Guy Daniels, TelecomTV (00:05):
Hello, you are watching TelecomTV. I'm Guy Daniels. To discuss the intersection of telecom infrastructure and the AI revolution, including multi-cloud architectures and AI factories at the edge. I am joined by Eric Van Vliet, who is VP Telecom System Business, Sale and GTM EMEA for Dell Technologies. Eric, really good to see you again. Can we start, I'd like to ask you what you think are going to be the top three main challenges that the telco industry has got to address right now?

Eric Van Vliet, Dell Technologies (00:44):
I think, well, the top three challenges for the telcos haven't changed over the last couple of years. I believe they're still much challenged with the increase of bandwidth. The driver behind that increase of bandwidth might be changing. AI is a big piece of it today, but it's still probably the biggest challenge they have is the increase, the continuously increase of the demands of the subscribers from traffic. I think the other two challenges are relatively new these days. One is obviously different use cases that are demanded by the subscribers and also maybe more importantly, the enterprise usage of the connectivity that they deliver. And the other one is talent. I think third challenge that they really need to address now is getting the right talent through the door to enable them to deliver the use cases that are so in demand within both the consumer and the enterprise space.

Guy Daniels, TelecomTV (01:52):
Okay. Well thanks Eric. Well, if we focus in now on this intersection between telecoms and ai, how would you say the secure, reliable and scalable infrastructure of telcos has enabled the development of AI use cases?

Eric Van Vliet, Dell Technologies (02:11):
So obviously AI use cases all live around the data. The data created by their customers of the telco. I think with a longstanding history with the customers and handling their data, starting at the beginning with voice going towards internet IP traffic, so real data over ip, which then they started caching, they started securing in a regulated way as they always have been. A regulated industry has meant that they've built it in a very reliable infrastructure to carry that data. Now because of modern connectivity with mobile and high speed fiber connectivity, we see quite a large disaggregated network now where a lot of the pieces that drive that connectivity are distributed throughout the network. This is ideal for a lot of AI use cases, maybe not training as we see the large language models mainly driven inside the data center, but use cases like inferencing or small language models, they are ideally placed close to the data and who else is close to the data? Then the connectivity provider, they're right there. They deliver either the fiber or the wireless signal to the user, to the customer. They're never further than a couple of miles away with a access point or a pop. And I think that's one of the key aspects there. Real estate, and as I mentioned, they are regulated industry, very hot on security and reliability. This is ideal for the next evolution of where we're right now with AI becoming a part of everybody's life and everybody's business.

Guy Daniels, TelecomTV (04:06):
And how do you think telcos actually view this? How do you think telcos see AI technology and what do they make of the resulting AI use cases?

Eric Van Vliet, Dell Technologies (04:17):
I think AI for the telco has a multifacet technology. Obviously they can leverage it themselves to drive more cost optimization via investment in ai. The telcos have been on a continual cost of optimization journey throughout the last decades to deliver better, more reliable, but also more cost effective connectivity to its customers. So one aspect is that it enables them to drive cost down and modernize that network by enabling the network to become smarter and more efficient and hopefully autonomous, as autonomous as it can, but also it gives them an opportunity to monetize on the connectivity. If you look back into the last two decades, a lot of the monetization of the network actually was happened by over the top providers as we know, because there was a lot of one way communication, and let's say what it is, it was media streaming was a big part of it.

(05:27):
With AI where it's continual interaction between the customer and the ai, there's a lot more opportunity for the telco NDC, that's to find ways to monetize on the need of that continual bi-directional communication. And I think it's an opportunity for them to really build out the edge now, whereas before, they were always looking for the edge use cases to justify the investment as we all were really looking for that. And AI is that use case, the data that the subscriber creates, the way it's handled by artificial intelligence, the way the token become, let's say that digital spine for AI is perfect, I think they're in a fantastic position to help us leverage AI more in our daily life. Enterprises leverage AI as a business and I think the telco seed as a way to monetize on the usage of their connectivity.

Guy Daniels, TelecomTV (06:25):
Well, lemme pick up on something you just said there, because very recently at Mobile World Congress, Doha, you described telcos as the digital spine of our connected world. What do you mean by this and why is this concept so important with ai?

Eric Van Vliet, Dell Technologies (06:41):
Well, and I mentioned this in Doha as well, if we think of intelligence, our collective intelligence only really works at the pace we are used to now is when we got connective. If we look back 50 years ago, the evolution of technology or evolution of medicine or any advancement that we were making, it was great, but it went at a speed which we now would class as very slow. So it's incredible to see how connectivity has helped accelerate our evolution in any form. And I think now that we're looking at artificial intelligence and we're bringing additional intelligence to the play, and in case it's digital, right? It's artificial connecting it to not just the users, but also the data creators, the entrepreneurs who create new use cases or they create new apps. If you see how fast an application can be built, they have to all still be connected.

(07:50):
And that's why it was referenced as a digital spine because we see the AI as the brain and we connect all the data that are, let's say, the limbs to that brain through the spine. And that's the telco. The telco is not optional. The connectivity is mandatory just as a spine towards the brain. The digital spine is there, is needed and is really important for its success. It's essential for its success. And same as reflexes if you think of it. That way we can build in let's say behaviors of that spine related to the data to help the artificial intelligence get better at handling the data as well. And that's part of the telco opportunity with artificial intelligence.

Guy Daniels, TelecomTV (08:40):
And as you say, the telco is not optional. So what unique advantages do telcos have when you compare them against hyperscalers or other players when it comes to becoming their backbone for ai?

Eric Van Vliet, Dell Technologies (08:54):
It's the access to the users, the data users, and the data creators. That's their biggest advantage. They have built an infrastructure that goes across a country that is not centralized, which is near the users and the creators of data. And that's their value. That's what they've built. And I think if we look at hyperscalers, even though they might have a great opportunity to build large power hungry data center to house the large language models and some of that artificial intelligence, without the data, without access to the data and without access to the users, it is irrelevant. So I think that's the big, big advantage that the telcos have. And they're doing a great job of continuously servicing those users, investing in technologies to provide faster connectivity, more reliable connectivity, and now it's time for them to build to maybe actually provide smarter connectivity that's more than just a simple pipe, but actually adds additional intelligence to that pipe to make sure that what we're building as a society and as businesses just keeps advancing and the momentum just keeps going. When we get things like physical AI such as the VR glass capabilities, robots, autonomous cars, and so on. The telco, again, is not an option, it's not optional. They need to be there and they're crucial to the success of all those use cases. Without the telco, there is no way to deploy artificial AI at that skill that we'd like to see.

Guy Daniels, TelecomTV (10:44):
Yeah, sure. And as you say, there's a lot still to come. And if we look at distributed networks, how do you see the interplay between multi-cloud architectures and the telco edge in delivering these AI services and services at scale?

Eric Van Vliet, Dell Technologies (11:00):
So the edge, and I think I was at one of your previous events I mentioned if we look at the distribution of compute right now within the network, many data centers that the telcos have built, those are ideal to house also some of the ai, let's say, systems to do things like inferencing or other elements around artificial intelligence. The reason why this is not yet a reality today is partly because running these distributed cloud technologies is complicated. If we're looking at multi-cloud architectures where we look at an infrastructure and see that infrastructure more than what it is beyond the physical we see it from, it is a cloud. First of all. Not every cloud is ideal for the same use case. So we see different clouds coming, hence where I now have clearly a multi-cloud architecture within any network, some focused on connectivity, some focused on control, security.

(12:10):
And we'll also see a multi-cloud focused on edge use cases that are delivering AI capabilities to the subscriber and to the enterprises. Now, without looking at your infrastructure's multi-cloud, and if we look at all those silos, it'll be very hard to manage all these different clouds and more importantly, manage them at an operational cost efficiency to make those use cases affordable for your subscribers. So the only real way to drive cost efficiency is start looking at your infrastructure as a multi-cloud architecture, accepting that the infrastructure is not as siloed as you might want to have it, because from a security and performance perspective, a silo is ideal because it's very contained. But from an operational perspective, a silo is very expensive and difficult to manage. So going to a full multi-cloud architecture in your infrastructure gives you the offshore, gives you the reality that an edge cloud is just a small piece rather than a whole new silo you have to build. So that's why I think multi-cloud architecture is needed to drive the right cost efficiency when building out an edge.

Guy Daniels, TelecomTV (13:24):
Well, final question to you, Eric, because you've spoken previously about AI factories built on pre-validated infrastructure blocks, but what does an AI factory look like in a telecoms context, and why are common building blocks so important to their success?

Eric Van Vliet, Dell Technologies (13:44):
So we just mentioned multi-cloud and that we have different clouds for different use cases. I think AI factories at scale, we cannot build AI factories for each use case. There will be a common AI factory that is leveraged by a telco operator to deliver both the use cases for their own network and their operations and the use cases for their customers. So without building it on validated building blocks, if you're not going to be able to leverage common applications or way of workings that are built for, let's say, an AI factory framework, if it's your own specialized framework, this means everything you do will be specialized. And that's just not how we can see the advancement going. So if we're looking at a telco operator, a telco operator, AI factory, we already see operators do this investing in one larger factory to run their own operations, the cost optimization, and then looking how to extend that factory to the field or to let's say aggregation points within the network.

(15:01):
But being part of that overall AI framework based on common building blocks to deliver those use cases to subscribers. That's the only way we believe that you can get economy of skill as a telco to run an AI factory. If you would run an AI factory for your siloed use cases, internal network, and then also for edge, the cost alone and the complexity of that would just be impossible to manage at a profitable space. So it's really important that we build AI factories based on common validated building blocks, and think of it as one whole infrastructure, AI infrastructure that we leverage for all the use cases that we see within ET telco, cost optimization, network modernization, and the edge use cases for the subscribers.

Guy Daniels, TelecomTV (15:52):
Absolutely fascinating. Well, we must leave it there for now. Eric, great talking with you. As always, thanks so much for sharing your views with us today.

Eric Van Vliet, Dell Technologies (16:00):
Thank you for having me. Happy to be here.