Over the last ten years, I played with Hadoop, GPU, FPGA and every kind of computing device to do bioinformatics. I also tried to answer biological questions using those computational tools. In my opinion, the biggest bottleneck of biology/bioinformatics is not larger and more efficient computer programs, but bioinformaticians trying to answer biological questions. Two biggest hurdles -
(i) bioinformaticians focus too much on the genome/transcriptome, which are only the informational blocks of the organisms and not the mechanical components.
(ii) Even those, who look into mechanical components, do not understand evolution and do not take evolution into account. That does not mean bioinformaticians are not developing tools to align sequences from multiple genomes, but that is very narrow perspective of evolution. The question of which sequences to align itself requires knowledge of underlying biological system, whereas those trying to develop tools seem to have the impression that they can build generic all-purpose tools. This is almost analogous to being a generic 'cook' or starting a generic restaurant to sell 'food' and not specific type of food.
I wrote a blog post on evolution vs mechanistic perspective of living organisms vs even narrower genome-focused view. Please feel free to criticize -