What you are describing is a multitenant application by the sounds of it as you want to use a single application but allow more than one customer to use the application as if it was their own.
There is a lengthy tutorial on medium that mihht give you some ideas.
https://medium.com/@ashokgelal/a-full-featured-multi-tenant-app-with-laravel-part-1-4049a3cc229d
Unfortunately telling you that one database is better than multi or vice versa is the best approach isn't possible as this isn't a one size fits all scenario.
My general rule of thumb would be, if you can manage on a single database then your overall application architecture will be simpler, no network dependencies, one place for all data, administrators can get an overall view of all customers without complex cross database considerations then you should.
Following the tutorial linked above will show that you will end up with a lot less tables than you currently have/anticipate.
If you do find you need to isolate your customers at the database level, like the scenario I found myself in, you will have to spend much more time diagnosing bugs, which may be harder to reproduce as your solution will have a lot of moving parts.
You will also have a overhead when adding a new customer as you will need to provision a new database/server instead of being able to just add a new customer in your existing database. This also means you will need to have a more complex backup process as you will have many different backups from various databases/servers that you need to manage and ensurd they arecworking correctly.
Notice though none of my comments reference database performance. Iv you follow ctonix advice if you have performance issues then the likelihood is the can be solved by either improving the queries you are running, reduce the number of queries (eager loading), increase hardware, especially memory, to allow more data go be loaded and indexed in memory, monitoring and tweaking you database configuration, cache queries that change infrequently in redis to avoid your database altogether.
I'm not saying don't ignore performance but it should not be the main driver in your design decisions.
For reference my current project has thousands of tables, some with tens of thousands of records and I have no bottlenecks with the database layer and I haven't spent much effort on optimising other than making sure my tables are joined correctly and indexed on all indexes and fields I search on. I don't even class this as a massive database.