Since the dawn of agriculture, cultivated soils have lost a vast amount of carbon to the atmosphere. Most of the carbon-rich organic matter in soil is microbial necromass, shaped by the traits of diverse organisms. To understand how soil microorganisms lead to stable soil carbon, it is critical to understand how microbial ecophysiological traits are linked to soil organic matter formation, and how cross-kingdom interactions—involving bacteria, fungi, archaea, protists, microfauna and viruses—shape soil carbon availability and loss. Yet our current ability to predict ecosystem processes from microbiome data is poor. Why? I will suggest several key reasons, and will present results from studies where we have used quantitative stable isotope probing (SIP) and metagenomics/transcriptomics to assess growth and mortality of wild microbial and viral communities and show how niche differentiation in space and time drives the soil carbon cycle.