Make serverless a true alternative to provisioned throughput
According to your documentation, serverless is intended for small workloads and containers have a 50 GB storage limit. However, most of us would benefit from a true serverless system in mission-critical apps that only charges for the actual throughput usage. Please make serverless a first-class alternative.
The limits today are for preview only. When serverless goes GA we will drastically increase these.
Yohan S. commented
Could it be a switch in the database interface (scale) so existing applications can take advantage of true serverless CosmosDB without having to do any migration path ?