How will CXL pooling and Kubernetes co-exist?

I have been reading more about the CXL 2.0 spec, specifically resource pooling. This feature enables servers to pool resources such as accelerators and/or memory that can be assigned to different servers depending on the workload. I have limited understanding of Kubernetes, but the basic idea to solve is being able to scale up or out depending on usage. Anyone have ideas about how CXL will change the landscape up on the stack?