Why Is It So Hard to Run a 5G Core on Kubernetes — And What Needs to Change for 6G#

Joel Studler, Swisscom & Ashan Senevirathne, Telstra

• Why: The sharpest critique in the program, from two operators rather than a vendor: most "cloud-native Network Functions" are merely containerized ones. For anyone watching telco-DC convergence, this names precisely where the abstraction is being faked.
• What: Why CNFs remain box-centric, illustrated with the argument that a User Plane Function is not inherently cloud-native, and what 6G would need to change.
• How: The specific indictment is that Helm install mimics hardware mounts, Multus wires up virtual cables, and NETCONF configures boxes — so these designs bypass Kubernetes' own IPAM, networking and high-availability strengths while being marketed as telco-grade. The forward-looking half argues for Kubernetes-first patterns and for exposing APIs so the network can be treated as a product.
• Where applicable: Telco-specific in its examples, but the underlying pattern — lift-and-shift dressed as cloud-native, bypassing the platform's strengths — is general and recognizable well outside telco.