Lumen Says AI Data Movement Is Defining Challenge

Lumen's Rob Ward told the Broadband Nation Expo keynote that moving AI data across data centers, clouds and edge is the central problem.

Updated on Nov. 18, 2026 in Data Centers

Isometric editorial illustration of a fiber-optic cable spool and amber strands stretching across flat prairie toward a distant modular data center.
Lumen SVP Rob Ward told the Broadband Nation Expo in New Orleans that moving AI data across data centers, clouds, and the edge is the defining challenge for networks. AI Illustration. Upload story photo >

Live Poll

Do you think AI growth will make your internet or electricity costs higher?

At the Broadband Nation Expo in New Orleans, Lumen SVP of global network engineering Rob Ward said moving AI data between data centers, clouds and the edge is the defining challenge. He said silicon and power are growing constraints on AI infrastructure.

Why it matters

Ward argued that AI models, data, users and inference are located in different places, so networks must link them. The shift affects how enterprises plan infrastructure across the United States as workloads spread beyond centralized facilities.

Lumen lit one fiber at about 20 terabits every 18 to 24 months five years ago and two fiber pairs at 50 terabits each every six to nine months two years ago. Newer equipment delivered a 12x space improvement, reducing 12 racks to one, but the source does not detail the full scale of future capacity.

The players

Lumen

Lumen is a major U.S. telecommunications and network infrastructure company that provides fiber and connectivity services to enterprises and cloud providers.

Rob Ward

Rob Ward is the senior vice president of global network engineering at Lumen, responsible for the company's network design and operations.

The details

Ward said customers historically bought a circuit between two endpoints, with humans designing and implementing it. Under the new model, a customer or the customer's agent defines the desired outcome and a programmable network creates it, which Lumen describes as its autonomy aspiration on TM Forum scale at Level 4.

Timeline

  1. Five years ago: Lumen lit one fiber at about 20 terabits every 18 to 24 months.

  2. Two years ago: Lumen lit two fiber pairs at 50 terabits each every six to nine months.

  3. Within about a year: 97% of surveyed enterprises plan edge inference, per Cisco and Omdia.

The Tech Race

Lumen's shift from humans designing individual circuits to customer-defined outcomes reflects the TM Forum autonomous network framework that the industry has used to grade automation. The move positions Lumen against rival network providers as AI workloads push traffic toward edge sites.

Enterprises that run AI workloads may see network provisioning shift from manual circuit requests to outcome-based ordering. The source does not describe pricing or availability changes for end users.

The takeaway

Organizations planning AI infrastructure should expect data to move across many locations rather than one central facility. Network capacity and programmability are becoming part of that planning.

Further reading

Explore more on Data Centers to see how AI workloads are reshaping network infrastructure across the country.

Source note: This article includes information reported by Fierce Network.

Live Poll

Do you think AI growth will make your internet or electricity costs higher?