Possibilities using current technology

• A fully 3D, navigable Google streetview. Right now each view is a panoramic photo wrapped around to give the illusion of perspective. You can navigate using the standard WASD keys, but the camera is stuck on a rail (the Google streetview van). There's software being developed that can automatically construct 3D geometry based on photos. Using the vast database of panoramas from Google, it is possible to recreate entire cities down to the pedestrian level, with freely changeable perspective. I'm guessing the Google team already has a working alpha.
  
  
• Streaming games that are rendered at powerful graphics servers. Why spend 1000$ for a machine that can play Crysis when you can stream video output from rendering farms? Most of us already have enough bandwidth to stream HD media. The main problem will be lag, due to the need to have a streaming buffer of a second or so. I'm sure programmers can find a way around that. Word is AMD is already working on this, but if this becomes successful it will destroy their market for PC video cards.
  
  
• Combine the two technologies above to get fully realized virtual cities. I hear Tokyo is beautiful this time of year.