Interoperability

We monetize on running the server for private re-training? Private Models? (does it matter)? Or in exchange for DP, we get anonymized data sets back for training?

How do those flowers federated learning folks fit in?