There are several MMOs in development using Torque. There's an MMO starter pack that is available from the Torque community. Minions of Mirth from Prairie Games uses Torque and has hundreds of players.

Torque's networking isn't going to let you do an MMO right out of the box, but that's no reason not to use it. The thing about Torque is that you get the complete source code to modify as you see need. It's the same case as the Unreal Engine, which is used in Lineage, Vanguard: Saga of Heroes, and other MMOs. UE is an FPS engine, but licensees modify networking, terrain rendering, and whatever else they need to get the job done. For Minions of Mirth, Josh Ritter integrated Python with Torque and used the Twisted framework for some of the networking, though Torque's networking is still at the heart of the game.

So Torque is not a shrink-wrapped solution for an MMO, but that doesn't mean it can't be used as a base to build one. That implies that you have to know what you are doing.

If you played the original EverQuest, it lagged pretty hard when you had more than 100 players in a single zone. The way they scaled the game to thousands of players per shard was to run each shard on its own machine (a few less popular zones can share the same machine, but run in different processes).

With Torque, suppose you could do 128 players per process -- that's not bad! Split your game on 15 zones, run each zone on a separate process, and you support up to 1500 simultaneous players! (In reality, they will not spread out that nicely, but the basic idea still applies).

My opinion is that you shouldn't attempt to scale by putting all your eggs in one super big, super fast basket that costs a lot of money and requires regular sacrifices of red-headed virgins to keep operating. You should scale by plugging in cheap, almost "disposable" server hardware, and farm the load out over a cluster.

Hplus speaks the truth here. Trust me, sacrificing red headed virgins get old fast, especially when your user base grows faster than the tech of the potential hardware you can buy to fix a particular problem. Then it's a race between your userbase growth and the speed optimisation you can make while trying to hack in some sort of multi server capabilities into an architecture that was never meant to support it.