Ray Le Maistre, TelecomTV (00:05):
So I'm here with John Harrington, he's senior VP of Nokia Europe. John, thanks very much for joining us today. Great to see you. And I understand that you've got quite a big announcement with the focus on AI RAN. Can you tell us all about that please?

John Harrington, Nokia (00:20):
Yeah, absolutely. It's the world's first AI native RAN platform that we're announcing, and it builds on something that we announced last year with our work with NVIDIA. And I think what's really exciting are a few key points. So first, it's going to dramatically improve the spectral efficiency. We're giving a roadmap of one and a half times ultimately up to 2X the spectral efficiency. And spectrum is really, really important to service providers as you know. Two, it's going to be open. And so it's going to be O - RAN compatible. And the third thing is what we're doing is abstracting what Nokia does into a software layer. So we're hardware abstracting if you think about it that way. And actually what we're going to have are three platforms. We're going to have the ability to put the GPU within the Nokia Air Scale platform, which we've deployed millions of units around the world.

(01:14):
The second is that we'll have an AI RAN accelerator, so a more purpose-built AI RAN product. And third, we'll be able to do this on COTS servers as well. So be able to put this inside edge nodes and data centers. And so fit for purpose for those three things. And it really will give a huge leap in performance and it won't come at the cost of power. We're also saying that we'll be meeting similar power envelopes as well.

Ray Le Maistre, TelecomTV (01:42):
So this comes sort of pretty soon after the greater NVIDIA involvement with Nokia. How much of this is based on what mobile operators are saying that they need from their future infrastructure?

John Harrington, Nokia (01:59):
We like to think about it as AI on RAN and AI for RAN. So AI for RAN means, okay, how can we make the product more spectrally efficient? How can we improve the performance of the product? And everyone's looking to do more with their radio portfolio. And that's something that we're delivering because of now we're working in software, we're improving our coding, we're looking at the algorithms. It's really dramatically improving the performance. And probably the key KPI of that is spectral efficiency. And secondly, AI on RAN, we now have a GPU at a radio based station. There's all sorts of services and applications that we could potentially think of starting with inferencing of course as we start to grow out the AI grid. Secondly, there are obvious things like sensing, other applications that we can use on the radio towers as well. So I think it's solving two things.

(02:52):
One, the performance and the efficiency that we're bringing. And two, it's potentially opening up new revenue, new applications, new services for service providers.

Ray Le Maistre, TelecomTV (03:02):
And what kind of timeframe are we talking about here in terms of when this might come to market?

John Harrington, Nokia (03:08):
Well, we've done a lot. We announced this last year within Nvidia. We've been doing some trials, some proof of concepts. We'll be going into the field at the end of the year. And so into 2027 is what we're talking about. It's very real. We're absolutely part of the AI RAN Alliance. We're working with leading operators like SoftBank in Japan, T-Mobile USA, and IOH in Indonesia. So it's real. I think people were not expecting it to come as quickly as this, but we're now moving forward very, very quickly with this.

Ray Le Maistre, TelecomTV (03:40):
So obviously in the few years that AI RAN has been a major topic in the industry, there has been some cynicism, some pushback from mobile operators as well as market commentators and analysts. What are the proof points here with what you're bringing to market and what will this mean for mobile operators? Because spectral efficiency, that's like a real massive target for all operators and this is something they're all going to be interested in.

John Harrington, Nokia (04:08):
Yeah. I think the spectral efficiency is probably the key thing. Service providers pay a lot of money for spectrum. They're probably one of their most valuable resources. So how can we squeeze more capacity out of that spectrum? That's the first thing. Secondly, power. People will assume, "Oh, this is going to take a lot more power." No, we're trying to fit within the existing envelope of our air scale platform and have very much comparable power consumption. So you'll have an efficient power powered machine, much better spectral efficiency.

Ray Le Maistre, TelecomTV (04:40):
Well, it seems like that Nokia and Nvidia are addressing some of the key requirements of the market here because it was always cost and power. There were some of the pushbacks from the operators and I'm sure you're going to get plenty of interest from operators wanting to check it out and see if it meets their requirements. So John, thanks very much for telling us about this development today and look forward to seeing where it takes the market and Nokia's involvement with the mobile operators. So thank you very much.

John Harrington, Nokia (05:10):
Thank you very much. I