in my humble opinion: supporting windows for the server side is like supporting IE 6 for the client side -> ressources best spend elsewhere
While I'm not an IE6 advocate, support for that browser generally comes down to the general percentage of it visiting your site. If there is convertable revenue there you will (99% of the time) end up supporting it; if not, that's your loss. This is a hard fact.
The same thing goes for Node. There are countless numbers of programmers operating on a Windows stack; the large number of Windows machines opens up Node to an adoption level that's actually far beyond what it could ever really gain with just Mac/Linux/BSD alone. It's possible Joyent has a vested interest in this, but I'd also wager it's just about being available to as many programmers as possible because the market is there.
i, as an egoistical dev-on-mac, would rather see a stable 1.0 for mac/unix soon, than a 0.8 for max/unix/win within the same timeframe.
as a thought experiment: would native max os x apps be as sophisticated as they are now, if all devs had decided to port them to windows (xp / vista / 7 / ...), too.