In 2020, Google DeepMind’s AlphaFold2 revolutionized science by predicting protein structures with near-experimental accuracy, solving a 50-year biology grand challenge. This breakthrough earned Demis Hassabis and John Jumper the 2024 Nobel Prize in Chemistry, with AlphaFold2 now used by over 2 million researchers across 190 countries to map 200 million proteins. AlphaFold 3 followed in 2024, predicting protein-DNA-RNA interactions with 50% better accuracy.
Today, Zuckerberg’s Biohub has surpassed AlphaFold with ESMFold2, an open-source “world model of protein biology” trained on 2.8 billion sequences that outperforms AlphaFold3 in protein-protein and antibody-antigen prediction. The new ESM Atlas maps 6.8 billion protein sequences and 1.1 billion predicted structures — eclipsing AlphaFold’s database by 800 million entries.
The open-source nature of these tools creates exponential impact: researchers worldwide now access state-of-the-art infrastructure without expensive licenses, accelerating discovery from years to months or even weeks. Biohub’s $500 million Virtual Biology Initiative puts this engine in everyone’s hands, enabling rapid iteration on disease targets that previously took decades to understand.
Hassabis boldly predicts AI could cure all diseases within a decade by cutting drug discovery from 10 years and billions of dollars to months or weeks—a vision now visibly unfolding. Between AlphaFold’s foundation, Biohub’s open world model, and Isomorphic Labs’ drug-design AI, we’re inching toward “radical abundance” where tailored treatments become routine. The journey from first prediction to disease-ending therapy has compressed from generations to years.
AI ISN’T JUST ACCELERATING DRUG DISCOVERY—IT’S DEMOCRATIZING THE CURE FOR ALL DISEASES.
Sanjay Sahay
Have a nice evening.