I know the max worker node are 10 per agent and we should use 50% for better performance
The recommended value for this setting should not exceed 50% of the available CPU cores (including hyper-threads if available) and 2GB of RAM per worker.
Is there a possible scenarios or in future release will this be reduced to better preform like selenium grid or docker ? 2GB per worker is actually not scenario we are looking at running our tests from could version of test project, In actual real life scenario we are looking at running 1000 test cases in parallel. any other possible scenarios, most of remote test project agent are 8core CPU with 16GB RAM.