I heartily look forward to when there are secure, privacy-respecting solutions to these problems… but I am not convinced the major cloud vendors are going to be a good friend in getting there.
On a number of occasions, there have been capabilities for integrating different jupyter technologies with different proprietary cloud solutions. However, each time, the vendors have deprecated the API. Using the most recent developer-level tools suggested to replace them, it is rather easy for an unwary user to generate a lot of writes, which can start costing real money.
With the yjs-backed CRDTs underlying the new JupyterLab collaborative editing features, it’s plausible someone could write a yjs-major-cloud-vendor connector.
However: I don’t see a first-party Jupyter integration for any of the major platforms’ sync technology on any core timelines, barring significant, sustained contributions from those vendors to maintain them. Unfortunately, i think we’re seeing the contrary happening, making it harder to share compute and data outside of their walled gardens.