No. Not in a sensible, usable, trouble free way.
What you are trying to do is really nasty and violates best practices but i'll give it a try anyway:
Connect ESXi Server 1, Port 1 to the Windows Server, Port 1 and use IP addresses in IP Range A
Connect ESXi Server 1, Port 2 to the ESXi Server 2, Port 2 and use IP addresses in IP Range B
Connect ESXi Server 2, Port 1 to the Windows Server, Port 2 and use IP addresses in IP Range C
Connect ESXi Server 2, Port 2 to the ESXi Server 1, Port 2 and use IP addresses in IP Range B
This creates a triangle so to speak. You cannot use IP addresses in the same subnet between hosts as every host is cross-connected directly to another and hosts do NOT forward packets in the same Layer 2. They are not switches after all.
If you try it anyway, and give ESXi hosts kernel-ports on the different physical interfaces, you will get strange behaviour as multiple interfaces in the same subnet will create funky problems.
If these are all the interfaces these hosts have. If they don't have extra 1gig Interface for example, you will not be able to reach anything else like Laptops or the Internet or anything.
Honestly, this is stupid. Don't even try getting this to work. I does not make any sense. Get a switch.
