Must be the data consumption part. AT&T starter package is 300 MB a month.
Edit: I mean to suggest that photos might be the optimization.
If software identified the user's face in live video and transmitted just that at high resolution, discarding or compressing the rest, how much efficiency could be gained?
Separately - as data gets cheaper and video chat is further optimized, video chat over 3G/4G becomes more feasible. Probably our answer to this is in the countries that have abundant bandwidth? What do high schoolers in Korea/Scandinavia use for video chat?