I'm assuming that aerodynamic forces are the only allowed sources of lift.
Aerodynamically, lift is obtained by redirecting airflow, which is essentially to accelerate air molecules. When you accelerate a mass $dm$ to speed $v$, you gain a momentum $dp$ given by
$$dp = v dm$$
Divide both sides by $dt$ to obtain the force:
$$F = v \frac{dm}{dt}$$
In order to hover the force needs to equal the weight of the flying machine,
$$Mg = v \frac{dm}{dt}$$
Now consider the work done by the engine to accelerate the air molecules:
$$W = \frac{1}{2}dm v^2$$
$$\rightarrow P = \frac{1}{2} \frac{dm}{dt} v^2$$
I assume that $v$ and $\frac{dm}{dt}$ remain constant. If that bothers you, replace these quantities with suitably defined averages.
Now we can combine everything to find:
$$P = \frac{1}{2} Mg v$$
So as long as we design our flying machine with $v$ as small as possible (and, therefore, a mass flux as large as possible), we can make the steady state power as small as we like. The only constraints come from engineering considerations that limit how large a craft can be made at a given weight, material strength, parasitic drag which I've neglected, etc.