Artificial intelligence is rapidly reshaping demand for network infrastructure, creating new revenue opportunities for carriers that know where to invest. According to Bill Long, chief product officer at Zayo, hyperscalers are driving near-term growth, particularly along long-haul routes that connect emerging data centers to power and compute resources. As AI training clusters expand into new geographies, connectivity becomes the critical enabler - without it, even the most advanced data center cannot deliver value. However, Long cautions carriers against chasing projects that rely on a single customer or overly optimistic financial models, emphasizing the importance of experience, scale and diversified investment strategies.
Looking ahead, Long sees significant upside in metro fiber as AI inference moves closer to where data is generated and used. Locations that once required minimal connectivity, such as warehouses, factories, hospitals and schools, now demand high-capacity, resilient networks to support robotics, automation and AI-driven applications. To succeed, carriers must go beyond simply providing physical infrastructure. Long argues that winning in the AI era requires delivering networks at software speed, with digital tools that allow customers to design, order and provision services seamlessly. As enterprises grow accustomed to cloud-like experiences, carriers that combine deep fiber assets with strong software capabilities will be best positioned to capture the next wave of AI-driven demand.
Mitch Wagner:
Where are carriers seeing the biggest near term revenue opportunities? Enterprises, hyperscalers, consumers?
Bill Long:
Speaking specifically for Zayo, we're seeing it with hyperscalers, we're seeing it particularly in the long haul network. So these data centers are popping up where there's available power that happen to be adjacent to our long haul network. If you build a data center in the middle of nowhere, if you don't connect it, then it's just a big expensive refrigerator. So we're seeing a lot of demand for long haul. But there are a couple words of wisdom for carriers that I would actually have. We're seeing what I'll call spreadsheet competitors out there who are able to draw a line on a map and get some PE money and think they can get in the long haul business. So a couple of words of wisdom, if you don't mind.
So first, beware of the bridge to nowhere. If you're trying to do a long haul route underwritten by a single customer, you're not going to be able to ... The economics will not pencil in. The second is expect the unknown unknowns. So whether that's a vendor going bankrupt, permitting that you don't know is out there, rock where you weren't expecting rock. Expect the unknown unknowns. Sock puppets don't make good customers, so we're not pets.com. Back in the day, when you look at the AI bubble, there are some customers out there that look not exactly like pets.com, but they look like they rhyme a little bit with it. So you got to be careful on who the customers you're underwriting. And the last sort of word of wisdom is build like you invest. You'll win at the portfolio level. If you're betting on a single route to be a home run, some routes are winners, some routes are losers. If you don't have the appetite to invest at the portfolio level, you're not going to win. So that's sort of where we're seeing it in my sort of quick words of wisdom.
Mitch Wagner:
Excellent words of wisdom they are. So talk a little bit about what you mean by spreadsheet competitors.
Bill Long:
I think people who haven't been in the long haul business think that if you can just draw a line on a map and you can go get some PE money and you can do a spreadsheet that says, "Oh great, we all get rich," that that's good enough to get into the business. And that's what we've seen, is folks who really have no credibility in the long haul business. This is a business that you can only get good at at the school of hard knocks. It's one of those businesses they say about like horse racing and airlines. The best way to make a billion dollars is to start with 10. So this is one of those businesses that the only way to be good at it is to have a long history of doing it. And so that's what I mean by spreadsheet competitors. You can put a spreadsheet together that'll tell you there's a great way to get rich. If you don't have the bruises and the sort of scars from being good at this business, I would stay away.
Mitch Wagner:
Let's talk a little bit more about the sectors for the revenue opportunities. When you're looking at enterprises versus hyperscalers versus consumers, which I guess isn't an issue as much for you guys, but how should carriers prioritize those segments differently?
Bill Long:
There's a little bit of a timeline here where of course, as AI is evolving, the first things that are popping up are these training data centers that's then leading to inference and being able to connect to those inference data centers is great, but I think the real opportunity is in the metro network as you get to inferencing. So if you think about where AI is going to be applied, whether that's in factories or a fulfillment center, or a hospital, or a school, locations that used to be, you would never think of as needing high bandwidth. We have a example, a customer who has fulfillment warehouses where they have in each fulfillment center, they have over $100 million of robotics and AI automation in that. They require diverse 100 gig connectivity into each of those fulfillment centers. You never would've thought that going into a fulfillment center would be a high bandwidth location five, 10 years ago. It's a big opportunity now. So I think there's a big opportunity for high bandwidth connectivity requiring fiber in the metro as enterprises really start to embrace AI more holistically.
Mitch Wagner:
So how do you see carriers' roles in the AI value chain evolving given that most telco leaders still identify primarily as infrastructure providers?
Bill Long:
I think their role obviously is at the fiscal infrastructure level, and that's the fiber that I talked about. But the folks are going to win who can figure out how to make infrastructure available at software speed. The only way you do that is with deep sort of software expertise. I think that if you're not operating a business that's big enough to have a significant software investment, which is really NAS. And when I say infrastructure at software speed, I mean having a great software experience where you can find, design, order, provision, deliver your network experience for enterprise customers all with a software driven experience. So if you don't have, again, this is one of those words of maybe cautionary tales that if you're not willing to get really good at software, that it's going to be hard to win because over time infrastructure is going to be expected to be delivered at software speed.
Enterprises have been acclimated. It started with the cloud, now it's going into networking. And if you're not ready to make a sizable investment in making your infrastructure available at software speed, it's going to be either a hard road to hoe or you just focus on infrastructure. Just sell the fiber and let other folks who are making the software investment help you monetize that physical network infrastructure that you have.
Mitch Wagner:
We did a massive survey of 500 communication service provider leaders worldwide a short time ago for an upcoming report on Telco AI. Nearly a third of those executives said the biggest barrier to AI growth is making the business case for AI infrastructure. How should they approach that challenge?
Bill Long:
That's shocking to me. I think if your addressable market and the thing you're trying to unlock is literally human cognition, you're not bounded by addressable market. And if you look at the smartest companies in the world, the hyperscalers, your Googles, your Amazons, your Metas, your others, literally the smartest people in the world are finding trillions of dollars to invest in AI infrastructure. They might be onto something.
Mitch Wagner:
Good. Thank you for your time and insight, Bill.
Bill Long:
Happy to do it. Thanks, Mitch.