In the world of cloud compute multi-cloud architectures are dependent on leveraging Kubernetes as an abstraction layer for running containers across cloud compute providers.
What happens however in the world of AI native applications?  Just as in the world of cloud compute we need Kubernetes as an abstraction layer to run containerized AI models across GPU providers in conjunction with running containerized application logic across CPU providers.
A new serverless cloud compute platform is needed for building and running AI native applications on Kubernetes to support portability of code and models, freedom, flexibility and choice across.