by x86. I mean 90% of it. In 2024, the market jumps from 220 billion to 350 billion. Amazing. And the accelerated computing component of that went from single digits to 25% overnight. It's amazing how the S-curve goes. And we see this growing at a 15%, the whole market, compounded growth rate. The accelerated portion, we call it extreme parallel processing, Jensen calls it accelerated computing, growing in the mid-20s over the next 10 years, hitting a trillion dollars by 2032. So we had a $200 billion market going to a trillion dollars in essentially a decade. And everybody I talked to says, "Yeah, that's kind of what we're seeing." Are you seeing that?

Sean Lawrence

>> Yeah, so now actually with those numbers, they're astounding numbers, but I understand better... Our global data centers CEO, Doug Adams, I understand his situation better now. I have been talking to him about how we can use these IOWN All-Photonics Network to improve the data center business, data center interconnects, et cetera. And he was telling me that his team... We're the number two or three largest data center provider globally, depending on how you measure things. But he was saying he is just building data centers as fast as he possibly can. He just can't build them fast enough to keep up with demand. And essentially it's all coming from the AI boom. He's just building data centers as fast as we possibly can, and he's just running 110 miles an hour with that.

Dave Vellante

>> And Bob, you and I have been talking about this all week, the importance of connectivity now and the network in order to support this AI boom, it doesn't work without that capability.

>> Exactly. Well, in the critical aspect of adding optical

Bob Laliberte

>> to this, some of the things that you're hearing now, because of the lack of power availability, as you're building those data centers, people start talking about distributed GPU clusters and how do you do that? So this is one of those areas where using that All- Photonic Network, all optical network is going to enable that because of the lower latency, the higher performance, higher throughput. And just to be clear, when we talk about All-Photonics, prior to this, there were optical net DWDM, things like that, but they had to go through a lot of optical electrical OEO conversions back and forth to get amplified and so forth. So being able to take out some of those components and have it be all optical will help streamline that and it'll enable these data centers as they're being built to... And there's always the speed of light times index of refraction of glass. So there's always going to be certain latency, but using optical, you can always ensure that you're going to have the best possible performance with the lowest possible latency over a copper line.

Dave Vellante

>> Are you working on that speed of light problem in the labs?

Sean Lawrence

>> That's the hollow core fiber thing. We are working on that. . But exactly, Bob, what you're talking about. I mean, our Docomo business did an experiment... I mentioned we have APN as a carrier service in Japan, and we had two data centers about 40 kilometers apart, and we had some GPUs in one data center and some GPUs in the other data center, and we connected with the APN and we did an AI LLM training on top of that. And the increase in training time, having this GPUs distributed versus all in one location, was less than 1% increase in training time, even though we had GPUs distributed. So with the right networking and the right design, we really can use distributed data centers, and we think that's a really important thing. Besides just the number of data centers, they're getting too concentrated in small areas and there's just not enough power. There's no real estate, there's no power. Different governments are saying, "No more data centers, we can't take it. " So we think this All-Photonics Networks can allow people to use more... It's like a virtual extension of the data center where you can move things 50, 70, 100 kilometers away from the core area, but with a good enough network, it's almost like you're in that same area. And we're hoping you can do that where you have sustainable power, not just... You need power, but hopefully it's also somewhere you have sustainable power.

Bob Laliberte

>> It's funny, it brings me back to the early 2000, when organizations were building out their synchronous data centers, so they had active- active environments and leveraging optical to stretch it out. About 100 kilometers back then was the... That was about it. That was the... If you wanted to keep it active-active, fully synchronized, obviously way more performance issues with a GPU cluster than just performing synchronous data replication and things like that in an active-active environment. But clearly it's great to hear that optical's coming back and going to be playing a bigger role in enabling this next huge transition and enabling organizations to be able to build out these high- performance GPU clusters in a distributed fashion and be able to take advantage of available power and space requirements.

Sean Lawrence

>> Yeah, we also have a couple of the IOWN Global Forum members doing experiments around redundancy and backup, et cetera, and what you can do, and a couple interesting cases. So MUFG Bank, it's a mega bank in Japan, kind of sponsored a POC, they're part of the OWN Global Forum. Could we move a running VM from one data center to another data center over the APN with less than one second of downtime. And could successfully do it, running VM, move it across, so one second of downtime, less than that. We also set up the first international APN for our R&D forum. We worked with Chunghwa Telecom in Taiwan. We set up an APN between Japan and Taiwan, 3,000 kilometers, 17 milliseconds latency, and doing different experiments across there. But one of them was really interesting, looking at database replication. So we took one and a half million 8K files, it's like the transaction files that you use to replicate the database, and doing it locally would take 47 seconds to replicate. Doing it over the 3,000 kilometer APN, 50 seconds. So you're pretty much there in terms of you can keep those databases synchronized even over really long distances.

Bob Laliberte

>> And business resilience, as we know, is critically important, so having that greater geographically distanced with the same performance, it's going to be incredibly important.

>> And coming back to the economics

Dave Vellante

>> and some of the numbers that we talked about, I mean even if they're directionally correct, which I think they are, do you have a base case for that crossover point, an assumption for when that happens? Are we talking near-term, mid-term, or long-term?

Sean Lawrence

>> The long-distance is there already.

>> Yeah, absolutely. - The long-distance is there already.

Sean Lawrence

>> So the short-distance, actually,

Dave Vellante

>> I don't have the economic modeling numbers there.

>> So it's still trying to figure that out in the labs,

Dave Vellante

>> experimenting with different things.

>> We're working on the science and miniaturizing,

Sean Lawrence

>> miniaturizing, getting the circuit board connections. But we think there's already a use case. I mean it's targeting high-performance computing. It's high-performance, low- power consumption, more flexibility. So we think there's going to be a use case for it soon, not 10 years out.

Dave Vellante

>> And I mean if this is right, the blue is accelerated computing. I don't know if you can see that. Probably can't. But if that's anywhere close, if it's all high- performance now, and that's again, a real challenge to you. I wanted to close, when you were talking before, you're in Japan, I haven't been to Japan. I went a lot in the '80s and '90s. Haven't been since the late '90s. What's it like these days? Culturally probably very similar, but there has been changes certainly since I've been there. The economy. I mean, back in the '80s Japan was going to dominate the world. That didn't happen. Now China's going to take over, we'll see. But still amazing culture, amazing place to live. What's it like? Give us a quick summary.

>> So it's my third stint in Japan.

Sean Lawrence

>> The first time I went there was the early '90s, and I was in Singapore for a while. I've been kind of back and forth. So it's my third time there. So first I wouldn't keep going back if I didn't like the place. It is a great place. I mean the culture, the nature, the food, the history, et cetera, fantastic. So it's a great place for all that stuff. Compared to the '90s when you were there, it's definitely easier to get around now. I mean, as a foreigner, I speak Japanese now, but when I first went there, I didn't speak Japanese. It used to be pretty rough to get around, just to order food. Things a lot easier now. Part of that is technology. I mean the AI translators, et cetera, they help so much. You just point it, it tells you, you can talk to it, it translates. But still, it's easier to get around. You can communicate with people more easily. Still, it's a fantastic culture. I guess the overall economic situation... I mean the population is shrinking. And we're a high market share, ultimately consumer- focused business in Japan as a telco and a mobile operator. So we got high market share in a shrinking economy, it's a bit of a tough place to be. So we definitely need-

>> You got to expand globally, yeah.

Dave Vellante

>> You got to expand, and we've been putting a lot

Sean Lawrence

>> of effort outside-

>> Keep having babies, people.

Sean Lawrence

>> So it is. That's not just Japan,

Dave Vellante

>> but most of Northeast Asia. Japan, Korea, China, Taiwan, everybody's got this population problem. So it is a bit of a drag, but it's also a driver for innovation. So using more and more robots, more and more automation, trying to keep away from labor- intensive, time-intensive work. So it actually is driving a lot of innovation in the technology side as well.

Dave Vellante

>> Sean, thanks so much for taking some time with us and great to meet you. I hope we can talk to each other in the future and maybe in Japan.

>> Fantastic. It's been my first MWC here, really exciting,

Sean Lawrence

>> and I hope to see you next year.

Dave Vellante

>> You bet. - Absolutely.

Dave Vellante

>> Okay, and thank you for watching everybody, this is Dave

Bob Laliberte

>> Vellante for Bob Laliberte. Savannah Peterson is up next. We have a special surprise for you. You're watching theCUBE MWC 2025. Be right back right after this short break.