The computer-aided design suite for molecules
HQ: Cambridge, MA, United States | Founded: 2024 | Employees: ~50–100 (est.) | Stage: Series C at ~$3.8B (July 2026); $600M+ raised across five rounds | Website: https://www.chaidiscovery.com
Chai Discovery builds AI foundation models for molecular structure and design. The company's models predict how molecules — proteins, DNA, RNA, small molecules, ions — fold and interact, and then use those predictions to design new ones: antibody candidates, optimized proteins, potential medicines. The founders call it a 'computer-aided design suite for molecules,' borrowing the analogy from engineering software.
Joshua Meier founded the company in March 2024 with Jack Dent, shortly after Meier's work on Meta's ESM protein language models helped define the field. Chai-1, released in September 2024, matched or beat the best structure-prediction systems across multiple molecule types. In August 2025, Chai-2 demonstrated zero-shot antibody design — proposing lab-validated antibody candidates directly from a target's structure, a capability with immediate pharmaceutical appeal.
Funding has come in tsunamis: a ~$30M seed from Dimension, OpenAI and Thrive Capital at founding; $70M Series A co-led by Menlo Ventures and Anthropic's Anthology Fund in August 2025; a $130M Series B at $1.3 billion in December 2025; and a Series C that valued the company at roughly $3.8 billion by July 2026 — more than $600 million raised in about two years.
Chai Discovery is the fastest value-creation story in AI biology. Joshua Meier — who led Meta's protein language-model work (ESM) — founded it in March 2024 with engineer Jack Dent to do for molecules what CAD did for hardware: a design environment where scientists model, predict and build molecular structures instead of guessing at them. Six months in, OpenAI and Thrive Capital wrote one of the largest seed checks in biotech history.
The products explain the pace. Chai-1 (September 2024) predicted molecular structure at frontier accuracy — proteins, DNA, RNA, small molecules and ions together. Chai-2 (August 2025) went further: zero-shot antibody design, generating viable antibody candidates from a target's structure alone, with lab validation. That is the pharmaceutical industry's holy grail — starting from a design rather than an immunization campaign and years of screening.
The money followed in waves: $30M seed (Sept 2024, Dimension-led with OpenAI and Thrive), $70M Series A (Aug 2025, Menlo Ventures and Anthropic's Anthology Fund co-leading), $130M Series B at $1.3B (Dec 2025), and a Series C around $3.8B by July 2026 — roughly tripling in seven months with more than $600M raised in total.
Biotech's graveyard is full of great predictors that never produced drugs. Chai's risks are the standard ones magnified by speed: wet-lab validation must keep matching the computational claims, pharma partnerships must convert to pipelines (and milestones), and the competition — Isomorphic Labs (DeepMind), EvolutionaryScale, Recursion and every big pharma's AI group — is extraordinarily well-resourced. But if 'design molecules like software' is even half right, Chai is positioned at the center of it.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki