Solmar Insights
Zayo has announced plans to construct, own, and operate more than 8,000 miles of new long-haul fiber optic routes across the United States, with Nvidia signed on as one of its first major customers. This initiative is designed to address soaring connectivity demands from AI-driven data traffic that is outpacing the capacity of existing digital infrastructure.
Key figures
8,000 miles of new fiber
Six new long-distance routes
Upgrades in 10 major traffic markets
Scope of the fiber expansion
Zayo’s new build will result in more than 8,000 miles of fiber-optic cable laid to meet surging demand for AI-specific applications. The planned routes will connect metropolitan areas where the companies anticipate the fastest growth in AI traffic volumes, according to the announcement. In total, the project includes six newly established long-distance routes and supplemental network upgrades to boost capacity in ten markets already experiencing high data traffic.
This expansion reflects a response to increased bandwidth needs from hyperscalers and enterprises deploying generative AI and other high-performance compute solutions, all of which require ultra-low-latency, high-capacity connectivity between data center hubs. The new fiber will help bridge digital infrastructure gaps along key US corridors, supporting emerging clusters of AI compute hardware.
Nvidia’s involvement as an early customer signals the build’s close alignment with demand for AI training and inference workloads. As GPU clusters and AI workloads scale, these connections improve the efficiency and throughput of data transfers between regional and national locations.
Analysis: Drivers of demand for AI fiber routes
The rapid introduction of AI applications in industries from healthcare to logistics is fueling significant strain on existing metro and long-haul fiber assets. AI workloads in particular are highly data intensive, involving constant movement of large datasets during training and operation phases. This has created new connectivity bottlenecks between distributed compute facilities and hyperscale data centers, especially in regions not previously known for tech-heavy traffic volumes.
Network operators like Zayo are targeting these pain points with next-generation fiber investments focused on speed, reliability, and route diversity. The Zayo-Nvidia build positions both companies to serve the increasing requirements of cloud providers, AI startups, and enterprises needing specialized network services for AI-centric computing. Nvidia’s role as an early customer further underscores AI’s advancing importance as a driver for network buildouts in US digital infrastructure.
By increasing underlying fiber capacity, Zayo can differentiate against peers and offer premium services for AI-specific traffic, which prioritizes latency and data movement over traditional consumer internet demands. This supports the company’s strategy to serve data center operators and software vendors investing in regionalized AI clusters and edge compute facilities for distributed training and inference.
Routing strategy and upgraded corridors
The plan’s emphasis on both new long-distance routes and upgrades within ten significant traffic markets points to a dual approach: targeting both underserved corridors and boosting established ones where congestion is greatest. While the specific cities and route maps have not been disclosed, the focus lies on creating new fiber pathways and expanding bandwidth along heavily trafficked backbones connecting major metro areas.
This approach allows for improved redundancy and route diversity, safeguarding against outages and single points of failure. For large AI operators, such route diversity is crucial both for security and for guaranteeing consistent throughput no matter the workload location. Each new or upgraded segment provides more options to harmonize east-west and north-south data transfers across the US, a key factor for AI model training that requires coordination between far-flung hardware clusters.
In addition to serving Nvidia, the new network is likely to attract further early adopters among cloud and colocation providers with similar high-bandwidth, low-latency needs. Tenants in upgraded capacity markets will benefit from a future-proofed platform prepared to handle AI-generated bandwidth surges for years to come.
Market implications for digital infrastructure
This new round of investment signals that the backbone of US AI infrastructure is expanding beyond just data center square footage and compute. Fiber networks, especially purpose-built, long-haul routes, are now tailored to the requirements of distributed AI systems. The magnitude of the Zayo project, at over 8,000 miles, highlights the scale of digital infrastructure needed to keep pace with emerging demand.
For institutional investors and developers, Zayo’s commitment underscores ongoing opportunities in high-capacity network infrastructure as a complement to traditional data center and cloud deals. Carriers with the ability to build, own, and operate their own networks are best positioned to capitalize on multi-decade secular shifts like surging AI adoption, while early anchor tenants such as Nvidia enhance deal bankability by de-risking upfront occupancy.
As additional AI hardware clusters come online in new geographies, or established regions see upgraded network capacity, the appetite for both greenfield fiber and brownfield network overlays should remain strong. Zayo’s move signals a market turning point where connectivity is placed on par with (or even above) power and location considerations in enterprise AI planning.
What to watch in next-phase buildouts
While the announcement identifies Nvidia as a core customer, the ultimate success of the expansion will depend on Zayo’s ability to attract a broader base of tenants, from hyperscalers to mid-market colocation providers. The buildout’s emphasis on connectivity between fast-growth AI traffic zones sets a template for where future fiber investments may cluster. Stakeholders will be monitoring how quickly Zayo can deploy new routes and whether other digital infrastructure investors follow with similar long-haul projects.
Attention will also turn to network interconnection strategy: how these new routes interface with existing backbones, neutral internet exchange points, and points of presence operated by major cloud and AI service providers. Given surging demand and the extended lead times for permitting and construction, delivery timetables will remain a focus for customers and capital providers alike.
Market observers are likely to track how new fiber capacity influences both lease rates and customer expectations for performance, particularly as regionalization of AI clusters drives demand for more diverse and resilient bandwidth supply chains. Each successful completion will lay groundwork for further expansion into secondary and tertiary US markets.
What this means for buyers
Buyers evaluating hyperscale and AI workloads should note that fiber connectivity is being prioritized alongside power procurement and location planning in digital infrastructure builds. Zayo’s project with Nvidia sets a precedent for integrating long-haul fiber investments with emerging compute clusters, creating new patterns of demand for dark fiber leasing and interconnection. Early engagement with new fiber routes can optimize AI data movement and reduce potential bottlenecks as capacity needs scale. This development points to a future where network strategy is as critical as data center footprint in supporting next-generation compute.
