Infant acute lymphoblastic leukemia harboring the fusion oncogene MLL-AF4, which arises in utero during embryonic development, is characterized by its dismal prognosis and short latency. The mechanisms of transformation are not amenable to analysis with patient samples because cancer is studied once the transformation events have already occurred. Many mouse models for infant leukemia have fallen short in achieving the goal of illuminating the human disease because they do not recapitulate key aspects of the actual human disease, indicating that the mouse model is missing essential ingredients of oncogenesis present in the human embryo. Here, we review the disease models currently available and propose the use of human embryonic stem cells as a scientific opportunity for modeling infant cancers with possible embryonic origin.