Antimicrobial peptides (AMPs) hold significant potential as broad-spectrum therapeutics due to their ability to target a variety of different pathogens, including bacteria, fungi, and viruses. However, the rational design of these peptides requires the molecular understanding of properties that enable such broad-spectrum activity. In this study, we present a computational analysis that utilizes machine-learning methods to distinguish peptides with single-target activity from those with activity against multiple pathogens. By optimizing a feature-selection procedure, the most relevant physical-chemical properties, such as dipeptide compositions, solvent accessibility, charge distributions, and optimal hydrophobicity, that differentiate between narrow-spectrum and broad-spectrum peptides are identified. Possible molecular scenarios responsible for the universality of these features are discussed. These findings provide valuable insights into the molecular mechanisms and rational design of multitarget AMPs.