Toward an Evolvable 6G Telco Cloud Infrastructure: Enabling Service Provision and Experimental Validation
The emergence of sixth-generation (6 G) networks demands a radical shift in how telecommunication infrastructures are designed, managed, and operated. This paper presents a comprehensive vision for an evolvable 6 G telco cloud and service provisioning platform that supports multi-provider, multi-technology environments across distributed sites. The proposed infrastructure integrates disaggregated network functions, open-source frameworks, AI-native orchestration, and end-to-end virtualization to deliver advanced services such as AI-as-a-Service (AIaaS), ISAC, Compute-as-a-Service (CaS), and Security-as-a-Service (SecaaS). Emphasizing sustainability, scalability, and openness, the platform enables dynamic service delivery, supports vertical-specific use cases, and validates performance against stringent 6G Key Performance Indicators (KPIs) and societal Key Value Indicators (KVIs). To validate the proposed platform, we present a simulation study of Service Function Chaining (SFC) latency path optimization across four representative 6 G service chains: Ultra-Reliable Low-Latency Communications (URLLC), Integrated Sensing and Communication (ISAC)+AIaaS, SecaaS+CaaS, and Holographic, spanning far-edge, near-edge, and core-cloud layers. Results show that, under the baseline delay parameterisation, URLLC and Holographic chains exceed their SLA budgets from zero load onward, ISAC+AIaaS exceeds its SLA beyond $\approx 14 \%$ network load, and only SecaaS+CaaS remains below its deterministic SLA budget up to approximately $\mathbf{6 5 \%}$ network load. Cross-layer Virtual Network Function (VNF) transitions increase latency by up to $40 \%$ per hop. Based on this per-hop penalty, we estimate as a design hypothesis that an AI-native orchestrator minimizing inter-layer placements could reduce SLA violation rates by $40 \%$ to $60 \%$ at moderate loads.