Monolithic optical frequency combs are sets of equidistant spectral components generated through the optical pumping an ultra-high Q whispering-gallery mode resonator with Kerr nonlinearity. They are particularly characterized by their conceptual simplicity, small size, and very low power consumption. This is why they are expected to replace femtosecond mode-locked lasers in applications where these features are of particular relevance. From a fundamental perspective, the theoretical understanding of Kerr comb generation is currently the focus of a worldwide and rapidly thriving research activity, motivated by the wide range of potential applications (metrology, sensing, aerospace and communication engineering). In this talk, we will review the main theoretical challenges related to this optical system, and propose new pathways that would enable to understand the complex light-matter interactions originating from the strong confinement of laser light in nonlinear media.