Cloud AWS re:Invent Coverage | Chivas Nambiar, AWS

Clips
News
More from Cloud AWS re:Invent Coverage

Chivas Nambiar

GM, Telco Business

AWS

Generative AI meets telecommunications: AWS redefines network agility and customer experiences

As industries evolve in the digital age, few sectors exemplify transformation as vividly as telecommunications. Key cloud players, such as Amazon Web ...

play_circle_outline Unlocking Revenue Growth: Business Model Transformation and Ai Agents in Telco Industry

play_circle_outline AWS advancements in networking and infrastructure

play_circle_outline Telecom Industry's 2025 Vision: Empowering Connectivity Through Programmable APIs and Gen AI Use Cases

Info
Transcript

Chivas Nambiar, AWS

Chivas Nambiar

GM, Telco Business AWS

TheCUBE coverage live in Las Vegas, day three. Swami finishes keynote on AI agents, infrastructure, and gen AI focus. Chivas, AWS Telco GM, discusses networking trends, cloud benefits, and business model transformation. Telco focus on generative AI, network transmission, and migration to cloud for agility and scalability. Growth areas include revenue opportunities and adjacency with partners like TELUS and Sunrise. Regional customization and control important in telco sector. Telcos facing challenges in moving from POCs to production, need for choice in model... Read more

explore Keep Exploring

What are some of the trends in the industry that are being driven by the revitalization of cloud and cloud scale technology? add

What use cases are telcos trying to solve for, and how might advancements in technology affect the development of edge inference use cases in the marketplace? add

What are we going to be talking about in terms of how gen AI is going to impact businesses? add

bolt Powered by CUBE AI

Chivas Nambiar, AWS

search

>> Welcome back, everyone, with theCUBE coverage here live in Las Vegas. I'm John Furrier, host of theCUBE. This is day three. We've got another whole day coming, but again, a lot of slew announcements. Swami's on stage finishing with his keynote. Again, this is about the AI agents, more of a deep dive and as you see the infrastructure focus, focus on agents. The focus on data around gen AI is going to be the big focus. Chivas is here, general manager at Telco business at AWS. Of course we follow greatly because of MWC, but also networking is the fundamental component of everything in telcos. We can talk about networking. That's the key one. Chivas, great to see you again. Thanks for coming back.

>> Yeah. It's fantastic to see you. Thank you for having me.

>> So everything that was hot and cool is still cool again, so networking, telco. Again, these industries just starting to see one of the big trends, and we've talked about this last time, is that just the revitalization of what cloud, cloud scale brings to these industries. Telco's one of them, but every industry, because you now have the ability to make the pre-existing stuff better. I'd like to get your perspective on that because as we go into Mobile World Congress next year, we've been already working on, "Okay. What's going to be the editorial angles?" And obviously one of them is business model transformation, which we talked about last time. It's not just technical digital transformation. The business model piece is now emerging. All these new tools, all these new capabilities are coming in at the infrastructure level, significant advancements and breakthroughs.

>> Yeah. And as you know, you've worked with AWS a bit, so you know we think in terms of flywheels. And you think about the normal telco flywheel at this point is usage and usage drives revenue and usage drives the network build out and then the network build out and the capabilities there allow you to go out and build better products. So if you think in that flywheel terms and now you say, "Okay. How do I apply generative AI? How do I apply cloud to this flywheel?" I think that's where what you're talking about is really coming to bear. In that entire business process, generative AI is helping you think through at the network operations level, at the business and the product creation level, at the actual customer experience level. All three are getting accelerated and no better place to do that than in AWS and the cloud.

>> Yeah. I want to get into some of the scale technical things, as well as the business model , but I first want you to set the table and share with the folks how you're looking at the business right now as general manager of the telco business. First, describe what that looks like. What's your focus?

>> Yeah. Yeah. So it's been an interesting year. We've spent a lot of time really focusing on what our customers are asking us for. And it's three areas and then a growth area. So it's like three areas on generative AI, obviously. Everybody's interested in figuring out how to apply this to as much of their business as possible. The second area is network transmission. We had some interesting announcements here this week. We announced that Comcast, for example, has moved their 5G core over to run on AWS. And that comes on the backs of an announcement with Telefonica where they've moved up a million subs over to a region in Frankfurt. And then they've also talked publicly about how they're going to add another 4 million subs in the next 18 months. So you can tell that 5G network core is now something that's a part of what we do. If you think about then the big migration modernization effort, we're hearing a lot of questions and conversations about how do you continue that journey, take all of that footprint of things that are running on some of the partner solutions and move them over to the cloud because you need that agility. You have all the insights coming from the gen AI work, but for you to act on it, you actually need systems that are addressable via APIs and that can scale up and down. So those are the three areas in the telco business itself. And then we're seeing revenue growth opportunities and adjacency. So things we're doing with, for example, TELUS, where we've built a smart home ecosystem with them. They're thinking about how do you scale that out globally for other operators. We've worked with Sunrise in Switzerland to build a small and medium business marketplace that allows them to address that marketplace with cloud solutions in much more targeted fashion. So those four areas I think are where we're spending a lot of our time.

>> Well, and also you highlight some of those names also. Global is obviously a big part of it, the regions. We had Tanuja Randery on from EMEA talking about the sovereign cloud. So these regional approach is also the nice piece for the folks to plug into Amazon. I love the message that you guys have around the flywheel, around giving people control of their destiny and at scale. And I think one of the topics that's coming up is... And you see this in the telcos, some of the names you mentioned. They're not small companies. They're running massive amounts of transactions. We had JP Morgan Chase CIO on theCUBE yesterday, Laurie Beard. She's moving 10 trillion a day. I mean, that's critical infrastructure. So for her... Even the bank. Right?

>> Yeah.

>> So talk about the scale piece of this, because your customers are operating at such scale, okay, and they also want to control their own destiny, control their money, so to speak, because they want to make that business model transformation. What are some of the comments that you have on that? Because this brings the confluence of the business model with the scale and technical challenges of keeping everything going fast, secure. I mean, Dave Brown was even mentioning some of the advancements at the silicon level coming in and helping out.

>> And I think what you mentioned there about the regions and customization for the regions is super important. So we'll touch on that in a second. If you think about what we're doing with telcos, telco's a nationally critical infrastructure in every country. They're typically highly regulated, so they have bounds on how they can use their data. And we're working with them and working backwards from those requirements. So you've seen what we've done in our European sovereign cloud and the announcement of what we build there that is very focused on what that set of users need. And now that you can see tier one carriers really say, "Okay. I can build a solution on top of AWS technologies that allows me to meet these regulatory needs, that allows me to actually build and scale, but it's not just I meet regulatory needs, then I'm actually able to build a platform and a set of solutions that works better than what I had today." And so back to Telefonica, it is one of the top three carriers in Germany. The expectation that they have been able to take the set of workloads that's core to their business, run it, and they've talked publicly about how they can scale up and down as needed, and actually put that in the regions, so which means we've solved for both the technical challenges, the security challenges and the regulatory challenges of that core system running in AWS. And we think our journey over the last 10 years of learning with telcos has actually put us in a position to help them scale. So we did DISH, which is the first nationwide network in the US. We learned a ton from our customers that way. And with Telefonica and now with Comcast, again, these transitions are getting much faster. So we think the scale is... Of course, at every point when you hit a certain threshold of scale, you have to work through some of these challenges, but we have a path.

>> And they have the data too. Tons of data. Talk about the problems. As you guys work backwards from the customer, what are they telling you are some of their core challenges that you guys are helping them overcome?

>> Yeah. A lot of interests and focus is really on the gen AI side of the house right now. So their primary focus is, "Well, last year was the year of POCs. I've seen customers who have 100s of POCs that are running. And then how do I take these into production?" Then you start to see all the challenges. So there are a couple of things that we're working with them on. One is how do we provide the choice? AWS, the way we think about it, customers are going to need to make different choices in terms of models, in terms of platforms they're going to use, in terms of ISV partner solutions that they'll use. How do we provide you with a platform that you can build that gives you the choice of the models, that gives you the ability to take all the work that you've done to protect your data, to provide for safe answers back when you send something into these models? And then how do you bind that to all of your corporate guidelines around how you use this data and then not have to redo it again? You make another choice of a different model, you don't have to go recreate all this. So our work with Bedrock and providing the model there is important. I'll tell you though in the summer this year we did a survey with TM Forum with about 200 of the decision makers at the telcos to say, "Okay. What are you seeing with gen AI? And how are you using it?" And it was interesting to see some numbers come out of it. There were only 14% of them that had done more than 10 use cases. And there were about 50% that thought they had a platform that they could start to build and scale. And only about 25% that actually had a point of view that said, "We have the talent and the capability to go drive this." So there's a lot of demand in that space. And so what we did with some of our leading customers, like a British telecom here, is we said, "Well, let's help figure out what that architecture looks like." So we've built the gen AI gateway model with them that allows them to send requests not just into Bedrock but also two multiple providers. It puts all the same guardrails around it and then it provides landing zones for multiple use cases to come on board and just activate, because that is the hard part, activating.

>> So they're actually... The numbers actually are more than I thought, but that's still compelling... There's more headroom there.

>> There's tons more headroom.

>> I mean, they're right now just trying to figure out their workflows, what data. And obviously the data becomes a big part. We're hearing here this theme at re:Invent is clear. The enterprise and/or anyone who has data is not just going to let it fly around uncontrollable and models that aren't theirs, or if they do use models, they're highly curated, highly specific with some resilience framework. I mean, this is what we're hearing. I mean, this is where we're at.

>> And that's the learning over the last year. If you look at... I talked about the telco flywheel. You look at usage in the network and the customer experience, all of that throws off data. And that data is what we've been using for the last two decades to really optimize what we do in each of these segments. But what generative AI is doing is it's accelerating that. It's accelerating the flywheel because you can now take a business use case and say, "Okay. What data do I have around it? What insight can I drive from it?" And then back to this conversation about if your apps are modernized, if your apps have been migrated, then you can act on them. And that closed loop is what is actually driving the acceleration.

>> So obviously gen AI is a focus and you're seeing that people starting to figure out these landing zones, these kinds of frameworks to get the innovation wave going. What about at the infrastructure level? Are you seeing any advancements there that is interesting to the telecom sector around some of the things you're doing? I mean, Dave Brown and I talked obviously training of two. We talked about training of three. They pre-announced it here. They didn't really give specifics. But if you look at how Amazon's going with their capabilities, it's getting smaller, faster, less expensive to run things. So natural progression would be hosting outpost-like capabilities, not necessarily outpost, but pushing it to the edge. That might be appealing to a telco because for form factor, I could co-locate Amazon in there. Is there any advancements that you can share around the infrastructure innovation?

>> So I think the way I would think about it is like what use cases are the telcos trying to solve for, right? Up to now, a lot of it has been high-value use cases where if you're able to take a transaction and that transaction cost you, let's say, $50 for a truck roll, can I spend a couple of dollars to get the insights to avoid that truck roll? That's great. As the cost of this infrastructure starts to go down, you can start to address more of these use cases, and then there will be edge inference use cases that I think long-term will develop in the marketplace. But if you look at where most of that engineering work is going... Think about the work we've done with South Korea Telecom right now, where they've had to go and build a Korean model, large language model, that is fine-tuned with their data and they've announced that they've built it, they've launched it. And that is a fair amount of work that's running on AWS infrastructure that is scaled. And we'll see more of those use cases I think as the cost of this inference comes down, as the cost of training comes down, especially with our next generation of inference and training chipsets going up.

>> Yeah. A lot of people don't realize that these language models are... Most of them are in English. And so foreign local characters, local language is a big opportunity.

>> It is a big opportunity. And the context in a specific country, the context in a specific society is important to get the right answers out. And then what they saw was when they did the fine-tuning, they saw dramatic improvements, up to like 70, 75% improvement, in the accuracy of the data that came back. And then they've talked about how it's standardized responses. You might get questions in many different shapes, but you want similar answers back because that represents your brand. And then the last thing they talked about is when they bring in new people into the ecosystem and into the contact centers, as an example, it accelerates the training process much faster. It reduces the amount of stress and trying to figure out data from multiple different systems. And so that's... All of it's starting to show up in the results for telcos as they take advantage of gen AI.

>> Any other things going on around AWS and your area telco that you see that's notable? You mentioned Connect. I mean, Connect is an AWS service that did extremely well during the pandemic. Now it's the core flagship in the new AWS Solutions Group under Colleen Aubrey. And so I can imagine that Amazon's got some goodness it could bring out to the field, so to speak, or areas. Anything there to report on your end? Are you seeing anything coming out of Amazon that you're going to be able to put through telco from a business perspective, I mean, Connect? Are they using Connect? They have their own? How do you see that working?

>> Telcos is some of the largest call center providers in the world. They host it for themselves. They host it for other parties. We are seeing with customers like VM O2 in the UK where they've been able to do this migration of Connect. We've talked about DISH doing a massive migration of 15,000 agents within a few weeks-

>> To Connect?

>> To Connect.

>> Oh. So they're adopting it?

>> They're actually using it and adopting it. And we actually had a session at re:Invent this year where the VM O2 team talked about how it is that they were able to move and migrate these workloads over. And now what that opens up is they're able to go take advantage of all the gen AI capabilities to improve that conversation and the life cycle consistently.

>> Tongue in cheek, I've said this, they move with the speed of glaciers, but I say that in a complimentary way, but I will also compliment the telco businesses because they have such a large business. So I mean, they have experimentation. They're not moving as fast as the nerds and the entrepreneurs want to move because of course they want adopt new gear, new stuff, so... But I think they're pragmatic in the sense of if there's a business model, customer value. Because there's a lot of stuff to do if you do some big change in telco. And I think that's my caveat. And I always joke because I love new technology, "Telcos should just go to the cloud." This is like 10 years ago, like, "Okay. A little slow." But now they see the value and because the impact is there, the real dollars and cents practical use cases... Actually Andy Jassy used that word on stage yesterday, practical ai. And I think that is a key story here. I mean, having actual practical implementations that either reduce costs or drive revenue.

>> Absolutely. And in the end, if you think about the consumer experience that they get, that all the consumers in the enterprise get from a telco, the question is what have we now been able to do in the last, let's say, 10 years? We've invested a lot of money in building out 5G networks. You've had the promise of 5G SA providing you with slices on the network, but that has required us to take all these systems and modernize them. And telcos have been doing that in the background. So I see that work has happened. The next step of this is how do you then drive revenue out of it through productization? How do you drive revenue out of it through the experimentation that requires productization? And with engineering teams now getting gen AI assisted ability to go build products, I think that'll ramp that up. The ability to go look at all the data that they have and then be able to use it in a seamless fashion that'll help ramp it up. And finally, on the core network itself, as you can see, the network workloads are starting to become more cloud-native, and that'll change again.

>> I'm very bullish on telecom. I've always loved the networking business. And I think networking is now moving everywhere up to stack, even the concept of networking. LLM routing is seeing that as a big part of what Swami was introducing today around all these new ways. Just let models talk to each other. That's a routing problem. I mean, this is networking, right?

>> Yes.

>> So networking comes up there. Chivas, I got to ask you to wrap up here because I know you're busy, but Mobile World Conference is coming up and we're doing our preparation. I know you've got yours. What's your vision as we go into 2025? MWC will be later in the first half of the year. Always a bellwether show in Barcelona. What's your vision for the industry? How do you see the agenda? What's the conversations that we'll be having going into MWC this year?

>> I think the conversations that we're all going to see is about how do you take the connectivity plus place. So if you look at connectivity as the baseline of everything we do, that is going to become programmatically accessible through APIs. That thread is really starting to gain momentum now. There was a joint venture that was announced with a bunch of operators, and we can see in partnering with them, we expect these APIs to start to become available globally. Once that's done, then I think you start to get industry use cases that can address it. And to take advantage of that, internally, all of these telcos have to go through and finish the migration process. They have to get really good at gen AI and using gen AI to drive their business. So I think that's what we're going to be talking about. We're talking about how gen AI is going to impact the business. We're going to talk about network transformation. And then we're going to talk about revenue-generating use cases. And I'd love to spend more time building new use cases for end customers.

>> And we love those customer success stories too because people want to see what their peers are working on. As general manager, you happy with the business? What are some of your business goals going forward? What are you trying knock down over the next few months?

>> We're always working backwards from the customer. So our customers are telling us they want to move faster. They want to build more revenue out of all of the capabilities they have. That's where we're laser-focused. I'm pretty excited about it.

>> Thanks for coming on theCUBE. Getting the telco angle here on theCUBE and AWS re:Invent. I'm John Furrier, host of theCUBE. Stay with us for another wall-to-wall coverage day here in Las Vegas at AWS re:Invent 20... Our 12th year with theCUBE. Thanks for watching.