All ion thrusters are quite small. Having read through the more approachable information about them, as they use physics pretty beyond me, I don't understand what it is about them that makes it necessary they be so small. Are they fundamentally limited in some way? Is there reason to hope what limits them now could be overcome at some point so their high Isp could be used for higher acceleration on larger spacecraft?
Edit (next day): The answers so far have helped me understand the basic issue that the power needs of these engines are huge. In some ways I think this deserves to be treated as a separate issue from the scalability of the engine itself - this article makes the point that energy could be beamed to the craft, for instance, meaning you don't need to carry the mass of the power source with you. And the point has already been made that nuclear reactors could be used, which have the required power density, if the details of how to make use of that was figured out (see supercapacitor conversation in comments under EchoLogic's answer).
The delta V imparted for their weight is very high. NSTAR engines weigh 48 kg including the thruster, power processing units, xenon feed system, and control interface. Three of those and 475 kg of xenon will impart delta V of over 10 km/s to the Dawn probe, itself weighing about 600 kg without those thrusters.
So if the power issue was handled, could you just make great big ones to push a large vessel fast, and take advantage of that efficiency?