Doug Roberts has a simple way of describing what’s happening to business networks right now: the traffic pattern has changed, and companies are finally seeing it.
Roberts, the CTO of Cytranet, says the AI boom is creating a new kind of broadband demand—less about occasional big downloads and more about constant, heavy, two-way data movement. That change is showing up in unexpected places: an accounting firm that now processes documents with AI tools, a logistics operation streaming warehouse video into analytics, a healthcare group leaning on speech-to-text and automated workflows, or a multi-site business that suddenly relies on real-time dashboards to make decisions.
“AI is a multiplier,” Roberts said in an interview. “Even when you’re not training models, you’re generating more data, sending more data, and relying on cloud compute more often. The network becomes a production system.”
This shift is one reason Cytranet has been expanding business-only fiber connectivity and strengthening high-capacity routes that tie offices and industrial zones to compute hubs. Roberts believes that as AI becomes ordinary business tooling, companies will judge broadband less like a subscription and more like infrastructure they build their operations on.
From “Internet Service” to Operational Capacity
Roberts says the most telling change isn’t in the tools businesses are adopting—it’s in the questions they’re asking.
Two years ago, many connectivity conversations started with price and advertised speed. Now, he says, businesses are asking about performance under load, upstream capacity, and how connectivity will behave when multiple critical systems are running at once.
“They’re not trying to win a speed-test contest,” he said. “They’re trying to avoid the Monday-morning slowdown, the unreliable upload, the voice system that degrades when the network is busy.”
He calls it the difference between a connection that looks good in isolation and a connection that holds up when the business is doing real work.
The Upload Side Is Where AI Changes Everything
Roberts keeps returning to one point that, in his view, is still widely misunderstood: AI makes upload capacity matter as much as download.
Businesses adopting modern cloud stacks are constantly pushing data upstream—backups, replication, application builds, security telemetry, logs, files, images, recordings, and system updates. Add AI tools—especially those that process documents, video, audio, or large datasets—and that upstream volume grows quickly.
“Consumer marketing trained everyone to think about download speed,” Roberts said. “But modern business operations are upstream-heavy. If the upstream is constrained or inconsistent, it creates friction everywhere.”
That friction can look subtle at first: slower backups, delayed synchronization, longer upload times for project files. But as AI workflows become embedded into everyday operations, Roberts says the tolerance for those delays evaporates.
“AI exposes instability,” he said. “It turns little performance problems into big workflow problems.”
Data Centers Are Becoming a Day-to-Day Topic
One of the more interesting changes Roberts described is how often data centers now come up in conversations with businesses that don’t consider themselves “tech companies.”
He attributes this to a growing awareness that compute isn’t just “somewhere in the cloud.” It lives in specific environments, and how efficiently a business can reach those environments affects performance, cost, and reliability.
“Businesses are realizing they need clean paths into compute,” Roberts said. “That could be cloud on-ramps, colocation, or facilities where critical systems run. The network is the bridge, and people are paying attention to the bridge now.”
As AI pushes more work into cloud platforms and data centers—whether for inference, analytics, or storage—latency and stability become practical concerns, not abstract engineering talk.
“When a system has to respond in real time, you notice jitter,” Roberts said. “When you’re moving large datasets regularly, you notice inconsistent throughput. That’s why data center connectivity is becoming part of business planning.”
Why Fiber Matters: Predictability Under Load
Roberts avoids the “fiber is faster” cliché. He says the real value of fiber is how it behaves when the network is busy—when voice, video, cloud applications, backups, and AI-driven processes are all happening at once.
“Businesses don’t run on speed tests,” he said. “They run on predictable performance. Fiber gives you a foundation that’s easier to scale and more consistent under real demand.”
That consistency becomes even more important as more companies depend on cloud-first systems and distributed teams. In those environments, the network isn’t just one service among many—it’s the backbone that everything else rests on.
Cytranet’s Business-Only Focus
Cytranet’s strategy is shaped by a decision Roberts says has become an advantage: the company serves business and enterprise customers only and does not offer residential service.
“Residential broadband is built around a household experience and high volume,” Roberts said. “Business connectivity is engineered differently. The expectations are different, and the consequences of downtime are different. We’ve chosen to specialize.”
That focus, he argues, keeps the company aligned with what business customers care about most: reliable performance, scalable capacity, and support that treats outages like operational events.
“When a business loses connectivity, it can stop revenue, operations, customer support—everything,” Roberts said. “That’s the reality we design for.”
The “AI Traffic Spike” Is Just Getting Started
Roberts believes many companies are only beginning to notice the impact of AI on their networks. As AI tools become more common—and as businesses move beyond experimentation into daily use—he expects the pressure on broadband to increase.
“People think AI is just a software change,” he said. “But it changes your data behavior. It changes how your systems talk to each other. It changes how often you reach into the cloud and how much you move.”
For businesses in Southern California and Las Vegas, where Cytranet is expanding fiber connectivity and supporting data-heavy operations, Roberts sees an opportunity: build infrastructure now that can support the next phase of demand rather than scrambling later.
“The businesses that succeed with AI won’t just have the right tools,” Roberts said. “They’ll have the connectivity that lets them use those tools every day—without the network getting in the way.”
