What's Happening?
Cisco has identified a growing opportunity for service providers as AI shifts from training to inferencing, which requires less power and latency. At the Cisco Live event, executives discussed how AI inference traffic could account for 25% of total network
traffic by 2035. This shift presents a chance for cable operators and telcos to become 'inferencing cloud' providers, extending beyond traditional connectivity roles. Companies like Comcast and Charter are already exploring AI edge initiatives, while Bell Canada collaborates with Nvidia on inferencing projects.
Why It's Important?
The transition to AI inferencing represents a significant change in network traffic patterns, with AI agents consuming more data and operating for longer durations than human-led actions. This evolution offers service providers a new revenue stream by enabling them to host and manage AI inferencing workloads. As AI becomes more integrated into various industries, the demand for efficient and secure network solutions will grow, positioning service providers to capitalize on this trend. The ability to offer 'inferencing clouds' could differentiate them in a competitive market.
What's Next?
Service providers are expected to continue exploring and investing in AI infrastructure to support inferencing workloads. Cisco plans to expand its network sensors to better track AI traffic, providing insights that could guide future investments. As the market for AI inferencing matures, service providers may develop new business models and partnerships to leverage their network capabilities. The ongoing collaboration between companies like AT&T, Cisco, and Nvidia on AI grid projects indicates a trend towards more distributed and localized AI processing solutions.
