I think you are misunderstanding how the speaker is rated... The LF driver is a KappaLite 3012LF which is rated by the manufacturer at 450 watts "RMS". The mid and/or HF driver do not contribute significantly to this number. There is an artificial "program power" rating that is used by the industry as an attempt to represent the additional 3dB specified crest factor in the continuous power test (which only requires that the driver survives for 2 hours by the way), and the program power is defined as 2x the "RMS" (also called continuous power) rating.
The reason I bring this up is that it's clear from your comment of 900 watts being "conservative" that you do not understand how this number is derived, and that you need to compare the power rating of the speaker and the power rating of the amp based on the same units of measure.
So, assuming the amp is rated at 2500 watts RMS at 4 Ohm's, you need to compare the speaker using the same units. Therefore, 2 of the F112's wound be rated at 900 watts ("RMS", 4 ohm load) connected to an amp that is rated at 2500 watts "RMS" into the same 4 ohm load which represents a matching that I would be very cautious of. Even assuming the drivers can withstand an additional 3dB of crest factor (2x the "RMS power, IME that's beyond their mechanical power handling, and I'm very familiar with this driver), that would mean that the drivers could handle only 1800 watts whereas the amp can deliver 2500 watts. I don't see how the word "conservative" can be used to describe this situation.
I went through this math because I have seen a fair amount of misunderstanding over power handling between dissimilar power rating methods, as well as the damaged drivers that can result from such misunderstandings.