MoleculeMind (分子之心)
Beijing AI protein-design platform founded by structure-prediction pioneer Xu Jinbo, selling its MoleculeOS platform to pharma and biomanufacturing partners.
1. Core Product / Service
MoleculeMind was founded in January 2022 in Beijing and incubated out of the Tsinghua ecosystem [1]. Its product line is a stack of protein models: MMFold (in-house structure prediction, claimed to beat open-source models on FoldBench and alphafold3 on high-precision cases), MMDesign (de novo biologics design, claimed >90% hit rate across 12 clinical targets) and the NewOrigin multimodal protein foundation model [2]. These are packaged into MoleculeOS (MOS), an "AI-native bioengineering platform" that links design, small-batch wet-lab validation and model retraining. MOS was used internally first and was formally opened to industry on 2026-07-02 [3]. Modalities covered: nanobodies, antibodies, cyclic peptides, mini-proteins and industrial enzymes.
2. Target Users & Pain Points
- Large pharma biologics teams that want to replace large-library screening with "programmable design": the company says 14–50 tested candidates are enough to obtain high-affinity, high-expression nanobodies [2].
- Biomanufacturing / industrial-enzyme companies: directed-evolution projects, with a reported several-fold strain-yield gain at one partner [2].
- Hard-to-express programs: one partner pipeline reportedly gained >400× expression with >90% monomer [2].
Business model. Three channels are disclosed [2]: (1) platform licensing, with MOS embedded in a pharma partner's R&D workflow; (2) joint R&D on difficult targets, typically structured as service fees plus downstream sharing (milestones/revenue share, terms undisclosed) [4]; (3) a small set of early in-house assets described as first-in-class. No public FTE pricing, deal values or named pharma licensees exist; revenue is not disclosed. Payers are therefore mainly pharma/biotech R&D budgets and industrial partners such as shareholder Cathay Biotech (凯赛生物).
Wet lab. MOS is built around a "small-scale wet-lab validation" loop, and the company reports wet-lab results on more than a dozen targets (TNFα, PD-L1, MDM2 and others) [1][2]. Whether this runs in an owned lab or through CRO/partner labs is not disclosed; the small candidate counts (tens, not thousands) mean the lab footprint needed is modest.
3. Competitive Landscape
| Player | Positioning | Vs. MoleculeMind |
|---|---|---|
| biogeometry | GeoFlow antibody/enzyme design, licensing + services | Closest domestic peer; similar size of 2026 round |
| biomap | xTrimo foundation model, Sanofi module deal | Larger, broader (cells, genomes), heading to IPO |
| tianwu-tech | Pro-series protein models, enzyme-heavy service projects | Overlaps on industrial enzymes |
| helixon-therapeutics | Tsinghua-linked AI protein design that moved to own pipeline | Chose asset model; MoleculeMind stays platform-first |
| generate-biomedicines / chai-discovery / nabla-bio | US AI biologics design | Far better funded; Chai/Nabla also sell design to pharma |
4. Unique Observations
- Platform-first in a market that pays for assets. Chinese AI-drug capital in 2025–26 flowed mostly to companies with licensable assets; MoleculeMind's in-house pipeline remains undisclosed, so the 2026 round is effectively a bet that platform licensing to pharma can scale in China [1].
- Industrial capital is a deliberate second leg. Cathay Biotech led a 2023 round and re-upped in 2026; COFCO's fund and Sepax (赛分科技, protein purification) joined in 2026 [2]. Enzymes and biomaterials give revenue paths with shorter validation cycles than drugs.
- Performance claims are self-reported. Hit rates, affinities and benchmark wins come from company and investor-sourced articles; no independent head-to-head is published.
- Founder credibility is the moat narrative. Xu Jinbo's RaptorX-Contact was an early deep-learning contact predictor that preceded AlphaFold2 [2].
5. Financials / Funding
- Apr 2022, angel: tens of millions of USD; Baidu Ventures and others participated [5].
- 2023, strategic: >¥100M, led by Cathay Biotech, with Lenovo Capital (联想创投) [3].
- Sep 2024, Series A: "several hundred million RMB", co-led by 谢诺投资 and Shenzhen Capital Group (深创投) [4].
- Jun 2026, Series A extension: reported as "over US$100M" (some outlets cumulative, some single-round), with 蓝桥资本, 浦东创投, COFCO emerging-industry fund, 东方富海, Fosun, Sepax and others; Cathay Biotech and 芯航资本 followed on [2].
- Valuation and revenue: not disclosed.
6. People & Relationships
- Founder: Xu Jinbo (许锦波), former professor at Toyota Technological Institute at Chicago; developer of RaptorX.
- Investors: Cathay Biotech, Shenzhen Capital Group, 谢诺投资, SenseTime-linked 国香资本, Baidu Ventures, Fosun, COFCO fund, Sepax.
- Peers: biogeometry, biomap, tianwu-tech, helixon-therapeutics, xtalpi.
Sources: [1] stdaily 2026-08-03; [2] pedaily 2026-06; [3] qbitai 2026-07; [4] bjnews 2024-09; [5] pedaily 2022-04.