[zahlman]

>humans tend to improve tech at exponential rates

A hypothetical "death star" project like this would require improvements in energy generation/storage capacity/etc., which haven't improved in nearly the way transistor production has (and are also much more limited by physical realities, such as the specific heat, enthalpy of combustion etc. of materials).

[hnuser123456]

Yes, extremely high sustained power lasers still have a hard time competing with hypersonic projectiles in energy delivered. The difference in being able to throw nuclei at the problem.