Abstract: Molecular gas (H2) is the primary fuel of star formation, and it is critical in determining both the morphology and evolution of galaxies, especially for the high redshift systems. The most widely used tracers of H2 gas mass are the emission of CO molecular lines and dust continuum, by using the CO-to-H2 conversion factors and the dust-to-gas mass ratios, respectively. These methods, however, have several caveats, which can not be calibrated or corrected. On the cold end of the thermal states possible for the H2 gas in galaxies, the rising Cosmic Microwave Background (CMB) will render the H2 gas distributions and the corresponding velocity fields, making it very difficult to image in dust continuum and low-J CO lines for z>3. And on the warm end, which is mostly heated up by intensive star formation, it now seems that the Cosmic Rays (CRs) can destroy the CO molecule in star-forming galaxies very effectively, making large H2 gas distributions (CO-line)-invisible. I will compare the emission of atomic carbon (CI), which seems to be the most promising alternative H2 gas tracer, with those of CO and dust continuum. I will also present our recent ALMA results in nearby starburst galaxies to testify the robustness of the CI-based method. In the end I will briefly discuss the implication of our results for the stellar initial mass function.