Argenx taps Chai Discovery's AI for de novo antibody engineering in immunology

San Francisco-based Chai Discovery has entered a [collaboration agreement](https://www.biospace.com/press-releases/chai-discovery-enters-into-collaboration-agreement-with-argenx-to-advance-ai-driven-immunology-breakthroughs) with argenx (Euronext & Nasdaq: ARGX) granting the immunology company early access to Chai's generative AI platform for de novo antibody discovery. The Chai Discovery argenx collaboration is the fourth disclosed pharma partnership for the AI startup in roughly seven months, following agreements with Eli Lilly, Pfizer and Novartis, and marks argenx's latest move to diversify its discovery toolkit beyond its core FcRn antibody engineering franchise. Financial terms were not disclosed. Under the agreement, argenx scientists gain use of Chai's platform to design antibody candidates computationally, without a pre-existing template, across unspecified therapeutic targets in immunology. No molecule, target or disease indication has been named, consistent with a discovery-stage tool access arrangement rather than an asset transaction. Chai's flagship system, Chai-3, is a multimodal generative AI model built on a diffusion-based architecture that predicts and generates three-dimensional molecular structures across proteins, antibodies, nucleic acids and small molecules within a single framework. Its predecessor, Chai-2, released in 2025, was described by the company as the first zero-shot antibody design platform to achieve double-digit experimental hit rates — roughly 16-20% in fully de novo antibody design against a target epitope, compared with sub-1% rates for prior computational methods. Chai-3 is reported to further reduce design failure rates and improve binding affinity and multi-specific antibody engineering. The distinguishing claim, as characterized by Chai, is generative rather than purely predictive design: instead of inferring how a known sequence folds, as tools like AlphaFold do, the model proposes new antibody sequences intended to satisfy a specified binding function before any wet-lab synthesis occurs. Chai previously signed a research collaboration with [Eli Lilly](https://app.allsci.com/news/ASC-NR-0000003859191-1.0-1782914439) in January 2026 that included a custom AI model trained on Lilly's proprietary data, then a [license agreement with Pfizer](https://app.allsci.com/news/ASC-NR-0000003834699-1.0-1781764466) in June 2026 granting early access to Chai-3 alongside a bespoke Pfizer-trained model, and a [collaboration with Novartis](https://app.allsci.com/news/ASC-NR-0000003876331-1.0-1783959757) in mid-July for antibody discovery across multiple programs. The argenx deal arrived one day after Chai closed a [USD 400 million Series C](https://app.allsci.com/news/ASC-NR-0000003878037-1.0-1784113894) at a USD 3.8 billion valuation, nearly tripling its worth in under eight months. That timing indicates the argenx agreement is part of a structured commercial rollout rather than an independently negotiated bilateral relationship, with Chai deploying fresh capital toward customer acquisition while using its pharma roster to support its fundraising narrative. *** This article was generated with AI assistance and reviewed and edited by the AllSci editorial team Explore more at AllSci News: [https://allsci.com/news/](https://allsci.com/news/) --- Spot something wrong? [Report an issue with this article](https://newsgen-prod.reframedata.com/feedback/ai-antibody-discovery-argenx-taps-chai-discoverys)