Data centers and airports

How would you create an effective network from scratch? Sure, you could experiment and then evolve over time. But what if you had to build a good network right now and couldn't experiment?
One idea came to my mind on my bike on the way to work: could you copy the size and location of airports for the efficient placement of distributed data centers? This might be a good predictor for suitable places. Why? One assumption is that airports are located at places where there is a need, either direct need or indirect need as a hub. One can also argue that over time more effective and better located airports grew whereas airports at unpopulated areas decreased in size and vanished. Finally, the cost structure is similar: high fixed costs and little variable costs.
Here are two examples: my previous hometown Charlottesville has only about 40.000 inhabitants and has a small airport. Considering the fixed costs of an airport (btw, you can compare those costs to a data center) one would only invest if there's enough return, i.e. traffic. Smaller town won't probably have an airport or only for recreational purposes. Of course large cities such as San Francisco have large airports to serve the local population. Now let's look at hubs. For hubs the economics are similar and to a part totally different from regular airports. Hubs channel traffic and require a larger size. Hence, location between centers and size are important. It is no surprise that Atlanta and Chicago are major hubs. Certainly real estate prices are cheaper than in Manhattan than in Georgia and it helps that both airports are in the middle of the country.
So why should we care? Both categories (data centers and airports) are totally different but share in their core the same characteristics. One can certainly draw useful conclusions from the case above and I challenge you to find similar cases for your business / situation.