AI is fueling an unprecedented expansion of data centers worldwide, and with it, the demand for high-capacity, low-latency optical networking. In this Fierce Network conversation, Daniel Frankel sits down with Mark Bieberich, VP of Portfolio Marketing at Ciena, to explore how the company is enabling hyperscalers, neoscalers and data center operators to scale for the AI era.
Bieberich outlines the challenges network operators face—from massive scalability requirements to limited in-house optical expertise—and explains how Ciena’s innovations in coherent optics, photonic line systems, and power-efficient transmission are helping customers overcome these barriers. With technologies like WaveLogic 6 Extreme delivering 1.6 Tbps per wavelength and next-generation pluggables driving metro and regional connectivity, Ciena is setting the pace for AI-optimized networks.
For telecom leaders navigating rapid growth in data traffic, this discussion offers a clear view of where the industry is heading—and how to stay ahead.
Daniel Frankel:
Hello, I'm Daniel Frankel, regular contributor to Fierce Network and CEO and co-founder of Next TMT. I'm very pleased today to welcome Mark Bieberich, who is the vice president of portfolio marketing for Ciena, a connectivity company that's needed in an era where data centers are popping up all over the place and massive compute computational horsepower is being established to underwrite a very data rich AI future. Mark, thanks for joining us.
Mark Bieberich:
Thanks for having me, Daniel.
Daniel Frankel:
So maybe before we launch into it, you can tell me a little bit about Ciena, you're globally situated, but Hanover, Maryland based, where are you based and what do you do for them? What does the company do?
Mark Bieberich:
I'm based near our corporate headquarters in Hanover, Maryland. My role is really to help our salespeople and our marketing teams position our portfolio of products and services through a wide variety of messaging and collateral content that we use to engage with our customers and many other stakeholders, both internal and external.
Daniel Frankel:
So you're situated against a growing ecosystem of cloud and AI technology companies, which are driving demand for optical networking, busy time, obviously. Maybe you can talk about that paradigm.
Mark Bieberich:
Obviously, over the last few years we've seen massive demand for AI applications and that has fueled a build out of optical networks, I would say first and foremost, among the top hyperscalers in the world. That's Amazon, Microsoft Meta, Google, Oracle. But now what we're seeing is the emergence of many other companies that we call neoscalers, but these are companies that are building out AI infrastructure services like GPU as a service. We also see a growing number of large language model developers that are also in need of optical infrastructure or high capacity optical services.
And there are also companies, for example, that are co-location providers, data center operators that really provide the real estate that where all of this can happen. And we are also seeing tremendous demand from those type of companies as well. So whether the company is a hyperscaler, a neoscaler globally, we see a rapid expansion of the number of data centers that are being built to host AI applications, deliver AI applications, train AI models, and provide them to end users at inference sites. And we're seeing this expansion fueled by tremendous investment from public equity, from private investors, from a variety of public investment sources as well. So that investment is giving rise to this rapid expansion of optical infrastructure that underpins all of the connectivity that's required to deliver these applications as that demand grows.
Daniel Frankel:
So you have a growing number of constituents who are trying to establish data centers and serve a growing customer base. What are the challenges they face as they build out these networks and try to establish this connectivity?
Mark Bieberich:
The first real challenge that we see, and that's really at the heart of our business at Ciena, is just the need for scale. I mean, there's just a massive amount of scalability required to deliver AI applications when and where they're needed. We see not just that demand for applications growing today, but we see optical builds that are coming together that are in anticipation also of future demand, and there's really no sign of any slowing down in that future demand. So the first main challenge that we see is just the scalability that's required to build these networks. I also think that for many, for example of our neoscalar customers, they don't have as much experience in the way of building optical networks as some of the companies that have been around a much longer time have, they don't have deep staff and expertise required to build, design, operationalize and maintain these networks.
So one of the challenges we see is just having the know-how to go and build the networks that are supporting their plans and that are really kind of the foundation of their business. So we're in a position, of course, to lend our expertise after many years of supplying the build out of hyperscaler networks to this growing ecosystem. I think also there are particular performance requirements in these networks that remain a challenge for many of the operators. I think as companies build out, for example, more and more inference sites, as we see the optical infrastructure need to cover a more distributed footprint, we see for example the need for low-latency connectivity to inference sites. We see the need for increasing security, for example, across submarine links and across many of the connections to training facilities. And we see increasing demand, for example, for more deterministic behavior to inference sites.
So you might think of, for example, even something as really readily apparent, especially on the West Coast now, is driverless cars. There's a tremendous amount of data that goes into the operation of driverless cars and a lot of AI technology is now being used to drive efficiency in those fleets. And we see that there's really no tolerance for things like packet loss or for higher degrees of latency. So we think that just the overall performance of these AI-driven networks remains a challenge for quite a few of these companies. So whether it's scale or whether it's the know-how or whether it's just boosting performance and security on these networks, the challenges are quite significant. We're really in kind of in I think the early phases of this expansion of the data center ecosystem. And whenever you're in those early stages, a lot of know-how new innovation is coming to the marketplace. And at Ciena, I think we're driving much of that innovation.
Daniel Frankel:
Right. You have to meet the demands of today, which is pretty challenging, but a lot of innovation is required to meet future demand. Maybe you could talk a little bit about that. What should we expect to see?
Mark Bieberich:
Yeah, so if you just think about the optical networks that are underpinning a lot of this expansion, I think first we see innovation in the performance optics that maximize the overall capacity per wavelength on these networks. So by performance optics, I mean, for example, our WaveLogic 6 Extreme technology, and what that does is it pushes 1.6 terabits per second per wavelength on a fiber, and that's an industry-leading performance mark. Why that's important is because especially for some of our neoscaler customers, like I said before, they don't have a tremendous amount of fiber, so they need to maximize the capacity per fiber pair that they have. And with a solution like WaveLogic 6 Extreme, they have the ability to not only push higher capacities, but do so in a highly spectrally efficient way. So we've seen a lot of take up of that particular technology. We also have a wide variety of pluggable coherent optics that our customers are implementing, especially over distances.
Again, metro networks, whenever they have to connect two sites across a metro footprint or maybe a regional footprint whenever they're connecting, for example, inference sites to training sites or let's say an enterprise site to their data center where they're serving up GPU as a service, pluggable optics is a viable solution and the right solution when space and cost factors into the equation. And we see tremendous take up of our WaveLogic 6 Nano pluggables for that purpose. I think over the next few years we're going to see a lot of innovation in the underlying components technologies as well as the pluggable form factors that drive a lot of this performance that's required and to enable all of the scalability that I was mentioning earlier. So coherent optics, whether it's performance or plugs, I think is a really important area of innovation. And then you have the photonic layer where the line systems reside.
And over the last few years, where Ciena has really had a lot of success is in deploying the line systems for a large number of hyperscaler customers. And through that experience, we've had great feedback from customers around what kind of capabilities they require to reduce footprint, to maximize scale, to handle things like multi-rail applications. What I mean by that is we have a growing number of customers that have many fiber pairs coming into and out of a particular physical location, what we might call a hut, and that's driven the innovation around the line system to support multiple rails, to be able to scale up in a highly space-efficient form factor.
I think also, and on the topic of space and power, that also of course remains a challenge. We've seen that first and foremost, and the hyperscalers as AI has grown so fast, some of these data centers really need to reduce the overall power consumption that they're seeing. So one of the areas of innovation, I believe where Ciena has been in the lead is in developing optical solutions that drastically reduce the power consumption required when transmitting all of this traffic over optical wavelengths. So all of these areas are key for building out these AI-optimized networks. And Ciena has really been in the lead in all phases of it.
Daniel Frankel:
We could go really deep on power consumption alone, but just winding this down, if somebody wanted to dive deeply with you and Ciena, call to action here, how do we get in touch with you? How would a hyperscaler or junior hyperscaler as the case might be, talk to Ciena and explore the future?
Mark Bieberich:
It's fairly simple to go on to ciena.com and engage with us directly on our website. There are a variety of different ways to connect with Ciena, just through ciena.com and on ciena.com are pages dedicated to all of the solutions that I just mentioned and more that are highly relevant to the growing ecosystem of data center operators, hyperscalers and neoscalers. But we also of course have a large dedicated account team that is trained and ready to serve this particular space with particular subject matter expertise and resources required to meet the needs of these customers.
So it's fairly straightforward to engage with Ciena. Would also just say this is an area that is of increasing importance to our business. Ciena is producing a lot of content, thought leadership material that's really designed to help educate this group of customers on where the technologies are going, what our innovation is all about, how we can help them, and most importantly, how we partner with them to address these particular challenges that I mentioned in the beginning of our call. That's where we really are focused, is in helping provide the level of expertise that helps them operationalize and grow their business.
Daniel Frankel:
Well, we have a ton of ground to cover over the next decade, so I am guessing, Mark, I'll be talking to you again. I want to thank you for chatting with us today. I'm Dan Frankel, Fierce Network on behalf of our [inaudible 00:15:09] wishing you a good day and we'll talk again soon. Thanks.
Mark Bieberich:
Sounds good. Thank you, Daniel.