Zayo Partners with Nvidia to Expand Long-Haul Fibre Network for AI Factories

Zayo Group has announced a strategic partnership with Nvidia to significantly expand its long-haul fibre infrastructure, a move aimed at supporting the rapid growth of AI factories, GPU clusters, and hyperscaler deployments across North America. The collaboration underscores the increasing importance of high-capacity connectivity in enabling large-scale artificial intelligence workloads.

The buildout will deliver six net-new long-haul routes across emerging AI corridors, alongside overbuilds of existing infrastructure in ten high-demand markets. According to Zayo, the expansion is specifically designed to connect AI factories, GPU clusters, and hyperscaler deployments as they extend beyond traditional data centre hubs into new geographic regions. Over the past 18 months, Zayo said its AI-focused build and overbuild projects have spanned more than 15,000 route miles across North America, reflecting the scale of demand driven by distributed AI infrastructure.

Steve Smith, chief executive of Zayo, said the company had invested significantly in modelling where AI-driven demand will emerge and was building ahead of that demand, rather than concentrating solely on existing routes. He added that the work with Nvidia was intended to extend high-capacity connectivity to neoclouds and enterprise AI users, not just the largest hyperscalers. This approach positions Zayo to serve a broader segment of the AI ecosystem, from specialised cloud providers to corporate adopters deploying inference workloads.

The announcement follows Zayo's recent acquisition of Crown Castle's Fiber Solutions business, which added 90,000 metro route miles and 40,000 on-net enterprise locations to its network. That acquisition has strengthened Zayo's metro density, a critical factor for AI inference workloads that require low-latency connections between distributed compute nodes and end users. The combination of expanded long-haul routes and enhanced metro reach gives Zayo a more complete network capable of supporting the full spectrum of AI applications.

Dylan Patel, chief executive of SemiAnalysis, said connectivity was becoming as significant a constraint on AI scaling as compute itself, with long-haul fibre needed to link distributed infrastructure in corridors where capacity is currently scarce. His observation highlights a growing industry consensus that network infrastructure must evolve in parallel with compute capacity to avoid bottlenecks in AI development. As AI models grow larger and training clusters become more geographically distributed, the role of high-bandwidth, low-latency fibre networks becomes increasingly central to performance.

In addition to the network expansion, Zayo recently launched its AI Infrastructure Blueprint, a framework intended to connect training, inference, and interconnection environments across the AI ecosystem. The blueprint provides a structured approach for enterprises and service providers to plan and deploy AI-ready network architectures, addressing both current needs and future scalability. Vladimir Troy, vice president of engineering for AI infrastructure at Nvidia, said network capacity was becoming as critical to AI progress as compute, reinforcing the collaborative focus of the two companies on building the foundational infrastructure required for next-generation AI.

The partnership between Zayo and Nvidia reflects a broader industry trend in which network providers are working closely with chipmakers and cloud operators to anticipate and meet the infrastructure demands of AI. By investing in long-haul routes that connect emerging AI hubs, Zayo is positioning itself to capture a growing share of the connectivity market driven by AI adoption. The expansion also signals that the geographic footprint of AI infrastructure is broadening, moving beyond established data centre markets into new corridors where land, power, and connectivity are more readily available.