A major unresolved question in the origin of life is whether there exists a continuous path from geochemical precursors to the majority of molecules in the biosphere, due in part to the autocatalytic nature of metabolic networks in modern-day organisms and high rates of extinction throughout Earth's history. Here we simulated the emergence of ancient metabolic networks to identify a feasible path from simple geochemical precursors (e.g., phosphate, sulfide, ammonia, simple carboxylic acids, and metals) to contemporary biochemistry, using only known biochemical reactions and models of primitive coenzymes. We find that purine synthesis constitutes a bottleneck for metabolic expansion, and that non-autocatalytic phosphoryl coupling agents are sufficient to enable expansion from geochemistry to modern metabolic networks. Our model predicts distinct phases of metabolic evolution, characterized by the sequential emergence of key molecules (carboxylic acids, amino acids, sugars), purines/nucleotide cofactors (ATP, NAD + ), flavins, and quinones, respectively. Early phases in the resulting expansion are associated with enzymes that are metal-dependent and structurally symmetric, consistent with models of early biochemical evolution. The production of quinones in the last phase of metabolic expansion permits oxygenic photosynthesis and the production of O 2 , leading to a >30% increase in biomolecules. These results reveal a feasible trajectory from simple geochemical precursors to the vast majority of core biochemistry.