In January 2026, Tsinghua University researchers published DrugCLIP in Science, a system that screened 500 million drug compounds against 10,000 human proteins in under 24 hours on eight GPUs. The approach is 10 million times faster than conventional molecular docking, and the results are public in GenomeScreenDB, covering roughly half the human genome.
Rentosertib, Insilico Medicine's lead candidate for pulmonary fibrosis, showed patients gaining a mean 98 mL of lung capacity in Phase IIa trials while those on placebo lost 20 mL. Insilico plans Phase III for Q4 2026 and in June 2026 signed a deal worth up to $2.5 billion with South Korea's SK Biopharmaceuticals to develop drugs for neuroimmune disorders. Isomorphic Labs, which raised $2.1 billion in May 2026, expects to dose its first patient in an AI-designed oncology trial before year end.
Why it matters
The first AI-designed drug is heading to Phase III trials, and a $2.5 billion partnership shows the industry's bet is growing.
17 events
Latest: June 22nd, 2026 · 2 weeks ago
Showing 8 of 17
JK to step
Tap a bar to jump to that date
Jump to
June 2026
Insilico Medicine Signs $2.5 Billion Deal with SK Biopharmaceuticals for Neuroimmune Drugs
LatestIndustry
At the BIO 2026 International Convention, Insilico Medicine and South Korea's SK Biopharmaceuticals announced a collaboration to develop AI-designed drugs for neuroimmune disorders, including neuroinflammatory and rare neurological conditions. Insilico will use its Pharma.AI platform to design candidates; SK Biopharmaceuticals handles late-stage development and commercialization. Insilico is eligible for $18 million upfront and near-term payments and up to $2.5 billion in total milestones.
May 2026
Verge Genomics Rebrands as Verge Labs After ALS Drug Failure
Industry
Verge Genomics launched as Verge Labs after its AI-designed ALS candidate VRG50635 failed Phase 1 testing—a third of trial participants dropped out due to tolerability problems. The company laid off about 90% of its staff and shifted from drug developer to AI platform provider, helping pharma partners identify which patients will most benefit from experimental drugs.
April 2026
Rentosertib Inhalation Formulation Clears IND for Direct-to-Lung Trial
Regulatory
China's drug regulator cleared an inhalation solution of rentosertib for Phase 1 clinical testing, the first direct-to-lung trial for an AI-designed drug. Preclinical data showed the inhaled form achieves higher lung exposure with lower systemic absorption than the oral tablet.
January 2026
DrugCLIP Published in Science
Publication
Researchers from Tsinghua University publish DrugCLIP, demonstrating genome-wide virtual screening 10 million times faster than traditional docking, with validated hits including compounds more potent than existing antidepressants.
June 2025
Phase IIa Results for First AI-Designed Drug Published in Nature Medicine
Publication
Insilico Medicine published full Phase IIa trial results for rentosertib in Nature Medicine, the first peer-reviewed clinical evidence that a fully AI-designed drug works in patients. Patients on 60 mg daily gained a mean 98 mL in lung capacity over 12 weeks; those on placebo lost 20 mL.
May 2025
China Grants Rentosertib Breakthrough Therapy Designation
Regulatory
China's Center for Drug Evaluation granted Breakthrough Therapy Designation to rentosertib for idiopathic pulmonary fibrosis, the second major regulatory milestone for the drug after its 2023 FDA Orphan Drug Designation.
April 2025
Rentosertib Becomes First AI-Designed Drug Named by USAN
Regulatory
The United States Adopted Names Council granted the official generic name 'rentosertib' to ISM001-055, the first drug where both the disease target and the molecular compound were identified by AI to receive a formal international drug name.
November 2024
Recursion and Exscientia Merge
Industry
Two leading AI drug discovery companies complete their merger, combining phenomic screening with automated precision chemistry into an end-to-end platform with over 10 clinical programs.
First AI Drug Shows Efficacy in Phase IIa Trial
Clinical Trial
Insilico Medicine announces positive Phase IIa results for ISM001-055 in idiopathic pulmonary fibrosis—the first time a generative AI-designed drug demonstrates efficacy in patients.
October 2024
Nobel Prize Awarded for AlphaFold
Recognition
Demis Hassabis and John Jumper receive the Nobel Prize in Chemistry for developing AlphaFold, recognizing AI's transformative impact on structural biology.
May 2024
AlphaFold3 Expands Beyond Proteins
Breakthrough
DeepMind releases AlphaFold3, which predicts structures and interactions of proteins, DNA, RNA, and small molecules, enabling more comprehensive drug-target modeling.
June 2023
First Fully AI-Designed Drug Enters Phase II Trials
Clinical Trial
Insilico Medicine's INS018_055 becomes the first entirely AI-discovered and AI-designed drug to enter Phase II clinical trials, marking a milestone for generative AI in drug development.
February 2023
FDA Grants First Orphan Drug Designation to AI-Designed Molecule
Regulatory
The United States Food and Drug Administration grants its first Orphan Drug Designation to a molecule conceived entirely by AI, confirming such drugs can meet rigorous regulatory standards.
July 2022
AlphaFold Database Expands to 200 Million Proteins
Data Release
DeepMind releases predicted structures for over 200 million proteins, covering nearly all known proteins and making structural data freely available to researchers worldwide.
February 2021
Insilico Medicine Achieves 18-Month Drug Discovery
Milestone
Insilico Medicine identifies a novel target for idiopathic pulmonary fibrosis and advances a drug candidate to preclinical trials in 18 months—a process that typically takes 4-6 years.
November 2020
AlphaFold2 Solves Protein Structure Prediction
Breakthrough
DeepMind's AlphaFold2 achieves near-experimental accuracy at the CASP14 competition, solving a 50-year grand challenge in biology and enabling structure-based drug discovery at scale.
January 2020
First AI-Designed Drug Enters Human Trials
Milestone
A drug molecule designed entirely by artificial intelligence enters Phase I clinical trials for the first time, proving algorithms can create therapeutics worth testing in humans.
Historical Context
3 moments from history that rhyme with this story — and how they unfolded.
1 of 3
April 2003
Human Genome Project Completion (2003)
An international consortium announced the complete sequencing of the human genome after 13 years and $2.7 billion. The project identified approximately 20,500 protein-coding genes, providing the first comprehensive map of potential drug targets. Researchers predicted a flood of new therapeutics based on genomic insights.
Then
The pharmaceutical industry invested heavily in genomics-based drug discovery, expecting rapid returns that largely failed to materialize in the first decade.
Now
The genome provided the target list, but understanding protein structure and function remained bottlenecks. Only now, with AlphaFold predicting structures for the entire proteome, can researchers systematically screen the genome for drug candidates.
Why this matters now
DrugCLIP represents the fulfillment of the Human Genome Project's therapeutic promise. The 10,000 proteins screened correspond to roughly half the protein-coding genes identified 23 years ago—now finally accessible to systematic drug discovery.
2 of 3
1985
Lipitor Discovery via Rational Drug Design (1985)
Warner-Lambert scientists used early computational modeling to design atorvastatin (Lipitor), targeting the HMG-CoA reductase enzyme's active site. The process took over 12 years from target identification to FDA approval in 1996. Lipitor became the best-selling drug in pharmaceutical history, generating over $125 billion in sales.
Then
Lipitor validated structure-based drug design as a viable approach, though computational limitations meant most work still required extensive laboratory screening.
Now
The success established the paradigm that understanding protein structure enables rational drug design—but computational power and structural data remained limiting factors for decades.
Why this matters now
Where Lipitor's discovery required years of computational work on a single target, DrugCLIP can now screen the entire human proteome in under a day. The fundamental approach—matching molecular shapes to protein binding sites—remains the same, but the scale has expanded by orders of magnitude.
3 of 3
1990-1999
High-Throughput Screening Revolution (1990s)
Pharmaceutical companies invested billions in robotic systems capable of physically testing up to one million compounds per day against disease targets. The approach promised to accelerate drug discovery by brute-force screening of chemical libraries. Major hits included gefitinib and maraviroc, now standard treatments for cancer and HIV respectively.
Then
High-throughput screening became industry standard, but hit rates remained low (typically 0.01-0.1%) and the approach required synthesizing or purchasing physical compounds.
Now
The method plateaued as companies exhausted their existing compound libraries. Virtual screening emerged as a faster, cheaper alternative, but remained computationally limited until recent AI advances.
Why this matters now
DrugCLIP represents virtual screening's maturation into a practical replacement for physical high-throughput screening. Where robots tested one million compounds daily, AI now evaluates 500 million computationally—against 10,000 targets simultaneously rather than one.
