I had asked basically the same question..and was pointed to this design as a reference:
I think this solution (one servo, one belt driving two screws) is performance/cost optimal. The gear idea would probably work, but sounds very expensive (if you want to minimise backlash). Also, I'm sold on using servos -- especially for a larger axis. The flatter torque curve across a wider range of (higher) RPMs seems like a better solution.