Ciena’s latest global survey with Heavy Reading, now part of Omdia, found that communications service providers (CSPs) expect artificial intelligence (AI) to become a dominant driver of network traffic within the next three years, especially on long-haul routes. Nearly one-third (29%) of respondents predict AI will account for more than half of long-haul traffic, while 52% believe it will surpass 30%. In metro networks, 18% expect AI to exceed half of total traffic, with 49% forecasting AI will make up at least 30% of metro traffic.
The study highlights growing demand for high-capacity connectivity to support AI workloads, particularly from enterprise customers. Half of CSPs ranked high-bandwidth wavelength services—at 100G, 400G, and 800G speeds—as the top growth area from AI traffic over the next three years, outpacing demand for dark fiber. Three-quarters (74%) expect enterprises to drive the most traffic growth, followed by hyperscalers and cloud providers.
Despite these growth expectations, just 16% of CSPs say their optical networks are “very ready” for AI. The majority remain in partial readiness, citing key obstacles including capex constraints (38%), business strategy alignment (38%), and network management (32%). The survey, conducted in February 2025 with 77 CSPs worldwide, underscores the urgency for network upgrades to meet AI-driven bandwidth needs.
• 29% of CSPs expect AI to contribute over 50% of long-haul traffic by 2028
• 18% expect AI to exceed 50% of metro network traffic; 49% forecast at least 30%
• 50% see high-bandwidth wavelength services as the fastest-growing AI-driven offering
• 74% expect enterprises to drive the most AI-related traffic growth
• Only 16% say their optical networks are fully ready for AI
“This research highlights the rapid rise of AI applications — from large-scale models to cloud AI services and edge-to-core workflows — that are set to become major drivers of both local and long-haul network traffic,” said Sterling Perrin, Sr. Principal Analyst, Heavy Reading – now part of Omdia.
The report is available here.
🌐 Why it Matters: AI traffic—particularly for model training, inference, and data movement between edge and core—requires ultra-high-capacity, low-latency optical transport. With most CSPs still in the “somewhat ready” stage, the industry faces a tight window to scale wavelength services, optimize network architectures, and secure capex for rapid upgrades. The shift could reshape optical vendor strategies and accelerate the rollout of 400G and 800G technologies.