We explore whether close-in super-Earths were formed as rocky bodies that failed to grow fast enough to become the cores of gas giants before the natal protostellar disk dispersed. We model the failed cores inward orbital migration in the low-mass or type I regime, to stopping points at distances where the tidal interaction with the protostellar disk applies zero net torque. The three kinds of migration traps considered are those due to the dead zones outer edge, the ice line, and the transition from accretion to starlight as the disks main heat source. As the disk disperses, the traps move toward final positions near or just outside 1~au. Planets at this location exceeding about 3~M$_oplus$ open a gap, decouple from their host trap, and migrate inward in the high-mass or type II regime to reach the vicinity of the star. We synthesize the population of planets formed in this scenario, finding that some fraction of the observed super-Earths can be failed cores. Most super-Earths formed this way have more than 4~M$_oplus$, so their orbits when the disk disperses are governed by type II migration. These planets have solid cores surrounded by gaseous envelopes. Their subsequent photoevaporative mass loss is most effective for masses originally below about 6 M$_oplus$. The failed core scenario suggests a division of the observed super-Earth mass-radius diagram into five zones according to the inferred formation history.