My thinking was over the next 10-20 years improvements in VR and AR and motion tracking and cameras etc. may be able to get us to the point where people can work almost as effectively from a remote location as they could in the office with their colleages.
At the moment a remote employee is just not as effective as the same employee working in the office.
Videocalls are just not as effective as face to face meetings at the moment, even for companies who spend ££,£££s outfitting their meeting rooms. Very few businesses are prepared to properly monitor remote employees (so some people will not work as much as they claim, which in turn makes management less trusting which causes further issues) etc.
The closer we get to parity between remote and in house workers the more options people will have in terms of where they live, and the less we will need to concentrate ourselves in a few major cities.
At the extreme, once we have Full Immersion VR working from home is identical/superior to working from work, and people will mostly stop building office buildings.
Of course I have no idea when the tipping point* will be.
*where time saved commuting and costs saved from not having a big expensive office in the middle of a city outweigh the lower productivity/company unity that WFH results in.
EDIT: This only applies to white collar/non physical work anyone needing to get hands on will need to go into work, or have a telepresence robot or something.