Your estimations assume way bigger uptime cost that it should. First, the team is way larger now, so maybe new people wouldn't be required at all. Secondly... what developers, really? We're talking here about just a "trip to the past" project, that would be not developed in any meaningful way. We don't need the same legion of GMs as we do in retail, this is a side project, people go and play at their own risk, so to say - and this is the way many enjoy it now, so it seems. And finding all those assets and code... I'm no genious here, but all companies I've worked in (as in... 2) kept their software versioned ALL THE TIME. There is no reason not to keep every single version of your releases. And I mean companies that hired like, 10 people. So what, would Blizzard have lower standards all of a sudden?
And you estimate 20k players... I'm not sure about this, but multiply this by 5. At least. That's my estimate for players who would gladly return to the game they once loved, who would leave private servers (there is more than a second WoW playerbase among those) to play on a working, safe and secure server, even paying a small fee every month. This isn't a thing we will be able to see for ourselves anytime soon, but with 3 first releases of the game (vanilla, bc, WotLK) being availible for playing, 100k players might be the number we're looking for at least.