As it was explained to me, when the precursor to 430.122 came out in the 2002 code (can't remember the reference number but it changed to 430.122 in the next cycle), things had to change. This was the first time there is mention of different sizing rules for the conductors FEEDING a VFD and it created a perception of a conflict with motor lead sizing, meaning that because the input conductor size had to be based on the VFD input current, not the motor current, you might end up with the OCPD for the VFD becoming too large for the motor leads. Although nothing in that 2002 or subsequent wording ever said you had to oversize the OCPD, just the conductors, people were nevertheless sizing the OCPD based on the conductor size, sometimes rendering it too large for the motor feeders. So to help address this, UL changed the testing and listing requirements for VFDs, I think in time to coincide with the 2005 NEC release, to require VFDs to provide motor branch Short Circuit and Ground Fault protection, essentially so that the DOWN STREAM circuit components are protected regardless of what happens up stream, that way the motor leads are sized as we always have; NEC table FLA x 1.25. This situation by the way also fixed the problem of using a VFD as a phase converter, which almost ALWAYS creates that mis-match of SCPD sizing, because if you double the size of the VFD to hadle having a single phase input, the required minimum OCPD for the VFD usually becomes at least 2x the motor circuit size already!
Aside from the single phase issue, where this gets tricky is when you get into multiple smaller motors down stream of the VFD. At or below 15A isn't a big deal because you simply run 14ga wire anyway, even if each motor is only rated 1A or something. At 20A it's usually simplest to just run 12ga wire as well. Above that it starts to get more and more difficult because you start to get into whether or not a bunch of really small motors can even handle having larger conductors going into them. I used to run into that with tunnel freezer applications where the OEM would attempt to run 30 motors of 2HP each, all off of one 60HP VFD. The VFD SC protection could take care of the conductors expected for a 60HP motor, but was NOT going to protect a 2HP motor adequately.
The most common work-around used involves putting in the IEC style manual motor starters for each individual motor. These are often called Motor Protector Switches or Motor Protection Circuit Breakers, the little devices with an adjustable OL dial and a switch that also has the SCPD built-in. Since you must provide separate OL protection for each motor anyway, this device kills two birds with one stone, allowing you to take advantage of the feeder tap rules as it would apply to a splitter block installation. Size the wires from the drive to the splitter point as per the VFD output current (since that is what the VFD's SCPD function is set for), then use the smaller conductors necessary to go to each motor. Better yet, most of the MPCB lines now offer a feeder bus bar system that takes care of that line side connection as well.
The one caveat to this is that after 25+ years of use of these and other adjustable bi-metal overload devices behind VFDs, there is mounting evidence of problems with the high harmonic output of the drive causing additional heating in the bi-metal sensing strips, essentially making them trip too early when near full load current. From a pure protective standpoint early tripping is not bad, but the human response is the problem... turning up the dial to make the problem go away. This often results in eventual motor loss. If you understand this and can instruct everyone involved to resist the temptation, it's fine. If not, you have to go to melting alloy OL relays, which don't have the SCPD capability, so you are back to the motor branch circuit problem again.
By the way, fuses? No no no... Single phasing on the output of a drive could be very bad for the drive. Most decent drives will trip off, but there is no guarantee.