# Chai Discovery

> Source: https://aiwiki.ai/wiki/chai_discovery
> Updated: 2026-06-28
> Categories: AI Companies, Drug Discovery
> From AI Wiki (https://aiwiki.ai), a free encyclopedia of artificial intelligence. Quote with attribution.

**Chai Discovery** is an American AI biotechnology company that builds [foundation models](/wiki/foundation_model) for predicting and designing the three-dimensional structures of biomolecules to accelerate drug discovery. Founded in early 2024 and headquartered in San Francisco, it is best known for two systems: [Chai-1](/wiki/chai_1), an open biomolecular structure-prediction model released in September 2024 and positioned as an openly distributed counterpart to [Google DeepMind](/wiki/google_deepmind)'s [AlphaFold 3](/wiki/alphafold_3), and [Chai-2](/wiki/chai_2), a 2025 system that designs novel antibodies and proteins from scratch.[1][2] Backed by [OpenAI](/wiki/openai) and [Thrive Capital](/wiki/thrive_capital) among others, Chai Discovery raised a $130 million Series B in December 2025 at a $1.3 billion valuation, making it one of the higher-valued companies in the post-AlphaFold-3 wave of AI-native drug-discovery firms.[1][3]

## What is Chai Discovery?

Chai Discovery describes itself as an applied AI research lab working to "reprogram the interactions between biochemical molecules," with the stated long-term ambition of building a "computer-aided design suite" for molecules analogous to the CAD tools used in mechanical and electronic engineering.[1][3] In practice this means training neural networks that take molecular sequences and chemical specifications as input and output atomic-resolution predictions of how proteins fold and how they bind to other proteins, small-molecule drugs, antibodies, and nucleic acids. The company's structure-prediction work sits in the lineage opened by AlphaFold and generalized by AlphaFold 3, while its newer design work moves from predicting existing structures to generating entirely new molecules computationally, in particular custom antibodies built from scratch ("de novo") against a chosen disease target.[2][4]

The company pursues a hybrid commercial model. Its first model, Chai-1, was released openly for research use, building a large academic and industry user base, while its design system, Chai-2, is offered to pharmaceutical partners under paid access arrangements alongside custom model-development partnerships.[4][5] Chief executive Joshua Meier has framed the underlying bet in language terms: "The question was, if these AI models can understand natural language processing, why can't they understand the real natural language, like DNA and proteins?"[4] The following table summarizes the company at a glance.

| Attribute | Detail |
|---|---|
| Founded | Early 2024 (incorporated March 2024)[2] |
| Headquarters | San Francisco, California[2] |
| Founders | Joshua Meier, Jack Dent, Matthew McPartlon, Jacques Boitreaud[2][6] |
| CEO | Joshua Meier[1][6] |
| Sector | AI for drug discovery; molecular structure prediction and design |
| Key products | [Chai-1](/wiki/chai_1) (2024), [Chai-2](/wiki/chai_2) (2025) |
| Total funding | More than $225 million (through December 2025)[1][3] |
| Latest valuation | $1.3 billion (Series B, December 2025)[1][3] |
| Notable backers | OpenAI, Thrive Capital, Menlo Ventures, General Catalyst, Oak HC/FT[1][3] |

## Who founded Chai Discovery and when?

Chai Discovery was founded in early 2024 and incorporated in March 2024 by Joshua Meier, Jack Dent, Matthew McPartlon, and Jacques Boitreaud.[2][6] The company emerged from stealth on 9 September 2024, the same day it published Chai-1.[7]

Chief executive **Joshua Meier** is an AI researcher with a background spanning machine learning and biology. He worked as a researcher at [OpenAI](/wiki/openai) during the GPT-1 and GPT-2 era, then joined [Meta](/wiki/meta_ai)'s generative-biology group, where he helped develop ESM1, an early transformer protein-language model and a precursor to the [ESM](/wiki/esmfold) family whose embeddings later powered Chai-1's alignment-free prediction mode.[2] Before founding Chai, Meier served as Chief AI Officer at the biotechnology company Absci, where he led the buildout of an AI-driven de novo antibody-design program.[2][6] President **Jack Dent** is a Harvard computer-science classmate of Meier who previously worked at [Stripe](/wiki/stripe), where he contributed to Stripe Link and Stripe Capital.[2] **Matthew McPartlon** worked with Meier on de novo antibody-design modeling at Absci, and **Jacques Boitreaud** studied RNA bioinformatics at McGill University before working as an AI scientist at the small-molecule company Aqemia.[2]

According to a January 2026 TechCrunch account of the company's origins, the idea traced back to conversations between Meier, Dent, and OpenAI chief executive [Sam Altman](/wiki/sam_altman) as early as 2018, when Altman had floated the notion of a proteomics company; Meier judged the technology premature at the time. After Meier spent the intervening years at Meta and Absci and Dent built product at Stripe, the four founders built Chai Discovery in 2024 while operating out of OpenAI's San Francisco offices, with OpenAI's investment arm joining as a seed investor.[5] Dent told TechCrunch that "every line of code in our codebase is homegrown," reflecting a decision to build custom architectures for molecular data rather than fine-tune off-the-shelf language models.[5]

## What are Chai Discovery's models?

### What is Chai-1?

[Chai-1](/wiki/chai_1) is a multimodal foundation model for biomolecular structure prediction, released on 9 September 2024.[7] It predicts atomic-resolution 3D structures for proteins, small-molecule ligands, DNA, RNA, multimeric complexes, covalent modifications, and antibody-antigen interfaces, reaching accuracy comparable to or exceeding AlphaFold 3 on several public benchmarks. On the PoseBusters protein-ligand evaluation set it reaches a 77% success rate, comparable to AlphaFold 3's reported 76%, rising to about 81% when the model is prompted with the apo (unbound) protein structure.[7][2] A distinctive feature is an alignment-free ("MSA-free") inference mode in which residue-level embeddings from a large protein language model substitute for multiple sequence alignments, allowing strong performance from single sequences alone.[2]

The model was initially released under a research-only license, with weights and inference code published on GitHub; in late November 2024 the company relicensed Chai-1 under the permissive Apache 2.0 license, permitting broad commercial use and making it one of the most freely usable systems in its class.[2] Chai-1 was the first openly distributed alternative to AlphaFold 3 to ship in functional form, predating the academic [Boltz](/wiki/boltz)-1 model by roughly two months.[2] By 2026 it had accumulated several hundred academic citations.[2]

### What is Chai-2?

[Chai-2](/wiki/chai_2), unveiled on 30 June 2025, extends the company's work from structure prediction into molecular design.[8][2] It is a system for zero-shot de novo design of antibodies and nanobodies, meaning it generates entirely novel binder sequences against a specified protein target and epitope without requiring prior binders or task-specific fine-tuning.[8] In a technical report accompanying the launch (also posted to bioRxiv as "Zero-shot antibody design in a 24-well plate"), the company reported that Chai-2 achieved roughly a 16% experimental ("wet-lab") hit rate when designing 20 or fewer antibodies or nanobodies per target across 52 diverse targets, which it characterized as a more than hundredfold improvement over previous computational methods whose success rates were on the order of 0.1%.[8][2] It said that in a single round of testing the system found at least one binder for about half of the targets (26 of 52), and that more than 86% of resulting full-length antibodies showed strong developability profiles.[8] In a related miniprotein-binder design task, the company reported a 68% wet-lab hit rate, often with picomolar binding affinities.[8] Chai-2 is offered to pharmaceutical partners under paid access arrangements rather than as an open release.[4][5]

## How is Chai Discovery funded?

Chai Discovery has raised more than $225 million across three disclosed rounds.[1][3] The company's seed round was reported by Bloomberg in September 2024 at roughly $30 million, led by Thrive Capital with participation from the OpenAI Startup Fund and Dimension, at a valuation of approximately $150 million.[7] In August 2025 the company announced a $70 million Series A led by Menlo Ventures' Anthology Fund, valuing it at $550 million; as part of that round, former Pfizer Chief Scientific Officer Mikael Dolsten joined the board.[2][9]

The Series B, announced on 15 December 2025, was a $130 million round at a $1.3 billion valuation.[1][3] Press coverage and the company's own announcement state the round was co-led by [General Catalyst](/wiki/general_catalyst) and Oak HC/FT, with participation from [Menlo Ventures](/wiki/menlo_ventures), OpenAI, Thrive Capital, Dimension, Neo, the Yosemite venture fund, Lachy Groom, SV Angel, and new investors Glade Brook and Emerson Collective.[1][3] As part of the round, Oak HC/FT co-founder Annie Lamont and General Catalyst chief executive Hemant Taneja joined the company's board.[1][3] The funding round raised the company's total capital to more than $225 million, and Chai said it would use the proceeds to accelerate research and product development and grow its commercialization efforts toward creating the equivalent of a computer-aided design suite for molecules.[1][4] The table below summarizes the disclosed financing history.

| Round | Date | Amount | Valuation | Lead(s) |
|---|---|---|---|---|
| Seed | September 2024 | ~$30 million | ~$150 million | Thrive Capital[7] |
| Series A | August 2025 | $70 million | $550 million | Menlo Ventures (Anthology Fund)[2][9] |
| Series B | December 2025 | $130 million | $1.3 billion | General Catalyst and Oak HC/FT (co-led)[1][3] |

The framing of the Series B as led by General Catalyst and Oak HC/FT is consistent with the company's announcement and TechCrunch's reporting; Menlo Ventures, which led the earlier Series A, participated in the Series B rather than leading it.[1][3]

## Why does Chai Discovery matter, and who are its competitors?

Chai Discovery is one of the most prominent companies in the wave of AI-native, foundation-model-driven drug-discovery firms that followed the May 2024 announcement of AlphaFold 3. Two features distinguish its trajectory. First, its decision to release Chai-1 openly, and later under a permissive license, contrasted with the initial release of AlphaFold 3 as a restricted web server, and helped Chai-1 become a widely used research tool.[2] Second, with Chai-2 the company moved beyond predicting existing structures to designing new molecules, positioning itself in the emerging market for computational antibody and protein design.[8][4] CEO Joshua Meier has summarized the company's framing this way: "We want to make biology look less like science and look more like engineering and then usher in as much progress as we can on drug discovery."[4]

The competitive landscape is crowded. [Isomorphic Labs](/wiki/isomorphic_labs), the [DeepMind](/wiki/deepmind) spinout led by [Demis Hassabis](/wiki/demis_hassabis), pursues a similar mission with substantially larger funding and pharmaceutical partnerships. [EvolutionaryScale](/wiki/evolutionaryscale), founded by former Meta protein-AI researchers and the developer of the [ESM3](/wiki/esm3) model, works on generative protein models. In the antibody- and protein-design segment, Chai competes with firms such as [Nabla Bio](/wiki/nabla_bio) and [Xaira Therapeutics](/wiki/xaira_therapeutics), the latter co-founded by Nobel laureate David Baker, as well as with open structure-prediction and design models such as [Boltz](/wiki/boltz) and [Boltz-2](/wiki/boltz_2). Both the prediction and design problems Chai works on sit at the center of contemporary [AI drug discovery](/wiki/ai_in_drug_discovery), and the breadth of well-funded competitors reflects the perceived size of the opportunity to compress the timelines and costs of finding new medicines.[1][2]

## References

1. [OpenAI-backed biotech firm Chai Discovery raises $130M Series B at $1.3B valuation](https://techcrunch.com/2025/12/15/openai-backed-biotech-firm-chai-discovery-raises-130m-series-b-at-1-3b-valuation/), TechCrunch, 15 December 2025.
2. [Chai Discovery Business Breakdown & Founding Story](https://research.contrary.com/company/chai-discovery), Contrary Research.
3. [Chai Discovery Announces $130 Million Series B To Transform Molecular Discovery](https://www.businesswire.com/news/home/20251214931432/en/Chai-Discovery-Announces-$130-Million-Series-B-To-Transform-Molecular-Discovery), Business Wire, 14 December 2025.
4. [Chai's the Limit for AI Antibody Designer After $130M Series B Funding](https://www.genengnews.com/topics/artificial-intelligence/chais-the-limit-for-ai-antibody-designer-after-130m-series-b-funding/), Genetic Engineering & Biotechnology News, December 2025.
5. [OpenAI-backed biotech firm Chai Discovery raises $130M Series B at $1.3B valuation](https://techcrunch.com/2025/12/15/openai-backed-biotech-firm-chai-discovery-raises-130m-series-b-at-1-3b-valuation/), TechCrunch (origin-story details), 15 December 2025.
6. [Chai infuses AI drug discovery efforts with $130M series B](https://www.fiercebiotech.com/biotech/chai-infuses-ai-drug-discovery-efforts-130m-series-b), Fierce Biotech, December 2025.
7. [Chai Discovery Gets OpenAI, Thrive Capital Backing to Bring AI to Biotech](https://www.bloomberg.com/news/articles/2024-09-09/openai-thrive-capital-back-six-month-old-ai-drug-discovery-startup), Bloomberg, 9 September 2024.
8. [Chai-2: Zero-shot antibody design in a 24-well plate](https://www.biorxiv.org/content/10.1101/2025.07.05.663018v1), Chai Discovery Team, bioRxiv, 5 July 2025; see also [MarkTechPost coverage](https://www.marktechpost.com/2025/07/05/chai-discovery-team-releases-chai-2-ai-model-achieves-16-hit-rate-in-de-novo-antibody-design/), 5 July 2025.
9. [Chai Discovery Announces $70 million Series A To Transform Molecular Design](https://www.businesswire.com/news/home/20250806670137/en/Chai-Discovery-Announces-$70-million-Series-A-To-Transform-Molecular-Design), Business Wire, 6 August 2025; valuation and board detail per [Contrary Research](https://research.contrary.com/company/chai-discovery).