The current internet infrastructure isn’t equipped to keep up with the demands of AI workloads and data traffic, according to Lumen Technology CTO Dave Ward.
The rise of AI — together with evolving enterprise demand for cloud computing — is driving a new cloud economy, a new cloud core and a fundamental network architecture shift, which Ward refers to as “Cloud 2.0.”
“Enterprises are constrained by how fast they can connect to cloud, where they connect to cloud and the workloads they’re running,” Ward said. “This whole new Cloud 2.0 concept is one around the fact that the economy is fundamentally changed, and the underpinning network architecture has fundamentally changed.”
The next evolution of cloud and what’s driving it
Ward leads the development and deployment of service provider Lumen’s global network and has more than 25 years of experience in the telecommunications and enterprise networking industries. He met with InformationWeek at the recent Gartner IT Symposium/Xpo in Orlando, Fla., and also delivered a presentation at the event.
In a recent white paper on the fundamentals of next-generation cloud computing, Ward discussed the demands driving it, including the following:
-
The enterprise’s reliance on multiple clouds.
-
The importance of processing data closer to where it’s generated (edge computing).
-
The integration of AI and machine learning into cloud operations.
-
Support for highly distributed systems.
-
The delivery of low-latency, high-bandwidth experiences to end users.
He forecast that the next-generation cloud that supports these drivers will emerge over the next three to five years.
In an interview with InformationWeek, he explained why the current Internet and cloud infrastructure must be transformed to meet the demands of AI and enterprise workloads. Ward said he envisions fiber and aggregation services being combined within a network fabric that supports workloads “depending on where they’re connecting.”
Data center densification will also be a key building block, demonstrated by the increase in data centers in Tier 1 markets, such as Northern Virginia — the largest and most concentrated data center market in the world — and an expansion of data centers in new “cloud regions” including suburban and rural locations such as in the Midwest and Southwest, Ward said.
Nearly 1 billion square feet of U.S. data center capacity will be added by 2030. By comparison, as of 2024 there was a total of 240 million square feet of data centers in the U.S., according to Lumen.
CIOs require redesigned enterprise networks
Meanwhile, CIOs must reckon with this cloud computing transition, Ward said. That entails rethinking how their enterprise networks are architected, such as their WAN and cloud infrastructure, as they adopt AI technologies.
One of the biggest questions Wards hears from CIO is regarding how they can connect their SaaS clouds and the data centers to access the services they need to train their AI models and run data analytics.
As part of the process of updating their enterprise networks to accommodate AI, he said CIOs need to move away from reliance on hub-and-spoke data center designs, where data flows are centralized through a router in a data center.
“They need a direct cut through in a multi-cloud design,” Ward said. “They do need point-to- point connectivity directly between data centers to create their own data cloud.”
He added that Lumen’s CIO customers are looking to the service provider for support in transforming their “connectivity architecture” to participate in the AI economy.
“The goal I have is that enterprises retain all design and control of their network, but don’t actually have to own, manage and operate their equipment,” Ward said. “This gets us to a point where it’s full flexibility for the enterprise, consumption-based economics, as well as pay for what you use, and allows them to design that cloud core as they see fit.”‘
