This would be very useful, as there doesn't seem to be a way to process a tabular model without fully loading, which isn't an option with very large models.

We need this as well - processing a 300M records table is not something you want to do every run.

We need this urgently, else we are forced to stop using TimeXtender for developing and processing semantic layers. Processing multiple large tabular models daily is just not workable.

But remember it's an Enterprise feature. 

https://docs.microsoft.com/en-us/analysis-services/analysis-services-features-supported-by-the-editions-of-sql-server-2016?view=asallproducts-allversions 

It's an enterprise feature or if you're on Azure you need at least the Standard price tier.
But once you have that, we use a simple Powershell script which uses the AMO objects to modify the tabular model and you can run manually or trigger it after a deployment. This script partitions every table in a model which isn't partitioned yet. The partitioning is automatically done on the modulo of the DW_id column if it doesn't exist yet.

We really need this. Especially now we are putting historical tables on AAS. It takes forever before the model is updated with all history. So this is really an important feature that is missing.

Any news on this feature? It is badly needed and completely basic.

We begged for this feature three years ago. Please implement it! Partitioning and Execution in SSAS Tabular – TimeXtender Support