Have the Option to run Dataflows Gen2 on Fabric Capacity / Use it for large use-cases

Currently it seems that Dataflows are still running on a shared capacity independent on your Fabric Capacity (like it was in Power BI). This leads to unpredictable and non-scalable executions. It would be much better if there's an option to run them on your own Fabric Capacity, then you would have some influence on how fast your DataFlows are processed and if it's reasonable at all to use them for large use cases (i.e. currently it takes 30min to just copy 20million records from CSV to Delta on OneLake).