Right, that's why I wanted to make sure to add the caveat about fluid velocity. But the density of water is almost 1000x that of air so there's a large benefit to water at lower fluid speeds. A rough back-of-the-envelope calculation shows the crossover is when airspeed is about 9.2x the speed of the water flow under perfect theoretical conditions.
It's also worth noting that when dealing with centrifugal devices, the realized power never actually follows the power law. It's usually degraded quite a bit due to linear losses. A common power used in air systems is 2.1 rather than the cube; using that value, the airspeed must be almost 24x the water velocity. I've never personally seen anything greater than 2.7 (the higher numbers, IMO, tend to be used as a sales pitch rather than an engineering decision), but again, I don't work in these particular systems.
It's also worth noting that when dealing with centrifugal devices, the realized power never actually follows the power law. It's usually degraded quite a bit due to linear losses. A common power used in air systems is 2.1 rather than the cube; using that value, the airspeed must be almost 24x the water velocity. I've never personally seen anything greater than 2.7 (the higher numbers, IMO, tend to be used as a sales pitch rather than an engineering decision), but again, I don't work in these particular systems.