Samsung agentic AI networks target autonomous Level 5 telecom operations

The shift toward agentic AI allows infrastructure to move beyond simple automation toward self-healing systems that manage mission-critical video traffic without human intervention. For the streaming ecosystem, this transition promises to mitigate service degradation during unpredictable traffic spikes by resolving bottlenecks before they reach the end-user. By integrating telco-grade models with O-RAN compliant architectures, Samsung is positioning network performance as a programmable asset for enterprise streaming applications. Watch for the TM Forum to release updated Open Digital Architecture specifications as more operators attempt to hit the 2027 deployment window for Level 4 systems.

Samsung's CognitiV Network Operations Suite enters a crowded field of vendors pursuing autonomous network operations. In March 2025, Nokia announced its MantaRay platform had been selected by three European operators for closed-loop automation, targeting Level 4 autonomy across radio access and transport domains by end of 2026. Ericsson has similarly scaled its Intelligent Automation Platform, which the company said in June 2025 was deployed across 12 tier-one operators for autonomous orchestration with a stated goal of reducing mean-time-to-resolution by 40 percent. These deployments establish the competitive baseline Samsung must clear to win operator commitments for its CognitiV stack.

The standards ecosystem around autonomous networks is tightening, directly shaping how Samsung's agentic AI roadmap will be evaluated. The TM Forum's Open Digital Architecture working group published updated intent-driven networking specifications in May 2025 that define the API contracts between agentic systems and network functions, creating a compliance framework that CognitiV must align with for multi-vendor interoperability. Meanwhile, the AI-RAN Alliance, which Samsung co-founded alongside NVIDIA and SoftBank, released its first technical report on AI-native RAN architectures in April 2025, outlining how inference workloads can be co-located with radio functions to reduce fronthaul latency. The O-RAN Alliance's Work Group 2 has also incorporated agentic AI use cases into its near-real-time RIC specifications, signaling that autonomous decision-making at the RAN layer is becoming a standards-track requirement rather than an optional enhancement.

On the technical side, Samsung's deterministic latency claims for streaming traffic face scrutiny from independent testing. A Heavy Reading benchmark study published in July 2025 evaluated closed-loop automation platforms across five vendors and found that mean-time-to-recovery for video-streaming QoS violations ranged from 8 seconds at best to over 45 seconds at worst, depending on the complexity of the traffic pattern. Samsung's CognitiV was not included in that particular study, but the results establish the performance envelope operators will expect. Separately, SK Telecom disclosed in August 2025 that its AI-based network optimization reduced peak-hour video buffering events by 22 percent across its 5G standalone core, providing a real-world reference point for the kind of streaming-quality gains that agentic AI platforms like CognitiV must demonstrate at scale.