Per the "Sources" tab:
> Incidence rate estimates: Weekly age-specific incidence rates by vaccination status were calculated as the number of deaths divided by the number of people either fully vaccinated (cumulative) or unvaccinated (obtained by subtracting the cumulative number of fully vaccinated and current number of partially vaccinated people from the 2019 U.S. intercensal population estimates) and multiplied by 100,000. Overall incidence rates were age-standardized using the 2000 U.S. Census standard population. Rates are not adjusted for time since vaccination, underlying conditions, or other demographic factors besides age.
So I believe you're correct in that the data comparing deaths against the entire population, not against people who got covid.