Following an analysis similar to Leigh's I convinced myself that
P_max = Torque on gyro * angular velocity of shaft wrt earth
= d(ang mom of gyro)/dt * w_e/2
= (ang mom of gyro * w_e/2) * w_e/2
= I_g * w_g * w_e^2 / 4
where I_g is the rotational inertia of the gyro about its spin axis,
w_g is the rotational velocity of the gyro about its spin axis, and
w_e is the rotational velocity of Earth.
To get some numbers, I considered a steel disk 1 m thick and 3 m in
radius spinning at 10 cycles/s (!?). Then
I_g = M R^2 / 2 = pi rho h R^4 / 2 ~ 1 MJ s^2
K_g = I_g w_g^2 /2 ~ 2 GJ
and
P_max ~ 100 mW
In other words, the rotational KE in the gyro all by itself would be
sufficient to supply the expected maximum power for many centuries.
Moreover, frictional losses would have to be kept to an impossibly
small amount. Most importantly, the capital expense of this 100 mW
generator would be, shall we say, prohibitive!