It's by no means a rock-solid conclusion. There are various data showing good returns, but not well matched to particular individuals' ex ante prospects.

For example, the combined wealth of all technology billionaires firmly surpasses that of all finance billionaires, and if one looks at fortunes earned in the last several decades the disproportion is overwhelming.

The mean returns of venture-backed entrepeneurs are good (while the risk-adjusted returns look comparatively poor), but that's a highly selected subset of entrepreneurs and startups (some discussion on 80000 hours).

The evidence for finance is more robust than for tech entrepreneurship (although 80,000 hours is trying to get a better picture).