With new and upcoming observing facilities (JWST and the ELTs), the exoplanet community is poised to precisely measure the chemical inventory of exoplanet atmospheres. This will allow, for the first time, to start investigating whether one of the greatest promises of atmospheric characterization studies holds up: inverting the atmospheric composition to infer the planet formation history encoded in it. In my talk, I will show how such measurements allow to run so-called formation retrievals, which constrain a planet’s formation history using its atmospheric abundances in a Bayesian retrieval framework. I will demonstrate how simple and popular models for the composition of the protoplanetary disk and planet formation could lead to interesting insights when applied in formation retrievals. At the same time, I will discuss how such assumptions are too strongly simplified for making the exoplanet atmosphere — formation connection in practice, and what the most pressing theoretical challenges are. Achieving this connection will be a formidable and interdisciplinary challenge, but the exciting exoplanet observations that lie ahead will allow the community to tackle it in earnest.