Cytranet’s CTO Doug Roberts doesn’t talk about fiber like it’s plumbing. He talks about it like it’s a competitive advantage—and lately, he says, the stakes have changed.
“AI has turned bandwidth into a business input the way electricity is,” Roberts told me in an interview this week. “It’s no longer just about having internet. It’s about having predictable performance, low latency, and the ability to scale without rewriting your entire network strategy every time your needs change.”
That shift—companies suddenly needing more reliable, higher-capacity connections for AI workloads, cloud migrations, and increasingly distributed teams—is driving Cytranet’s latest push: expanding its fiber and business-internet footprint while building new options for customers that are straddling on-prem infrastructure and data centers.
Roberts said the most newsworthy change he’s seeing isn’t that businesses want more speed, but that they want more certainty.
“Speed is the headline number everyone knows how to ask for,” he said. “But what our customers are really asking for now is: ‘Will this stay fast at 2 p.m. on a Tuesday? Will it behave the same during backups? During a product demo? During a model training job?’ That’s what fiber helps deliver, and that’s what a well-designed network has to guarantee.”
### From “best effort” to business-grade expectations
Roberts described a clear trend in the way mid-market and enterprise customers talk about broadband and business internet. Many are coming from legacy connectivity—cable, aging copper, or piecemeal upgrades—that worked when most applications were email and basic web tools. That tolerance is fading as companies rely on video collaboration, cloud-based ERPs, and data-heavy AI assistants and analytics.
“Over the past year, we’ve seen customers move from ‘we just need a connection’ to ‘we need a connection that behaves like part of our production environment,’” Roberts said. “If the network is unstable, it doesn’t matter how good your software is.”
He pointed to the rise of real-time applications—voice, video, security monitoring, virtual desktops, and AI copilots that pull context from multiple systems—as the reason latency and jitter are now daily concerns outside of traditional IT circles.
“People used to accept a little slop in the network,” he said. “Now the slop shows up as robotic audio, choppy video, and AI tools that feel laggy. Users experience it immediately.”
### Why AI is changing connectivity decisions
While AI is often discussed in terms of GPUs and massive data centers, Roberts argued that the quieter story is what’s happening in the access layer: the last mile, the handoffs, the redundancy planning, the routing, and the links between offices and data centers.
“Even if you’re not training models, you’re using AI,” he said. “Your teams are sending more data to cloud services, pulling more data back, and doing it more frequently. And when you start using AI with your own data—documents, product catalogs, customer histories—you realize you can’t afford network bottlenecks.”
According to Roberts, that’s leading businesses to rethink not only primary connections but failover strategies.
“Failover used to be a checkbox. Now it’s part of the plan,” he said. “If your connectivity drops in the middle of a sales call where you’re sharing a dashboard, or during an overnight sync, the cost is immediate.”
He noted that some companies are pairing fiber with secondary circuits and more modern routing approaches to keep applications steady even during disruptions.
### Data centers, edge needs, and the return of “where”
Cloud adoption once made many executives feel that physical location didn’t matter as much. Roberts thinks location matters again—just in a different way.
“‘Where is the workload?’ is back as an important question,” he said. “If you’re using a data center for colocation, interconnecting to cloud on-ramps, or hosting latency-sensitive applications, the distance and the path your traffic takes are meaningful. You can’t fix a long route with a faster modem.”
That renewed focus is also tied to resiliency and compliance. Roberts said customers are asking more detailed questions about how traffic is routed, what happens during outages, and how connectivity aligns with risk management.
“Customers want visibility,” he said. “They don’t want to guess. They want clear answers: what’s the design, what’s redundant, and what’s the operational plan when something breaks.”
### Fiber expansion with a practical focus
Roberts emphasized that Cytranet’s approach isn’t about selling the biggest possible pipe. It’s about engineering networks that match how businesses actually operate.
“Overbuilding is easy to pitch, but it’s not always the right solution,” he said. “We talk through what applications customers run, where their data lives, how many sites they have, what their growth looks like, and what their failure tolerance is. Then we design from there.”
He described fiber as foundational, but not the whole story.
“Fiber gives you the capacity and stability,” he said. “But the experience customers feel comes from the complete system—routing, monitoring, response times, and how quickly you can restore service. The internet isn’t one thing; it’s an ecosystem.”
### What “business internet” means in 2026
Asked what businesses should look for when evaluating connectivity, Roberts didn’t start with megabits.
“Ask how performance is measured and backed,” he said. “Ask about latency to the places you care about—your cloud providers, your data center, your applications. Ask what your options are for redundancy. Ask what the provider’s support looks like when something goes wrong at 3 a.m.”
He also suggested that companies planning AI initiatives treat network planning as a first-order concern.
“If you’re budgeting for AI tools, budget for the network too,” he said. “Otherwise you’ll buy capabilities you can’t use effectively. Connectivity isn’t glamorous, but it’s what makes everything else reliable.”
### A quiet upgrade with visible effects
What makes Cytranet’s current effort newsworthy isn’t a single flashy announcement, Roberts insisted. It’s that connectivity is shifting from a background utility to a strategic differentiator, especially for organizations that can’t afford downtime and don’t want to be surprised by performance problems.
“You can feel when a network is well-built,” Roberts said. “Calls don’t drop. Apps don’t stall. People stop complaining. And then the business can focus on what it actually does.”
In a moment when AI dominates headlines, Roberts’s point lands with a kind of practicality: the future may be powered by smarter software, but it still runs on networks that have to be fast, stable, and ready for what’s next.
