I think the problem is many people assume that "nano" is a sort of magic solution you can sprinkle over a problem like fairy dust.
Nano means "one billionth", so things at the nano scale are on the order of sizes of bacteria or Virii. A typical fusion reactor design like ITER or the laser fusion devices at Lawerence Livermore Labs fill factory sized buildings are are the size and cost of aircraft carriers. Even "new" fusion devices like Polywell or Focus Fusion are still large devices which fill rooms. Building devices these sizes from nanoscale assemblers really only adds quality control (precision placement of atoms), but the physical principles behind these machines isn't changed by this.
What we should be looking for is a way to leverage nanoscale phenomena to make a fusion reactor. The Foresight institute published a paper: Non-Statistical Fusion Reactions In Atomic Scale Accelerators
If we imagine the individual atoms as marbles, this can be thought of as taking a wooden board and carving straight channels for the marbles to roll down. By rolling the marbles fast enough at each other on the tracks or at fixed targets, they are much more likely to impact directly and have a much better chance to achieve fusion.
For actual power generation, we would have to create a device with hundreds or thousands of parallel channels to accelerate the nuclei at each other, and of course there will have to be some method of extracting energy from the thousands of individual reactions, but using nanoscale technology, we potentially can build a fusion reactor the size of a laptop computer (including all control and energy extraction equipment), which is certainly much more practical than something the size of an aircraft carrier.