VMware HA does great with your typical application / web / file / print etc servers. You lose a host, within 60-90 seconds the VMs that were on the lost host are running again on a different host, provided you are N+1 or better. Database servers like Exchange, SQL, Oracle etc will not be as tolerant of a host failure, but with a little work you will get them back up and running reasonably quickly.
As already mentioned, FT will only work with uniprocessor currently and all FT will protect you against is a host failure. A failure in the VM protected by FT, will be replicated to the 2nd VM immediately therefore if you are trying to protect against a bad patch or something, FT will give you nothing. As soon as your source VM dies or BSOD's so will your secondary VM.
Again, I am not an Exchange admin. I hung that hat up shortly after we migrated from 5.5 to 2003. So I am not familiar with all the current Exchange resiliency options available today. But I can tell you that if you must have five 9's or better of uptime, even if FT supported SMP, you would still not be guaranteed that. Physical servers with multiple, well distributed DAGs is probably a much better option.