We study the Spitzer Infrared Array Camera (IRAC) mid-infrared (rest-frame optical) fluxes of 14 newly WFC3/IR-detected z=7 z_{850}-dropout galaxies and 5 z=8 Y_{105}-dropout galaxies. The WFC3/IR depth and spatial resolution allow accurate removal of contaminating foreground light, enabling reliable flux measurements at 3.6 micron and 4.5 micron. None of the galaxies are detected to [3.6]=26.9 (AB, 2 sigma), but a stacking analysis reveals a robust detection for the z_{850}-dropouts and an upper limit for the Y_{105}-dropouts. We construct average broadband SEDs using the stacked ACS, WFC3, and IRAC fluxes and fit stellar population synthesis models to derive mean redshifts, stellar masses, and ages. For the z_{850}-dropouts, we find z=6.9^{+0.1}_{-0.1}, (U-V)_{rest}=0.4, reddening A_V=0, stellar mass M*=1.2^{+0.3}_{-0.6} x 10^9 M_sun (Salpeter IMF). The best-fit ages ~300Myr, M/L_V=0.2, and SSFR=1.7Gyr^{-1} are similar to values reported for luminous z=7 galaxies, indicating the galaxies are smaller but not younger. The sub-L* galaxies observed here contribute significantly to the stellar mass density and under favorable conditions may have provided enough photons for sustained reionization at 7<z<11. In contrast, the z=8.3^{+0.1}_{-0.2} Y_{105}-dropouts have stellar masses that are uncertain by 1.5 dex due to the near-complete reliance on far-UV data. Adopting the 2 sigma upper limit on the M/L(z=8), the stellar mass density to M_{UV,AB} < -18 declines from rho*(z=7)=3.7^{+1.0}_{-1.8} x 10^6 M_sun Mpc^{-3} to rho*(z=8) < 8 x 10^5 M_sun Mpc^{-3}, following (1+z)^{-6} over 3<z<8. Lower masses at z=8 would signify more dramatic evolution, which can be established with deeper IRAC observations, long before the arrival of the James Webb Space Telescope. Comment: 6 pages, 3 figures, 2 tables, emulateapj, accepted for publication in ApJL