The leading candidate for dark matter that is thermally produced in the early Universe is the Weakly Interacting Massive Particle (WIMP). However, increasingly stringent bounds on WIMPs are now motivating the exploration of viable alternatives. One interesting possibility are DM candidates with sub-GeV masses. In this talk, I will present two such examples. First, I will focus on models where the dark matter abundance is set by mutual annihilations among multiple species. I will show how sizable mass splittings between the dark matter states naturally point to masses exponentially lighter than the weak scale. Light dark matter from coannihilation evades stringent bounds from the cosmic microwave background, but will be tested by future direct detection, fixed target, and long-lived particle experiments. Second, I will illustrate another viable thermal dark matter candidate with sub-GeV masses which has been overlooked in the literature: a cosmologically stable dark Higgs.