I think you should try to make all necessary corrections to use your cooling systems efficiently, as a first measure (i.e. seal those holes in the floors since they cause cool air waste, re-locate the air ducts as needed to optimize air flow, put the hot air return in place for that rack, and look for any other problems that may exists such as damaged air ducts, etc…).
On the other hand, I don’t think one should talk about the ‘datacenter’ temperature, because you have a very high temperature in the exhaust side of the racks, which in fact affects the overall room temperature, but it is the temperature of the incoming air which really imports, because it is the one that will cool the equipment, so the temperature should be measured in the front side of the racks, and 100 degrees is too high (I think it should not be over 75 degrees F).
Also, there could be room for server consolidation and virtualization, which would redound in less cooling (and power) requirements.
Also, you need to plan for the future, based on growing expectations.
There are merits to both arguments. If the a/c units are already maxed out, there is no redundancy if the a/c rack fails. but before anything, you should seal up your floor tiles.