Enjoying a 3 day weekend so I bread boarded the little thing this morning and it actually works just fine as drawn. I used an old Intel 875 PBZ board in an old P3 system that I use for experiments. The board actually had three fan speed control headers but now has two.
I toasted one but it matters not. Anyway I used the PWM signal to drive the circuit and the MOSFET doesn't even begin to get warm.
Something to note (and where I screwed up) is the PWM is tied to the fan Negative much as drawn in the add on circuit. The CPU fan positive is maintained at +12 volts constant and they PWM the negative side of the fan.
If you wanted to get slick about this rather than use the PWM from the motherboard you could drive the MOSFET using a 555 timer chip to generate a PWM signal and have a small pot to adjust fan speed for a half dozen fans or more. I powered three.