The math here is doable. Approximately the inner angle of the cone of light comming from a laser beam is 1.22 2lamda/d (source: airy disk wikipedia). Where lamda is wavelength and d is the aperture.

Lets pick lamda so 1.222lamda is 1000nm (blueish light). Then by smal angle approximation, the size of a beam at 500km (starlink height) is 500km * 1000nm / d

Say we want our beam to be 50m wide. That gives an aperture requirement of 1cm.

Huh, that seems reasonable. So in a vacuum, it could perhaps be done. Atmospheric effects would make this worse ofcourse. And a 50m beam is not going to dump much heat, but it seems somewhat plausible in the near future.