We consider alternative history scenarios in which explicit climate mitigation begins before the present day, estimating the total costs to date of delayed action. Considering a 2(1.5) degree Celsius stabilization target, peak costs are greater and reached sooner with a later start to mitigation, reaching 15(17)% of global GDP in 2085(2070) for a 1990 start and 18(35)% in 2080(2035) for a 2020 start. Further mitigation delay costs a best estimate of an additional 0.5(5) trillion dollars per year. Additional simulations show how optimal mitigation pathways evolve without imposing a warming limit, finding that median abatement levels and costs are not strongly dependent on start date. However, whereas 18(5) percent of optimal solutions starting in 1980 meet the 2(or 1.5) degree target, 5(or 0)% of 2020 simulations meet the goals. Discounted damages due to delayed mitigation action rise by 0.6 trillion US dollars per year in 2020.