It's not like amazon is going to airlift in 50k people on day 1.
Anyway, Northern Virginia is still the obvious place for second HQ, given metro access to DC, Dulles airport, the millions of datacenters, fios penetration and large numbers of local it workers.
BUT I think it's fair to point out that Amazon going anywhere where it represents a large change in the population will have a notable housing impact. Construction lags behind demand and the current markets in high tech areas have lots of problems with difficult local regulations, extreme rents, very limited availability, and general gentrification issues. Amazon will likely be a net positive for wherever they go, but it's a good idea to anticipate and try to reduce the inevitable negatives.
I want to disagree with your NoVa conclusion, but I can't. It has location, population, infrastructure, though transportation is notoriously bad - almost always ranked worse for traffic than San Francisco and Seattle, usually beating Boston, and challenging LA depending on what you measure. (that's from memory and a few google searches to make sure I'm not crazy, take with salt) North Carolina would be the next natural match, but I don't see Amazon taking on that political mess.
If Amazon moves to Podunksville, it will "work". Given that they're looking for deals, it will almost certainly be successful as far as "does the office function, is it a net positive to our productivity". But what is already at their new location will definitely impact what happens for the first few years as the local economy shifts. Pulling in 90% of their workforce from remote with a tiny hub airport that has only 2 non-stop destinations is very different from pulling 50% of their workforce from remote with an international airport with non-stop flights to most major cities. The reverse effects are true as well - getting people in where the local cost of living makes every dollar worth $1.20 grants a benefit vs a local cost of living giving you $0.90 spending power per dollar.
Heck - Moving to Seattle was wonderful for my wife and I in lots of tiny ways: People aren't confused when asked about vegetarian options, assuming it's not already clear on the menu. Our various hobbies and entertainments are well represented (Heck, I play tabletop RPGs and a good chunk of the major companies are actually _here_) It'd take a notable bit more to get me to go to somewhere where eating out is a chore, where I have to deal with the 5 local adults willing to play a "game" regularly and have to put up with their quirks because I have no other options, where I have to deal with religious evangelism on a daily basis, where you have to drive everywhere, where the weather is extreme, etc.
This isn't a question of success vs failure, it's a question of how to maximize their success. And the success of the local area, including the current residents.
1. Transportation: the vast majority of workers in NoVa drive to work, either to drive to Metro or drive to their workplace. Traffic is unacceptably bad now. Going from Bethesda to McLean, which is about a 15 mile drive, can take over an hour.
Solution: Amazon locates west of Manassas, in Front Royal or Leesburg.
This still puts them close to their us-east datacenters, but it’s still far from nightlife and city attractions for their workers. Maybe Richmond would be better, except there is a dearth of tech workers there.