By examination, the first function is 0 for all t < 2000 . That cannot possibly be 0.876. So the question comes down to finding (1-((1- wblcdf(t-delta12_origin,eta_hat_2,beta_hat))./(1-Age_t1_origin))) == 0.876 and if the answer is < 2000 then discard that answer.
F_2_origin_cond = @(t) (1-((1- wblcdf(t-delta12_origin,eta_hat_2,beta_hat))./(1-Age_t1_origin))) - 0.876;
t_target - fzero(F_2_origin_cond, [max(2000,deltal2_origin), inf])
The max() is there to select between the greatest of 2000 (the part at which the equation becomes valid) and delta12_origin, as wblcdf(t-delta12_origin,...) will be 0 for t < delta12_origin .
