Rapid determination of molecular structures can greatly accelerate workflows across many chemical disciplines. However, elucidating structure using only one-dimensional (1D) NMR spectra, one of the most widely used techniques for the characterization of organic compounds and natural products, remains an extremely challenging problem because of the combinatorial explosion of the number of possible molecules as the number of constituent atoms is increased – for example for molecules with up to 36 non-hydrogen atoms, the number of possible structures consistent with the typical bonding rules of chemistry has been estimated to range from 1020-1060. The task of determining the molecular structure (formula and connectivity) of a molecule of this size using only its one-dimensional 1H and/or 13C NMR spectrum, i.e., de novo structure generation, thus appears completely intractable. I will show how it is possible to achieve this task for systems with up to 40 non-hydrogen atoms across the full elemental coverage typically encountered in organic chemistry (C, N, O, H, P, S, Si, B, and the halogens) using a deep learning framework, thus covering a vast portion of the drug-like chemical space. Leveraging insights from natural language processing, I will show how our transformer-based architecture can overcome the combinatorial growth of the chemical space while also being extensible to experimental data via fine-tuning.