The problem wasn't buildings. The problem was new hires. Most of grants goes to pay people's salaries. More researchers means more grants means more money for everyone, so during the doubling most places used it to expand the number of researchers. So the available funds got absorbed on an increase in the number of people competing for it, then we were hit with an inflation rate cut, so the percentage of projects funded hit a sharp decline. This is especially true because grants span multiple years, so you must finish funding grants that were already accepted before you can fund new ones.
Think about it in terms of a bubble like the dot com bubble or the housing bubble. If you see the prices of something double, then suddenly stay flat, most other sectors would see people frantically running for the doors crying about a burst bubble. Of course people half way into a PhD can't exactly leave that fast, but the danger is that in the long run the same thing will result, with students running to other fields (perhaps less beneficial for the long term health of the nation) where they can actually have a chance of making a living.
A better approach would have been to double more slowly and then keep increasing funding at slower and slower rates each year until biomedical funding hit some target percentage of overall GDP.