I think the idea is tic-tac sized probes with nano circuitry (that doesn't exist yet)

accelerated by lasers so they don't have to carry the power source

Obviously stopping is the problem, they can never stop but at some point no need

Before someone mentions IBM's (completely BS) "sub-nanometer circuitry": it isn't nearly sub-nm, except if you look at it from one particular way. The real estate is still many nanometers in dimension. The gates aren't even sub-nano; gate-to-gate interactions are, kind of.