Is it 65,000 objects (for instance, 65,000 customers) or 65,000 entities (65,000 different types of objects inside of Mendix).
I am guessing it is 65,000 objects. If this is the case, and your microflow retrieves all 65,000 objects in one retrieve action and then tries to loop through those to do some processing, you are likely running out of memory. I would suggest running this in smaller batches. I have found the Community Commons Java action executeMicroflowInBatches to be useful when processing large numbers of objects. You can create a microflow that processes a single object, and then have another microflow that uses this Java action to call that microflow. You can specify how big you want the batches to be (maybe 500 or 1000 objects at a time).
This is just one approach - I am sure the community has other ways to accomplish this type of batch processing.