I have to think that computational models will be infinitely cheaper than animal testing. If they worked as well as animal testing surely some for profit pharmaceutical company would leverage them to reap enormous profits. Because this is not happening I conclude the the models aren't as affective.

But it literally isn't here. We need live organisms to test drugs, we can't do it all on computer chips and disconnected cells

If it was just inertia, it means the options have similar outcomes and the only factor is the cost of switching. As you pointed out, there are other trade-offs.