cookyy2k said:
The problem with shrinking transistors any further is quantum tunneling. Any smaller and the tunnelling current basically fries them.
True, though we'll learn how to build transistors on the quantum scale long before we get scalable quantum computers.
It might involve a radical redesign of the semiconductor junctions we use now, for example some researchers have proposed building classical cellular automata with quantum dots:
http://nd.edu/~qcahome/
The bits are stored as confined electrons, and computations are done through the bits interacting electrostatically.
cookyy2k said:
As for this new advance the theory is great and all but the cost will be huge, they may be available in 10 or so years but affordable? STMs are not cheap pieces of equipment to buy, maintain and use.
Agreed, the real innovation would be a fast method for reading and writing these kinds of ultra dense atomic memories. Parallel to the discussion about building bigger processors, achieving a huge storage capacity is as simple as keeping 100 1TB drives, the real breakthrough would be a 1 TB drive with the kind of access times we are used to seeing in RAM.