Re: re: No word on actual data transfer rates.
It also sounds like it's not random-access either, so you'll just have a soup of DNA strands and have to decode them all, to get the data you want.
I suspect the time to decode an exabyte of data will be pretty high.
an entire human genome is about 3.1-3.2 billion base pairs
At 2 bits per base-pair that's about 0.8GB.
So your sugar-cube exabyte could encode the DNA of about 1.25bn humans without compression. Allowing for the huge amount of duplication in that dataset, it should easily be able to encode the DNA of the whole human race.
The time and cost of sequencing the entire human race should give you an idea of the cost of data retrieval.
OTOH: having a backup of the entire human race in a sugar cube is an interesting idea :-)