Clarence Reynolds, TelecomTV (00:03):
The telecommunications industry is witnessing a revolution as AI transform network operations and infrastructure. The strategic partnership between HPE and NVIDIA is at the forefront of this change, creating new possibilities for operators. And joining us today to discuss these innovations is Chris Pinrose, global Vice President of Business Development for Telco at Nvidia. Chris, thank you very much for being with us today.

Chris Penrose, NVIDIA (00:28):
Great to be here.

Clarence Reynolds, TelecomTV (00:29):
So as it relates to AI ran, how is the Nvidia HPE collaboration transforming network infrastructure and what key benefits are operators starting to see?

Chris Penrose, NVIDIA (00:39):
So from a network infrastructure perspective, we're really seeing this move toward what we call AI ran and this opportunity to put the radio access network and AI on the exact same infrastructure. So this gives massive benefits to its telco to be able to fully utilize and get, instead of flying the plane path full, you're now able to use it for telco when you need to, but use it for AI when you don't need it for telco to get full utilization of that compute infrastructure. So it is a great basically TCO for the telcos, and then also they're able to take that same infrastructure and apply it to the radio access network to make the network more performance. So we think it's an amazing opportunity to collaborate as

Clarence Reynolds, TelecomTV (01:25):
5G evolves. How do you envision AI ran technology advancing through this

Chris Penrose, NVIDIA (01:29):
Partnership? So we're really focused on making sure that we can apply the technology of AI into the network. So think about things like making it more energy efficient, more spectrally efficient, and ultimately being able to do better root cause analysis. So make a better performing network with ai. At the same time, we're able to have that shared infrastructure so we have the best way to get maximum utilization out of it, and then we can layer all these new AI applications, how do we make those perform really well on top of the network? And so assess all three of those things are really coming together to help transform and really shape where we're going with the next generation wireless.

Clarence Reynolds, TelecomTV (02:07):
So as it relates to scale, how does combining h HP's computing infrastructure with NVIDIA's AI capabilities help service providers scale AI workloads?

Chris Penrose, NVIDIA (02:17):
Yeah, so I think HP has done some really great work in being able to build out a reference architecture for how to build large scale compute and working closely with nvidia, we're designing not only from a chip perspective and our GPUs, but also the CPUs and our duss, how do we tie all those together with the right networking and the right software to do AI at scale? And so when you are putting all these pieces together, you really get the ultimate token producing factory and getting the best return on the investment of building out these large compute clusters. So that's a lot of the great work we're doing together. How does the private cloud solution accelerate production AI deployments? So with private cloud, a lot of telcos are looking today to say, how do they apply AI and generative AI to their own operations? And what we really recommend to most of the telcos is they just set up an AI center of excellence to be able to really become experts at applying AI and working closely with HPE and their private cloud solution. We're able to roll in with an infrastructure that enables their data scientists to begin to start working on all the different AI opportunities inside that business and making them way more cost efficient and way more productive. So with

Clarence Reynolds, TelecomTV (03:36):
This platform supporting diverse applications, which use cases are you seeing the greatest impact?

Chris Penrose, NVIDIA (03:42):
Yeah, so what we've done is we're seeing a wide use across the telcos, but the most popular applications are customer experience. And how do we transform that customer experience for the end consumer and support the representatives out there taking those calls. The second place is how do we apply it into the network operations itself? And so how do we help those technicians be more effective? How do we do better network planning? So all sorts of ways in which we can drive improved productivity and network operations. And then the third place is just actually making all the employees more effective and efficient in the roles that they do. By taking the corpus of data that every telco has on how they do methods and procedures and operate, and putting an AI front end on that, it really is transforming getting the information of what they need to do to their fingertips much faster.

Clarence Reynolds, TelecomTV (04:28):
So how does sovereign AI enable telecoms to function as regional AI factories, and what

Chris Penrose, NVIDIA (04:33):
Challenges does it address? So with sovereign ai, you're hearing a lot of interest in how do we create the right AI infrastructure for our nation to be able to support this big wave around generative ai And telcos are an unique spot. They're a trusted partner to the governments and the countries in which they operate, and they also have access to data center space and power that you need in order to begin to stand up this infrastructure. So what we're really seeing is that this is a new monetization opportunity for telcos where they can begin to bring the infrastructure that's vital to the nation that can help the startup community. They can help researchers, they can help government businesses and consumers be able to bring AI services into the market and be able to provide the latest and greatest computing infrastructure between HP and NVIDIA to make that happen.

Clarence Reynolds, TelecomTV (05:24):
How does your infrastructure support multi-tenant AI stacks?

Chris Penrose, NVIDIA (05:28):
So we're working very closely with HPE to make sure that we can not only support a single tenant architecture, but also multi-tenant architecture. And really we're down addressing that at kind of this different software packages we would layer on top. So we can serve people buying maybe a small cluster or buying as many as hundreds, thousands of GPUs. We want to be able to make sure that we can get the optimal utilization of that infrastructure that's been put in place. Chris, thank you for joining us today. Thanks for having me. It's great.