MIT researchers trained AI models to learn yeast DNA sequences and optimize protein production workflows, aiming to lower costs in vaccine and biopharmaceutical manufacturing. The model decodes regulatory sequence features to boost yield in industrial yeast strains. Separately, researchers led by César de la Fuente at the University of Pennsylvania are using AI to mine genomic and natural-product space for novel antimicrobial peptides, including candidates from archaea and extinct-species sequences. Their approach seeks to replenish a thin antibiotic pipeline by uncovering molecules difficult to find via traditional screening. Together, these studies exemplify AI’s bifurcated impact on biopharma: improving biomanufacturing efficiency and expanding small‑molecule/peptide discovery funnels—two avenues with near‑term translational potential.