You haven't included much information about your System, VMWare Workstation version, preferences or even the VMs themselves.

My first guess would be that this could be related to the 'Process priorities' options on being grabbed vs ungrabbed.

For example: If you have them set to run as high for grabbed, then it could end up making the host itself seem to lag for longer periods - Especially if over extended on the total actual cpu core count over the multiple VMs (and not leaving any free ones left over for your actual system)

The above doesn't seem to be what you are describing but it could be as simple as having 'any' priority differences set up that trigger the extra cpu usage while changing the priorities while also triggering the grab or ungrab. If it's just a short spike in VMWare Workstation 'doing' something and not creating an extended lag spike or an unresponsive system then tbh I'm not sure what else to say atm.