I think its cubic feet per hour
Okay, I went and googled the term like I should have and here's what I found:
Converting air flow velocity "V" to cubic feet per minute "CFM" is easy!
So, by their formula, an 80mm Panaflo L1A which provides 24 CFM over a surface of Pi*80mm^2 would have a FPM rating of:
Area in Sq Ft. x FPM = CFM
Pi * (80mm / (25.4mm per i x 12 i per ft))^2 x N FPM = 24 CFM
0.216 sq ft x N FPM = 24 CFM
N = 110 (units cancel)
So, extrapolating from the graph for the heatsink linked above, we see that with the heatsink in an "open duct" configuration (which I assume describes an attached or nearby Panaflo better than "closed duct"), you get a thermal resistance of about 0.47C/W, which puts it about dead even with the Zalman 6000Cu. (Based on the thermal resistance of the Zalman in the SPCR heatsink roundup with a Panaflo at 12V.)
Anyway, the point is that the "passive" coolers you see around designed for server systems aren't "passive" in the sense that they require no airflow. It's more that they are designed for server farms that have crazy ductwork that forces air into the systems, without the need for tiny, breakage-prone fans all over the place.
Sorry to put a damper on the party, but if you want to dissipate the heat from a 30W+ CPU, you need airflow (or waterflow). Although anyone who wants to prove me wrong is welcome to try.