Ciena Analyzes Impact of AI Applications and Traffic Growth

Ciena and Heavy Reading (now merged into Omdia) recently conducted a global survey aimed at gaining a deeper understanding of the impact of artificial intelligence (AI) applications and their traffic growth on the metropolitan and long-distance networks operated by communication service providers (CSPs). The interviewees include fixed line, mobile, converged network operators, and cable TV operators.

The research results show that CSPs generally expect AI to bring significant traffic demand in their metropolitan area networks in the next three years: 18% of respondents believe that AI will contribute more than half of the total metropolitan area traffic, and nearly half (49%) expect AI to account for more than 30%. In terms of long-distance traffic, CSP’s expectations are more optimistic: over half (52%) of respondents believe that AI traffic will exceed 30%, and nearly one-third (29%) expect AI to contribute over half of long-distance traffic.

Sterling Perrin, Senior Chief Analyst at Heavy Reading, said: “This study highlights the rapid rise of AI applications – from large-scale models to cloud based AI services, and to edge to core workflows – that are about to become the main drivers of local and long-distance network traffic. In metropolitan area networks, AI will compete with video, web, and IoT traffic, and its predicted growth is particularly astonishing. Given that AI is expected to occupy a larger proportion of long-distance capacity within three years, it is evident that AI data streams for training and inference will pose unprecedented demands on CSP networks. ”

Connection services for AI traffic

Research has found that as the market matures, CSP is expected to play a broader role in AI connectivity, especially for enterprise customers, providing 100G, 400G, and even 800G high bandwidth wavelength services.

50% of respondents ranked high bandwidth wavelength services as the top priority in the “services with the fastest growth due to AI in the next three years” option; By contrast, only 25% of respondents believe that dark fiber will grow the fastest due to AI. 74% of CSPs expect enterprise customers to be the biggest driver of their network traffic growth in the next three years, ahead of hyper scale cloud service providers and cloud providers.

Optical Networks Meet the Challenges of AI

Research indicates that global CSPs face various challenges in embracing the AI traffic boom. The top three challenges are: restricted capital expenditure (38%), market entry/business strategy (38%), and network management (32%).

Although 16% of surveyed CSPs believe that their optical networks are “very mature” and can meet AI needs, the majority of respondents acknowledge that there is still work to be done: 39% say the network is “ready,” but “most devices have been deployed and still need improvement; 40% believe that “only partially ready and there is still a lot of work to be done”; 5% admit that their optical network is “completely unprepared”.