Using DNA is using "atoms" themselves for storage. DNA is about 50 atoms/bit, and nature has given us machinery for replicating this data and even some types of error correction. And the linear structure and base-pairing action of DNA gives us the capability to perform data retrieval physically.

I will be surprised if anybody can come up with a nanotechnology scheme that comes even close to matching DNA's density in the next few decades. Look at the difficulties of even reading DNA, we can stick it through a nanopore, but it's not a trivial task to read off the various shapes of molecules as they poke through. Can you come up with a linear structure that is as stable but also allows for easier readout? If you start with DNA, but then add various modifications, how are you going to read out those modifications to the bases, and how are you going to replicate and copy that data?

I'm sure that it will happen that we eventually best DNA's density when it comes to storing information in covalent bonds, but our nanotechnology isn't even close to being able to come up with something like that now. Reusing biology's inventions are our only hope for the near future.