As @LipChean_Soh already mentioned that more the robust design less the performance issues. Anaplan handles it quite well even if there are dozens of concurrent users on the same dashboard provided we'd followed the best practices in the model design.
On Data Load - See if you can break the load in chunks - you can take it to 30, 40 chunks etc. and make sure that the load doesn't happen during peak time.
Data Load from End users - Now this is something that is the bone of contention. Let me share my views: If there are end users who start loading their data at the same time then it might take sometime (here we are talking about 10 seconds to 20 seconds - based on the size and the number of users). If your processing time is higher than a minute or two then it might be worth to break the calculations link based on the input line item and stage/store end user driven numbers into standalone module and then pull the data for calculations later (schedule the job later during offpeak period) - this you have to take a buy in from end users that there will be a trade off between the performance and the real time