IDEALLY, yes, but not necessarily.
This can get tricky.
You might use one EA on one account trading multiple pairs. That would be okay, but if you want to know how the EA is doing with just one of those pairs you will not be able to generate a report for that, your report would include the trades for all the pairs. For example, you're trading the Euro, Yen, and the Loonie with the same EA. You have no way of knowing how the system is performing on the Yen, just the whole basket. It may be doing well on the Yen and poorly on the Loonie. You need to know that so that you can make adjustments. You could manually break down your account summary but that would be a lot of work.
If that's what you want, that's okay.
The more important issue is running one EA per account, because you don't want the EAs conflicting or competing for the money in that one account. So again, if you have the resources- one EA, per pair, per account will avoid any confusion. Admittedly that is a lot of resources and it could get expensive.
Once again, if you demo test your EAs with the currencies you want (yes, it will be time consuming) and can narrow it down to the best few, take those live to your host and keep it simple.

A more important point to make is the nature of the EA itself. If an EA generates an order that includes a TP and SL, you have less at stake than using an EA that generates just market orders and then waits for a certain condition to close the trade. With the first scenario, your broker will have the TP and SL in his server. If your PC crashes, the worst that can happen is no new trades will be taken, but your TP and SL are there with the broker, they'll get executed. Not so with scenario number two. If there are no directions to close the trade and you crash, you are at the mercy of the market. This is where you need the bulletproofing.
You need to assess what types of EAs you're using; you may need a VPS for some, but not all of them.