Under what plausible circumstances would you consider DNA-based or chemistry-based computation to be a superior solution compared to silicon-based computation?
And nowadays ASiCs for accelerating deep learning is the best we have for massively parallel computation, I think.
It's more about thinking what we could do if we had gazillions of little things swimming in a solution or sitting on a die that could do computation for us. For me one important result is that it doesn't scale for huge NP complete problems. Add a few hundred more variables to a SAT problem and you don't have enough water in our whole world.
This reminds me of the old story of doubling the amount of rice for each square on a checker board.
Aside from this, imagine taking a "smart" medication that uses biological computation to determine exactly where/when in your body to optimally release the medicine.