Zayo Building 8,000 Miles of Fiber for Nvidia AI Corridors

Zayo is constructing more than 8,000 miles of new long-haul fiber infrastructure across North America with Nvidia as its anchor customer, addressing emerging capacity constraints as AI data centers spread beyond traditional tech hubs.

The Denver-based digital infrastructure provider will build six new long-haul routes and expand capacity across 10 high-demand markets, bringing its broader AI-focused program to more than 15,000 route miles. Nvidia will have significant access to capacity on the new routes, while Zayo retains ownership and will make remaining capacity available to other customers including AI developers, cloud providers, and enterprises.

Why it matters

As AI factories chase available power into secondary markets, the network layer is becoming a potential chokepoint. Existing fiber routes were designed for legacy data center topologies, not for the distributed GPU clusters that now need to synchronize training data across geographically separated facilities. Building new fiber routes takes 12 to 24 months—roughly the same timeline as data center construction—meaning providers must anticipate demand or risk leaving completed, powered facilities without adequate connectivity.

Capacity constraints emerging on two fronts

Network bottlenecks are developing through different mechanisms, according to industry analysts. Ron Westfall, vice president at HyperFrame Research, said existing routes are already hitting capacity limits as AI facilities move into new markets. "Existing fiber routes were designed around legacy data center topologies," he noted, with wide-area optical capacity becoming a rate-limiting factor for distributed computing clusters.

Meanwhile, optical equipment availability is constraining efforts to add capacity to existing routes. Jimmy Yu, vice president at Dell'Oro Group, said pump lasers used in amplifiers and coherent transponders face near-term supply constraints, with Ciena and Nokia reporting product lead times beyond 12 months and growing backlogs. While Yu has not observed widespread wide-area network bottlenecks from AI traffic yet, he acknowledged the industry remains in early stages of the AI infrastructure cycle.

Purpose-built routes for AI workloads

Zayo CEO Steve Smith said AI is "fundamentally reshaping where and how network infrastructure needs to be built across the U.S." The company has spent 18 months expanding its AI-focused network footprint, targeting corridors where existing long-haul capacity is limited or nonexistent.

AI workloads generate distinct traffic patterns compared to traditional cloud networks. Distributed GPU clusters require substantial bandwidth to coordinate workloads across regions, while training runs can produce traffic bursts as systems synchronize data between geographically separated clusters. Yu said purpose-built routes with low latency and thousands of fiber pairs could become critical infrastructure for connecting AI data centers to other facilities and exchange points.

The timing challenge is acute: a data center can be complete, powered, and equipped with GPUs while lacking the connectivity needed to operate as part of a larger distributed AI system. "Providers need to begin building these new fiber plants now otherwise data centers will sit idle," Yu said.

Zayo's expansion follows its acquisition of Crown Castle's Fiber Solutions business, which added 90,000 metro route miles and 40,000 on-net enterprise locations. The company said the expanded network will serve hyperscalers, neocloud providers, frontier model developers, and enterprises across healthcare, finance, manufacturing, and other sectors.

These details were first reported by Data Center Knowledge.