The main issue with running AutoLCDs in a large installation is real time response.
The more display surfaces a single instance of AutoLCDs manages, or the more commands that it has to process, the less responsive it becomes to changes on the grid in real time.

Suggestion for improving performance.
* Run the latest version of AutoLCDs.
For instance, the v2.0200 Update definitely improves response times.

* Use block names and limiters when appropriate.
Specific block / block groups names used with the C, T, and G limiters or combinations thereof will significantly reduce the number of blocks to be queried to execute a command.

* The TextLCD command is your friend.
Using a single textLCD command per display surface to display the same information at various points on a ship can be more efficient than using multiple identical command sets on each display surface. It also makes keeping the command set updated a lot easier.

* A cascade AutoLCDs instance.
This is a second instance of AutoLCDs running on the same grid with a different LCD tag. This instance handles display blocks that are 90-100% TextLCD commands. In a large installation, the benefits are that the Cascade AutoLCDs initializes faster than the primary AutoLCDs, and it updates the cascade displays faster than the primary AutoLCDs would as it is running very few AutoLCDs commands that interact with the grid. This in turn takes the load of managing the Cascade displays off of the primary AutoLCDs instance, making it a bit faster as well.
Example: I currently have a Cascade managing 60+ display surfaces, this resulted in a noticeable improvement to the speed of the primary AutoLCD displays.