What this demonstrates is a manufacturing method for producing nanoscale machinery. The concept is, with a series of simple experiments conducted in many labs, enough data is obtained to create an accurate nanoscale design simulation. (many of this experiments have been done already). You would then design a set of nanoscale machine parts - sensors, motors, gears, and so on in the software then create prototypes very laboriously and at very high cost using a tool like this microscope at IBM.
Once you have tested the prototypes and debugged them, correcting mistakes in the software model of them as you go, you'd then design a nanoscale machine that could place atoms down according to a template. You'd build that machine at even higher cost, using hundreds of microscopes like the one at IBM, and test it.
Anyways, a series of iterations later (and about a trillion dollars, realistically) you'd have a machine that could PRINT ITSELF and you no longer need the microscopes. By print itself, I mean it can slowly manufacture itself if kept in a clean vacuum chamber, supplied with clean DC power, atomic intermediates as substrate, and a high bitrate stream of the design files for itself from a control computer system. Nevertheless, exponential growth would be possible and you'd be very close to changing the world forever.