Clarence Reynolds, TelecomTV (00:13):
Generative AI is rapidly transforming the telecommunications landscape in customer experience, network operations, and other real world applications already in use. Manish Singh, CTO of telecom systems business at Dell Technologies. And Ronnie Vasishta, senior VP of telecom at Nvidia, join us to discuss the evolving role of AI in telecom. Welcome to you both. And Ronnie, I want to start with you as we open this year's AI native Telco summit, there's a lot of talk about generative AI and what it can do for telecoms. How much has evolved so far this year?

Ronnie Vasishta, NVIDIA (00:52):
Well, a year is a long time in AI for sure. What we've seen this year is actually some dramatic improvements and advancements, I'd say in the area of telecom and the use of gene ai. About a year ago, it was a lot of evaluation and even some skepticism I would say about AI and the benefits of AI to telecoms. But what we've seen over the course of the last year is much more of adoption, moving in many cases from proof of concept into some real deployments. And if you look at some of those areas that we're seeing in the use of genai of customer improvement, customer service management, and then of course what we've started to see as a lot of infrastructure deployment to service generative ai, and as we'll come onto maybe a little bit later, is telecom companies are playing a very, very important role in that infrastructure deployment and the infrastructure build initially.

Clarence Reynolds, TelecomTV (01:59):
And Manish, what changes have you observed in the last 12 or so months?

Manish Singh, Dell Technologies (02:04):
Yeah, I mean, I would agree with Ronnie. I mean, what we are seeing is the pace is picking up and different telcos pursuing the adoption of generative AI for different domains, and I'd probably put them in three big areas. I mean, there are clearly a set of telcos, especially across Europe and Asia Pacific, who are building AI infrastructure, building sovereign AI capabilities to offer GPO as a service, as a service providing those capabilities in their markets. And I think in many of those cases, telcos really have this prime real state called the edge. And I think bringing AI to the data where the data is lot the data getting created in the physical world and bringing those AI capabilities closer to the data. So we see that we of course, see customers building solutions now and deploying them in customer experience, and we should talk a little bit about that later. And then third one, I would say is in the domain of the network itself to drive a lot more automation lifecycle management of the network itself. So going from forecasting to planning to install, commissioning, deployment, configuration optimization, the application generative AI across that network's lifecycle is now continuing to build and grow for the top.

Clarence Reynolds, TelecomTV (03:35):
So Ronnie, what top two or three objectives are telecoms trying to achieve with gene AI in customer experience and network operations?

Ronnie Vasishta, NVIDIA (03:45):
Yeah, so I would say in the area of customer experience, the first is that actual word experience. It's the ability for customers to get access to issue a resolution as fast as possible, and also to avoid issues for the customers. And generative AI can play a very important role in that. And so we're starting to see the focus on reducing customer churn using generative AI tools, methodologies to engage the customer in a more efficient and an experience for them that isn't as frustrating as traditional methods when it comes to network operations. That's something that I would say is a little bit earlier in the phase of deployment. We're starting to work with ISVs and so independent software vendors and global system integrators that have access and have ability to engage with telecom companies in terms of the network operation data. And that network operation data lends itself to AI in the form of how do you improve the productivity of managing a network. That's still in early days, but we're starting to see some extremely promising results from that.

Clarence Reynolds, TelecomTV (05:19):
Manish, what are the most important objectives you see for gen ai?

Manish Singh, Dell Technologies (05:23):
Yeah, I mean, especially Clarence, coming to your question around customer experience, what we are seeing is telco's really now starting to leverage the power of large language models for customer care, billing care and more. And the unlock here really is to really understand the in intent and the context of the customer and then really delivering a much better experience to the customer. I mean, we have had chat bots forever, but what we are seeing now is migration towards chat agents, leveraging the large language models and delivering much better user experience to the customer, much better. Customer engagement is very good in understanding intent. I mean really understanding if there is a churn prediction, initiate mitigation plans accordingly. And I would like to highlight some of the work that we've done with Esk Telecom in particular around this in the context of a TM forum catalyst project, where we've put together a chat agent leveraging the large language models, but also an m and o AI platform, which then integrates southbound with the BSS systems without requiring BSS transformation. And the outcome is you can leverage generative ai, the large language model for your customer engagement, but then take it further, offer new plans, new devices. If you know the intent for the customer is traveling, you can offer roaming plans, all of that being done dynamically. But with that southbound integration with the BSS systems also automate that whole process. And so it's really about customer engagement, improving that experience, and then understanding the intent and actioning on that all in an automated fashion.

Clarence Reynolds, TelecomTV (07:22):
So Ronnie, what use cases are you seeing in real world applications that use AI in both customer service and network operations? And what benefits are those companies beginning to realize?

Ronnie Vasishta, NVIDIA (07:35):
Yeah, so we're seeing quite a few already getting deployed, and we work with partners, as I mentioned earlier, to enable these to be implemented MOCs, for instance, on some of their platforms for their customer billing services. As mentioned earlier, that's been one that I would say has been leading in enabling telcos to see productivity improvements in terms of customer experiences. We talked earlier about the ability to provide a better experience for a customer through the ability to deploy large language models trained in their local language such that they can see and experience something that is better than the press one, press two, press three for getting access to information. In terms of network operations, we have partners today that are already deploying, say, service tickets. ServiceNow is a great example of that. ServiceNow is being able to deploy and customize using the Nvidia AI foundry frameworks, their own customized version for each telco customer, such that the Telco customer now can see and talk to their network essentially and generate service tickets that are meaningful in a much faster way.

(09:06):
We also have some telecom customers that are using accelerated compute for routing technicians and the ability to route technicians in an order of magnitude faster such that they have a much more efficient fleet management system. So really I would say AI is becoming pervasive across customer experience, across network management. Our ecosystem of partners are starting to become very involved. We've just been announcing more and more partners in the area of system integrators, for instance, that are now training on the NVIDIA tools to be able to bring those tools out to telcos specifically for telco use cases.

Clarence Reynolds, TelecomTV (09:53):
Manish, a few weeks ago your company launched Dell AI for telecom with Nvidia as the accelerated computing partner. What makes this partnership unique and how will it enhance CSPs network transformation efforts?

Manish Singh, Dell Technologies (10:07):
Yeah, sure. I mean, we've been working with NVIDIA to build what we call as our AI factory construct, and think of a factory which would take in raw materials, produce finished goods out in the case of an AI factory. The raw material is your data, the finished goods out are your use cases and business outcomes. And what sits inside the factory floor is the AI infrastructure. What we are doing is working and pulling together a curated ecosystem of telco for delivering the right kind of use cases and the business outcomes, and then also putting services around it to make it easier for telcos to get started and scale their AI deployments. And so whether that's for GPO as a service, whether those are for customer experience use cases, whether those are for network lifecycle management use cases and more. And so as part of that announcement, of course, working closely with Nvidia, we started and announced our first set of partners, Amdocs.

(11:12):
We are working with them on their Amaze platform, enabling a whole suite of use cases there. I mentioned about esk telecom, working with them on the chat agent. We are also working with some of the, call them, the innovators in the space like Kinetica working with them as an example on network troubleshooting. AI is very good in anomaly detection, looking at large call flows, message flows, logs, all structured unstructured data, pull all of that together and identify where the problems are within the network and message flows and quickly troubleshoot them. Similarly, we are working with another partner fy, another innovator I would call, who's looking at the network data and doing fault prediction with very high accuracy, seeing not of 90% accuracy in that. And you can see the implications of that. If you can predict 10, 12 hours ahead, then a network fault going to happen, that power fault for any network operations to reduce downtime, improve network KPIs, optimize their truck rolls and repair on a mediation work that they got to do. So there are more use cases, but the key construct is bringing that AI factory with the right infrastructure, a curated ecosystem set of services, and really focusing on the use cases on the business outcomes that the telcos need, be it for their customer experience, be it for network lifecycle management, or be it for offering GPO as a service.

Clarence Reynolds, TelecomTV (12:44):
What's next for gen AI and telecom? Ronnie, let's start with NVIDIA's perspective.

Ronnie Vasishta, NVIDIA (12:48):
So good question. As I said earlier, a year is a long time in ai. And as we look at what comes next, we can learn a little bit about what's been happening over the last year project forward, but there are some new advancements that are coming as well. But first, let's take a look at learning from what's happened. We've seen that accelerated compute AI infrastructure, as we call AI factories, are becoming great opportunities for telcos to service their sovereign AI needs, but also as business model opportunities for telcos that haven't existed before. We have now in the region of 14 telcos around the world that have engaged with us on building and actually starting to implement and deploy AI infrastructure in Indonesia, for instance, linta, which is a great example of that. And within a very, very short timeframe using the NVIDIA reference architectures, a reference architecture doesn't just mean the hardware architecture, but it's the full software stack to service generative ai, the full product cycle of generative ai.

(14:06):
And there they'll be deploying government services in a local language that has been trained a large language model in their local language. And that's really critical because sovereignty means culture language, it means the learning of that country is retained within that country. So we expect that and we know that's going to continue infrastructure initially centralized, but now getting more and more deployed at the edge as well. And that's a very exciting space. It's that edge of the infrastructure and the edge of the network because that space is a space that telcos occupy, whether it be with point of presence today or mobile switching offices and managed services can play a role there as well as the applications that sit on top of that infrastructure. We're starting to see now the deployment of generative AI onto devices. We're going to see that accelerate rapidly, and it can be obviously beneficial to the telecom operators as well, and the ecosystem around telcos.

(15:19):
So I would say over the course of the next year, we're going to see infrastructure deployment centralized, moving to the edge. We're going to see that we're able to provide software-defined overlay of the radio access network as a purely software-defined overlay on that infrastructure running together with ai, generative AI applications deployed to end users in the form of personal assistance on your phone or in the form of providing AI applications into headsets or glasses. And that's actually not far off now, but once it seemed futuristic, but the technology pieces all now exist to make that happen, and we're working with our ecosystem to enable that over the course of, I would say, this year and next year.

Clarence Reynolds, TelecomTV (16:12):
And Manish, what do you see as the future for generative AI in telecom?

Manish Singh, Dell Technologies (16:17):
Yeah, I mean, I agree most of what Ronnie just laid out, I mean, clearly the rise of sovereign AI is here. Again, bringing that local language culture context into the mix, and telcos in many, many markets are very uniquely positioned to deliver these capabilities. Edge is the priceless prime real estate that the telcos have because most of the data is going to get created in the real world edge, gives them a opportunity to bring AI to the data closer to where the data is getting generated. So clearly we see that. I also will say the adoption in network automation and on customer care is going to grow. And what I mean by that is here and now we see still a lot more of the copilot kind of models where you have AI working as an assistant to a network engineer, to a field technician, and more driving productivity, reducing network downtime, improving KPIs and more. Where I see this moving going forward in a year or so is the rise of agents, right? I mean the agentic frameworks coming into the play, driving more complex autonomy in terms of planning reasoning, and then execution on very complex tasks. So I think the future looks very exciting, whether you're looking at it from a offering sovereign AI capabilities to really looking, driving more and more full autonomy in terms of your network planning, network lifecycle management, network operations, or enhancing customer experience.

Clarence Reynolds, TelecomTV (18:01):
Ronnie and Manish, thank you both for your insights today.

Manish Singh, Dell Technologies (18:04):
Thank you.

Ronnie Vasishta, NVIDIA (18:05):
Thank you.