AlphaFold: Five Years of Impact (deepmind.google)

Five years on, AlphaFold has shifted structural biology from slow, labor‑intensive structure determination toward rapid, AI‑driven prediction and hypothesis testing. The original AlphaFold and AlphaFold 2 have been cited in over 35,000 papers and influenced methodology in 200,000+ publications; an independent Innovation Growth Lab analysis found users submit 40% more novel experimental protein structures, with those structures tending to be more dissimilar to known folds. Work linked to AlphaFold 2 is twice as likely to be cited in clinical articles and more frequently appears in patents, and practical applications—from plant genomics to crop‑resilience studies—demonstrate accelerated discovery timelines and higher‑quality outputs. The open AlphaFold Server has already produced more than 8 million fold predictions for thousands of researchers, democratizing access to structure prediction. Building on that momentum, DeepMind and Isomorphic Labs have pushed the platform into “digital biology” with AlphaFold 3 and a unified drug‑design engine. AlphaFold 3 extends prediction beyond single proteins to DNA, RNA, ligands and entire molecular complexes, enabling joint 3D models of targets, genetic material and small‑molecule binders. Technically, this means holistic structural context for rational drug design, better in silico screening of binding modes, and faster translation from molecular hypothesis to experimental validation—potentially reshaping discovery pipelines in academia and pharma and accelerating the path from structure to clinic.