Licensing/Cloning

things are going to change too fast on your server to make keeping a dead box in storage very good method of disaster recovery imho. keep your installation software and backup data copied offsite (takes discipline) and be ready to rebuild onto new hardware you can order overnight from dell, imho. that is just as fast a way to recover smashed server room as way you propose and cheaper and i got new server hardware now instead of poopy old server that was in closet for 5 years and i can't get the drivers for now.... keep live failover server on-site imho. you can use it to test new stuff on in addition to being failover server. ms has it all built in to keep it updated and fail it over. check it out at technet at www.microsoft.com . ask for training ask for books ask for technet subscription.
even tho you don't protect aganst building disaster using on-site box you protect wonderfully well against hardware/software failure on the server which is higher probablility of happening imho. this server failure tolerant configuration you want is all built into widows 2003 now...but they expect you to have failover server on premises (simply a second ad server as you guessed)....they also have it built in for mirroring a remote server which is your idea, just add power and internet to failover server (ask to keep it at your house?)think what you want is to check out clustering servers at support.microsoft.com. complete how-tos are available to do exactly as you wish.
you could consider keeping failover box running in another location ready to 'failover' if main box goes down. you could always schlep it over to main building but why bother? search www.microsoft.com for 'disaster recovery howto'

WOW, the Sgt really hit a lot of good points, but I wanted to add
a few.

You solution leaves a few problems. The license allows one
install. A second install offline would still be illegal I would think.
Also, The idea of a pre-installed box does not allow for
matching the patches and SP's and software updates etc the
main server would have had applied to it. The second server
would be better as a parts box. I have known many to keep
duplicate hardware and if a box dies pull the matching part.
This is not the best solution by far.

So, I would suggest a few backup systems.

One would be the RAID system of your choice. If the system is
mission critical and avoiding down-time is key I would do RAID
5 with an online spare. Any HDD failure and the system would
continue. Even without the online-spare, as long as there is
spare drive it could replace the failed drive and regenerate the
RAID. RAID 1 is fine but slower than the RAID 5 too. Also RAID
1 may require downtime in the event of HDD failure, short
time, but some.

Two, I would have a good backup of all data and even a Disaster
Recovery backup if possible. ALWAYS use some form of off-site
storage for backups.

Three, might Ghost the system, and update the image anytime
you make OS changes such as SP's or patches. Now you could
recover the base system with the image, or the backup, and the
data with the backup.

Four, and this is just my preference, I would use the "My Drivers"
shareware product to backup all your drivers.

The last option, and I think your best solution, is to use a
Cluster. The features depend on your server version, but as the
SGT mentioned, this would do what you want and the only cost
addition would be another server license and some thought
about your CALS being user or device. Microsoft has a great Doc
file called BDMTDM.doc that you should read through. I will put
the link below but I am again on the Mac and the URL will