TM Forum CEO warns telcos to avoid 'betting the farm' on edge

Willetts' skepticism signals a necessary recalibration for the streaming infrastructure stack. If telcos pivot away from public edge compute in favor of internal 'AI-native' network autonomy, the burden of low-latency video processing and AI inference may shift back toward hardware-heavy consumer devices or hybrid on-premise solutions. This highlights a growing divide between Western operators, who face higher human and regulatory barriers to automation, and Chinese telcos who are already operationalizing Level 4 autonomy. Watch for whether high-bandwidth streaming use cases, like real-time spatial video, can survive without the distributed edge infrastructure telcos were previously expected to build.

The shift toward 'AI-native' operations is a central theme at DTW Ignite 2026, where TM Forum recently evolved its Open Digital Architecture (ODA) from a cloud-native blueprint into an environment tailored for AI execution. This transition coincides with the launch of a joint white paper by Vodafone, Google Cloud, and TM Forum in June 2026, which outlines a framework for shifting from manual, reactive operations to intent-based network autonomy. Research from STL Partners suggests that achieving full network autonomy could provide more than $800 million in annual upside per operator by reducing the current reliance on human intervention for routine tasks.

While Western operators work toward these efficiency gains, Asian carriers have already established a technical lead. Per TM Forum and CAICT reporting from March 2026, Chinese operators have begun validating Level 4 autonomous networks in live environments, benefiting from a lack of legacy labor constraints and a streamlined vertical integration of cloud and network assets. This regional gap is driving Western vendors like Nokia and Ericson to deepen partnerships with hyperscalers; for instance, Nokia recently integrated its autonomous fabric natively onto AWS to help Western clients reach Level 4 status faster.

Compounding the pressure on telco edge investments is the rapid advancement of localized hardware. Per NVIDIA and Microsoft, May 2026, the newly unveiled RTX Spark platform delivers 1 petaflop of AI performance directly to Windows PCs, potentially negating the need for telco-hosted GPUs for many AI use cases. As compute power decentralizes further to the device level, the strategic focus for streaming and infrastructure providers is shifting from 'where' data is processed to how autonomously the delivery network can manage its own performance and reliability.