Nokia’s James Watt talks scalable, simple, and secure networks at OFC50

Absolutely. Great big year. Big year for OFC, bigger for us. Great confluence of events. The acquisition, for the past several months we've been talking about the whys and the hows and the whens. We're past that now.

Yeah, sure. Let me start with the challenges. There's a lot going on right now in optical networking. The big challenges the customer has, on one hand, is capacity — different kinds, different forms, different drivers, but a lot of it is capacity challenges. Power is another one. We hear a lot about that in the data center space, but in general it's just a challenge as speeds get higher and higher. And the other thing, we've been talking about this for the last few years — we've been edging up on Shannon's limit, and really we can now easily fill a fiber, and that makes for a bit of a change as well in how you want to think about building networks. If you take all those and all the other stuff customers are dealing with, we boil it down to three pieces: you need to build networks that are scalable, simple, and secure. And we need to do that in a way that gives tailored solutions to different customer segments. The challenges the CSPs have are related to those hyperscalers have, but not the same. Similarly, there's some relationship between the challenges those two face and the mission-critical enterprises' challenges, but they're not the same either. Obviously, different parts of the network have different relative importance of things. A lot of this is getting bandwidth to the right place in the right way. So that's why we say scalable, simple, and secure networks — that's what we're about.

We've got a richer set — two things. We've got a richer set of end-to-end solutions today already. We're a month in and already the solution set is much richer. But also, by not having to do everything twice going forward, we can deliver more innovation that matters to the customers sooner — things that make a difference in their networks and how they deliver services. So we've got a better set of solutions today and we're going to be able to expand them more rapidly than we could on our own.

There's two dimensions of that excitement, right? There's the volume, because we see it and we talk about it, and there's also the technological excitement, which is always the risk in optical networking. But those technologies also play in different parts — how we end up with systems and how we deploy them in different customers or different parts of the network segment change, and that's always been the case. Now that multi-dimensional cube is just getting bigger, and there's more to remember, if you will.

The line system has to be there eventually — we'll fill it up with transponders, but it's got to be there. It's a good example of where all the pressures come. Capacity — we've been doing C++ for about 10 years, but we still need to find more. As I said, we can fill up fibers now, we need more. Now you start talking about multi-root or multi-rail, maybe further out, fiber types changing on us as well. And there's room for simplicity, as the different kinds of networks need different structures and being able to tailor the line system to give the right structure. Maybe very simple in a high-capacity, reasonably fixed purpose, more flexible when you're trying to distribute bandwidth in a metro or as a CSP. Being able to have the right different pieces of line — again, scaling up is clearly one of the big things, but also scaling down. We did a lot of work a couple years ago to get a four-port ROADM at the cost point of a FODM to give that flexibility down. At the same time, we're doing massive amounts of work to figure out how to get multi-rail in an operationally effective manner.

Yeah, pluggables. Again, another thing that's getting a lot of attention, including from us. If we back up a little bit, the other part of the equation, as you were saying earlier, is the transponders — coherent engines, get that bandwidth. Pluggables — we've taken what used to be 15 years ago a three-card set for 100 gig single-pair coherent. Now we can do eight times of that in an OSFP or QSFP-DD. So massive compaction of volume. Again, the core goal there is to get as much into that wavelength as we can, get the right spectral efficiency and the right power point. For us, it's a great way to deliver transponders, whether in a transport system — we've been doing that for a couple generations of the metro — whether directly in routers or otherwise. Having a massively scalable line only makes sense if you can get lots of bandwidth on it. That's where customers need the best transponders, whether embedded or pluggables. Right now, there's a big gap. We're busy filling it — the industry is filling it on the pluggable side, and we're going to have ours mid this year. We're seeing a lot of demand for what we could deliver.

800 gig.

The way I think about it is, what we call embedded is a packaging choice. If you need the extra space for a particular application or the extra power — because the pluggables are both space and power constrained — then you need to use it. If you can fit it in an existing package, it's even easier for a whole bunch of reasons, operationally for the customer, operationally for ourselves. The embedded versus pluggable is a bit of a packaging choice. If you look at the PSC-6S packaging, that's a proprietary large pluggable that we did to get the operational benefits while still having the power and space we needed. As we go forward, it's really a packaging choice. The core is performance and power, and that's what the customer needs. We keep focused on that, and then we do what we need to in packaging to make it as simple both for them and ourselves. Our focus overall is just on building great coherent engines, great transponders, and then packaging is a second-order term.

I wouldn't — no more embedded. Not at all.

Maybe I start with the second one. We've got to do a great job, as I was talking about earlier, for different customer segments and different network segments. On the other hand, if you look at our technology pool and our capabilities, this inside-the-data-center opportunity and the need to take power out of that part of the space is something we believe we can make a significant dent in. There are other places where we can probably apply our technologies, and this is a great example of where we can and where we will. It's a great complement as well to all the efforts we're making on the IP side on the data center fabric work. If you can think of it as that part of a bigger overall focus on the data center that's going to continue, that's something Nokia more broadly is very focused on.

I think that's where we start. You'll see the working die packaged into different applications on the booth today, and that's where we start. That's where we make our contribution to taking power out of the data center, or reallocating it back to the GPUs. We've got work to do to do that, and after that, we'll see what we learn. We do have a habit of, once we get into a space, expanding. Even in this case, think about the data center fabric as part of that bigger approach to the data center.

That's a great question, Sterling. It's a great example — where did that come from? The answer is our customers. We had customers, again, we have a rich customer base as I said earlier, we had customers actually in the mission-critical enterprise space who had challenges they needed solved. Our first work was to solve them. Then you step back and say, they're not the only ones with these problems — they're just the canary in the coal mine. So we've kept our focus on making sure we can do what they need, making sure that it is simple to operate, but also can expand over time as the technology shifts. As you mentioned, QKD — there are a lot of players, a lot of rapid innovation there. Our focus has been: don't preclude that for our customers, allow them to have it, but don't require it. Allow them to get started today. This is really about making sure our innovation matters, taking a customer need, our capabilities, and then continue to do that. We've done this for multiple generations of transponders and we'll continue, because, as you said, it's an ever more important need. PQC is coming, but if you think about rolling out anything across even a small enterprise, the time to get that rolled out is a lot less than standing up encryption on a set of links. That's how we think about it.

Absolutely. It's an important priority for many customers.

Great catching up. We're one month in, so there's a lot more to come. Again, it should be a great show. Hope you enjoy it.