I'm more and more convinced we will see a reversal here. My guess is this: society as a whole will eventually reject the idea that they should be burdened with a massive expensive infrastructure just to have people sit in offices in centralized locations, staring at screens and doing the same job they could do from their homes. Or that shops need to be in centralized locations. Roads, railways, brick and mortar shops will all be gone and replaced with rural settlements in natural surroundings and virtual "travel" (VR) for the most part, air travel (drones, helicopters, flying cars) for the cases when absolutely necessary (transport of goods). It will probably be a gradual, but accelerating process as the bits fall into place.
So no, I don't believe urbanization is a good bet, not even in the medium term at least not in developed countries. Goods and services that support the above scenario on the other hand ...
1. Urbanization
More than 60% of the world population will live in cities and existing cities will become bigger ceteris paribus. I could write a thesis on this but you need to think very broadly about this. A lot of unicorns are playing in this space think Uber, instacart, Zillow etc
2. Global Southern shift Basically demographic and economic centre of activity is shifting to global South. This is different from southern hemisphere. Pretty much any business focusing on consumption sector in these areas is a good bet. Think FMCG FMEG
3. Ultratrend - climate change Climate change will play havoc with so many layers of our economic systems. While most people think agriculture, you can also add transportation, manufacturing, renewable energy, reinsurance, fishing, textiles, pharmaceuticals
Any solution that is climate defensible is gold
4.Technology breakthrough
Here an IP defensible stack with a focus a sector or niche in above 4 areas would be a force multiplier.
Here I guess basic framework could be data + operational expertise + combo{ML, AI, serverless computing.. }
Chemical companies could be an interesting bet. There is a lot of work happening in using AI for molecule discovery, process discovery. Any company which cracks this will be the Google of "atoms"