Great to see you, Justin. How are you?
Nice to meet you, Steve. I'm doing well.
Yeah. I mean, there's a lot of energy around AI and things like data centers and data center factories and places where we're going to do the training. That's obviously great business for Nokia. Are you worried that your company may become overexposed to those markets which are fantastic now, but what happens if the super cycle crashes, as it were?
Yeah, look. I think first of all, I don't see the super cycle crashing. Will growth slow at some point? I mean, of course, right? We just know that the math of physics and all of this. It's also, you know, but I think a few things. First of all, when you look at it, you know, the demand has actually been constrained from the outset of the ChatGPT moment. You know, first it was GPUs, then it was energy, now it's memory. There's been pressure.
Water. Yeah, exactly. And you know, and we're and when you look at the industry, I mean, we're in long lead times for silicon still. Um and so, you know, there's a bunch of constraints that I think are actually tempering the demand. The other thing is, when you look at the...
Do you think there's more to give?
Yeah, I think there's more to give. And I also think when you look at underlying usage, the reality is the usage is sticking. Now, does that mean that we're not going to have winners and losers? You know, we're going to have some consolidation that always happens in these markets. Right. Will growth slow? Yes, but I think we end up in a much bigger market. And that's part of that is because AI becomes the dominant workload in networks, in data centers. Right.
And ultimately, you know, drives that will drive the innovation cycle, drive a lot of the demand cycle. And the new growth, it's almost like there's a lot of growth coming from, if I can even use that term, a traditional hyperscaler that haven't been around that long, but you know, the well-known established hyperscalers, there seems to be a lot of stuff coming from the neo cloud players as well. A lot of investment going into that. Do you see that as well?
Yeah, we see it on both sides. Absolutely, hyperscalers and neo clouds. All of that, you know, today driven by either models, you know, foundation models or applications, users of those models driving the demand. What's interesting though, when I, you know, I think about this, you know, quite a bit is we really only have scratched the surface. So, we have a tremendous amount of data or you know, trillion token, 100 trillion tokens a month. Um you know, 77 exabytes a month of data. But, we've really only scratched the surface because most of what we're doing is chatbot engagement or a little bit of agentic. Yeah. We haven't really touched the massive scaling of, you know, machine-to-machine autonomous agents, you know, at scale. We haven't touched physical AI in any meaningful way. Yeah. And that's why I think long-term there's so much runway and impact coming.
We're at a telecom show. I mean, largest telecom show in the world. Uh what role do carriers play in this future that you're describing? Because there is a real danger they're going to get in some ways they could get disintermediated, right? Certainly into data center or you know, AI factory, that's already happening, right? So, so where do they play? What's their value in this new world?
Well, I think today they own a very valuable asset, which is the access network. And you know, our research suggests that over, you know, 50% of the traffic right now in AI is being generated on mobile. So, it's in their networks today. Um the question is how they capitalize on it. This is why, you know, at this show and you know, our big conversation is around designing AI native networks. Not, you know, recognizing that first of all, you can't just transform these networks overnight. But, if they're already carrying the majority of the traffic, as that traffic inevitably grows and right and moves into the physical world, they're either going to have to innovate or they're going to be disintermediated, right? And I think there's many of them, you know, some of the customers that we're partnered with on AI RAN, for example, that are really thinking about how do we innovate and take advantage of this opportunity.
Yeah, AI RAN is that's the glue, right? I mean, that's the thing which brings these two universes, compute and carrier, together. And that's why you're putting I mean, I don't mean to put words in your mouth, but it seems to be that that's why you're putting so much emphasis on it at this event. Is that right?
Yeah, I mean, AI RAN is a great example of what building an AI native network means because you have to do it at every layer of the stack. You know, there's a lot of talk about let's just layer intelligence on top of the network, but keep the old stacks. Well, then I'm not building the kind of dynamic deterministic fabric that I need to deliver tokens, right? And one of our customers, John Saw, you know, coined the term kinetic tokens, this idea that, you know, a token, you know, its existence matters temporally, right? It's going to have It's got movement as an attribute. And that's a part of the shift is that the network is also going to have to be a compute engine. You know, latency obviously matters, but it's all about delivering that token at the right time with the right trust, the right security profile, and obviously the right network capacity. And that's a very different dynamic for a network, which is why it's a different stack. You can't just put it on top.
Different stack, really. And that seems to be where you're really focused at the moment at this event. You've talked about cross-domain innovation. That can be technology. It can also be culture. I mean, such huge changes at Nokia since you arrived. How long have you been there now?
Uh this is 11 month 11, on month 12. Just starting month 12.
Okay. I mean, really, it's a very different company now. I mean, is part of what you're trying to inspire there really like a hyperscale model? Not in terms of the technology and cloud, but in terms of a more horizontal approach to excellence and surfacing new ideas. And is that part of your goal?
