Also note that the velocity cannot be increased to infinity, as it is limited by the first law of thermodynamics (even though the Mach number can go to infinity in theory). If your flow starts from rest, then it's total enthalpy is
h_0 = c_p T_0.
For an isentropic system (as is typically the case for nozzles), the first law of thermodynamics states, as previously mentioned by
@Chestermiller, that
\Delta h = 0,
so between the initial rest state and any point in a nozzle, you have
c_p T_0 = c_p T + \dfrac{u^2}{2}.
In the case where the flow has been expanded so far as to reach absolute zero (##T=0##), which is the limiting factor on velocity, this gives
c_p T_0 = \dfrac{u_{\textrm{max}}^2}{2},
or
u_{\textrm{max}} = \sqrt{2 c_p T_0}.
So, the maximum velocity in a nozzle is limited by your starting temperature, ##T_0##, and the specific heat at constant pressure, ##c_p##. In that sense, yes, a nozzle converts the internal energy into velocity, and the maximum attainable velocity scales as the square root of these two variables and shows marked diminishing returns.
Of course, you also need a pressure difference to make a nozzle operate, and this pressure ratio, ##p_0/p_e##, grows very rapidly as you increase the Mach number of a nozzle. So, a nozzle is theoretically limited by the total temperature (total enthalpy) of the working gas, but in practice is generally limited to the pressure you can obtain (safely) in your reservoir.
Either way, though, you aren't going to be able to use a converging-diverging nozzle to decrease the power consumption of a compressor located upstream in any useful sense. For one, a converging-diverging nozzle is going to undo the compression obtained by the compressor, as the purpose of such a nozzle is to
expand the flow to a higher velocity/Mach number and lower pressure, temperature, and density.
Second, once the nozzle "starts", the compressor can receive no information about what is occurring downstream of the throat of the nozzle. If the throat is sonic and the expansion is supersonic, then information about that downstream flow, which can only travel at the speed of sound, cannot travel upstream faster than the air is coming out of the nozzle. Essentially, a nozzle operates only in one direction, so whatever outlet pressure is supplied by the compressor is going to drive the nozzle.
Now, for a given reservoir pressure, a nozzle may operate in one of several modes depending on the downstream conditions. Assuming the compressor supplies a pressure high enough to choke the throat, then the mass flow rate through the nozzle is no longer sensitive to downstream conditions and depends only on the reservoir temperature and pressure. If the compressor is not capable of keeping up with this mass flow rate, then the nozzle won't function properly because it will try to draw gas away faster than the compressor can supply it and you will end up with subsonic flow throughout. If the compressor supplies gas faster than the nozzle can expend it, then you might see a reduction in power consumption since the compressor doesn't have to supply gas as fast as it possibly can, but I don't see the use here.
So, I guess then the operative question is this: what exactly are you hoping to use this compressor for? You might reduce the power consumption slightly, but all of that gas is going to end up just moving through the nozzle and being expanded back to ambient pressure, so I am not sure that you will get any use out of it unless that use is simply making a supersonic wind tunnel.