High-mass star formation, a key process in controlling the energy budget of galaxies, remains one of the unsolved problems in astrophysics due to the observational difficulties caused by their short evolutionary timescales and the distant locations where they form. The primary debate in high-mass star formation revolves around two competing models (Fig. 6.1): the monolithic collapse of a high-mass core [core-fed; 1] and the accretion of gas onto low-mass cores [clump-fed; 2, 3, 4]. To address this issue, we conducted a statistical study of the cores embedded in IRDCs using ALMA observations, focusing on the core properties prior to the formation of high-mass protostars. IRDCs are massive, dense, and cold regions, the best candidates for the birthplace of high-mass stars. In particular, clumps (the denser parts of IRDCs) without bright emission at <70 µm are thought to lack even embedded protostars and are the best candidates for studying the initial conditions of cluster formation.