AI Workloads Have Driven Fiber Demand Higher
Zayo reports a surge in fiber connectivity needs as AI training infrastructure stresses legacy networks.
Updated on Oct. 6, 2026 in Data Centers
Live Poll
Is now a good time to invest in major national digital infrastructure upgrades?
Zayo has identified a significant spike in demand for fiber connectivity, driven by the massive infrastructure requirements of AI training and inference. The company notes that long-haul dark fiber demand doubled between 2024 and 2025 as existing capacity reached its limit.
Why it matters
Modern AI training requires high-speed, distributed connectivity between compute clusters and data centers that the previous era of network investment cannot support. This transition creates a bottleneck that forces providers to navigate complex construction and regulatory landscapes to expand.
Dark fiber demand has increased up to 20 times in specific metro markets compared to legacy usage patterns. Individual customer requirements have now surged to reach thousands of individual fibers per project.
The players
Zayo
Zayo is a global provider of fiber-based communications infrastructure that serves data centers and large-scale enterprise customers.
The details
Projects now face significant construction hurdles as firms must navigate municipal rights-of-way and local regulatory requirements. Because these projects must unfold in sequence, any delays in the early construction phase impact the entire broader delivery timeline.
Timeline
The initial industry network investment wave occurred throughout the 2000s.
Zayo delivered new long-haul capacity routes between 2021 and 2026.
Long-haul dark fiber demand doubled during the period from 2024 to 2025.
The Tech Race
This spike represents the final exhaustion of the network capacity built during the 2000s investment wave. Zayo now positions its new route deliveries to replace these aging systems with high-density fiber capable of supporting modern AI workloads.
As companies rush to expand connectivity, users may experience improved AI processing speeds and more reliable cloud services. However, the reliance on intensive construction could lead to localized delays in network upgrades for regional data centers.
The takeaway
The transition to AI-focused networking requires a complete reimagining of physical infrastructure capacity. Organizations must now account for multi-year lead times in construction to keep pace with the exponential growth in data transmission needs.
Further reading
Learn more about evolving infrastructure requirements at the Data Centers section.
Source note: This article includes information reported by RCR Wireless News.
Live Poll
Is now a good time to invest in major national digital infrastructure upgrades?
