Neonatal sepsis (NS) kills 750,000 infants every year. Effectively treating NS requires timely diagnosis and antimicrobial therapy matched to the causative pathogens, but most blood cultures for suspected NS do not recover a causative pathogen. We refer to these suspected but unidentified pathogens as microbial dark matter. Given these low culture recovery rates, many non-culture-based technologies are being explored to diagnose NS, including PCR, 16S amplicon sequencing, and whole metagenomic sequencing. However, few of these newer technologies are scalable or sustainable globally. To reduce worldwide deaths from NS, one possibility may be performing population-wide pathogen discovery. Because pathogen transmission patterns can vary across space and time, computational models can be built to predict the pathogens responsible for NS by region and season. This approach could help to optimally treat patients, decreasing deaths from NS and increasing antimicrobial stewardship until effective diagnostics that are scalable become available globally.