Why Earendil Labs matters
Most AI drug-discovery startups are still selling a story about models. Earendil Labs is one of the few that has converted that story into large, named biologics checks from a top-ten pharma. In twelve months Sanofi paid $125 million upfront for two bispecific antibodies, then came back with a platform collaboration worth up to $160 million in near-term cash and $2.56 billion in total potential value, and then participated in a $787 million private raise. That sequence is rare even among well-funded computational biology companies: the buyer is not licensing a black-box screening tool, it is taking worldwide rights to specific molecules and paying again to run the same engine on more autoimmune targets.
The reason to track Earendil is not that it has an approved drug. It does not. It is that the company is now a live test of whether an AI-native antibody engine, built in Beijing and listed through a Delaware vehicle, can produce clinical-stage biologics fast enough to justify those checks. HXN-1001 is already in Phase IIa ulcerative colitis. HXN-1011 has entered Phase 1. A Hong Kong listing is being shopped at a reported $5 billion valuation. The next 12-18 months decide whether this is a real clinical franchise or a financing story that peaked before the first efficacy readout.
From a Tsinghua-incubated protein model to a US listing vehicle
Jian Peng, a former University of Illinois Urbana-Champaign computer-science professor, ISCB Overton Prize winner and RECOMB 2021 program chair, founded Helixon Therapeutics in Beijing in June 2021 with backing from Tsinghua's Institute for AI Industry Research. The scientific calling card arrived a year later: OmegaFold, a protein language model plus geometry transformer that predicts high-resolution structure from a single amino-acid sequence, without the multiple-sequence alignments AlphaFold2 depends on. Helixon released the method as a bioRxiv preprint on July 21, 2022 and open-sourced the code, arguing that antibodies and orphan proteins, which often lack clean evolutionary alignments, were exactly where single-sequence prediction mattered.
The company then had to turn a structure model into drugs. In February 2023 Zhenping Zhu, previously global head of protein science and design at Novartis and president of R&D at 3SBio, joined as co-founder, president and co-CEO to run the wet-lab and pipeline side. Earendil Labs was incorporated in Delaware in late 2024 as the overseas face of that group. Press releases now come from Wilmington; contact emails still go to helixon.com; BioPharma Dive describes the firm as Delaware-incorporated with offices in Beijing. That dual structure is not a footnote. It is the reason a Hong Kong IPO is the listing path being discussed, and the reason Chinese regulators later flagged the red-chip holding company.
The platform: design at sequence scale, then test in the lab
Earendil's public technology page describes a closed loop rather than a single model. A foundational protein network is asked to predict antibody-antigen structure, epitopes and docking poses, to remodel CDRs, and to score biophysical properties, immunogenicity, half-life, expression and stability before a sequence is made. Those designs feed a high-throughput biology stack: naive and immune libraries, hybridoma and single B-cell screening, AI-designed libraries, then binding, blocking, epitope mapping and formulation. The company line is that AI is not a research accessory. Peng's March 2026 fundraising quote put it as a production engine for real therapeutic programs.
That claim is only as strong as the molecules it has produced. Earendil says the engine has generated more than 40 programs. Its pipeline page lists 19 of them across immunology and oncology, spanning half-life extended monoclonals, biparatopic antibodies, bispecifics, a CD19 x BCMA x CD3 trispecific, dual-payload ADCs and T-cell engagers. The page itself is stale: it still stops at Phase 1 and does not show HXN-1001's Phase IIa start. Treat the table as a catalog of named assets, not as current clinical status.
Two drugs in the clinic, and Sanofi on the rest of the IBD stack
HXN-1001 is the asset that turned Earendil into a clinical-stage company. It is a half-life extended anti-TL1A monoclonal aimed at ulcerative colitis and Crohn's disease, formulated at high concentration for subcutaneous injection. Phase 1 healthy-volunteer dosing began in Australia in July 2025. By April 2, 2026 that study had completed enrollment, and the first ulcerative colitis patient had been dosed in a randomized, placebo-controlled Phase IIa trial. Zhu said single doses up to 1,200 mg and multiple doses of 450 mg were well tolerated, with an about 8-week half-life at projected therapeutic doses and a low anti-drug-antibody rate. Those are safety and PK claims from the company, not yet an efficacy result. TL1A is a crowded IBD target: Merck's tulisokibart is already in Phase 3, and several other anti-TL1A antibodies are in the clinic, so HXN-1001 has to win on potency, convenience or both.
HXN-1011 is the second internal clinical program, a biparatopic antibody that binds two epitopes on TSLP and is meant to block both chains of the TSLP receptor. First-cohort Phase 1 dosing was completed on May 18, 2026. The company positions it for asthma, atopic dermatitis, eosinophilic esophagitis, chronic rhinosinusitis with nasal polyps and COPD. The rest of the listed immunology stack is still preclinical or partnered: Sanofi holds exclusive worldwide rights to HXN-1002 (TL1A/α4β7) and HXN-1003 (TL1A/IL-23p19) from the April 2025 license, and the January 2026 collaboration gives Sanofi development and worldwide commercialization of additional bispecifics discovered on Earendil's platform. Oncology programs targeting CDH17, CEACAM5, GUCY2C, DLL3, B7-H3, TROP2, EGFR and PD-1/IL-15 remain undisclosed as to IND timing beyond a general 2026-2027 filing plan.
The money: two Sanofi pacts, a $787 million raise, and a stalled listing
Helixon's early capital was Chinese. Dealroom records an about $10 million 2021 round from Hillhouse, Xianghe Capital and Qingzhi Capital, followed by a June 2022 Series A of nearly CNY 500 million (about $75 million) led by 5Y Capital with Gaorong, Neumann, Xianghe, Hillhouse and Qingzhi. The March 20, 2026 raise of $787 million, with Sanofi and the Hillhouse/Pfizer Biotech Development Fund in the syndicate alongside Dimension, DST Global, INCE Capital, Luminous Ventures and Miracle Capital, is the round that put Earendil on the global map. Disclosed equity across those rounds is about $870 million. On top of that, the two Sanofi collaborations carry up to about $4.4 billion in headline biobucks, of which only the $125 million first-deal upfront, a $50 million near-term milestone, and up to $160 million of second-deal near-term cash are described as close-in; the rest is unearned.
A listing was supposed to be the next step. Bloomberg reported on March 13, 2026 that Earendil was working with CICC and Morgan Stanley on a Hong Kong IPO that might raise as much as $500 million. The Wall Street Journal reported on August 5, 2026 that the company was planning to raise close to $400 million at about a $5 billion valuation and had submitted its application to Chinese regulators, who had raised concerns over the red-chip holding structure. No prospectus has been posted, and no listing date has been announced. Manufacturing is being outsourced: WuXi XDC granted an exclusive global license to its WuXiTecan-2 payload-linker for multiple undisclosed targets in a deal worth up to about $885 million plus royalties (February 27, 2026), and WuXi Biologics signed on March 12, 2026 to provide end-to-end CMC and GMP manufacturing for multiple bispecific, multispecific and ADC candidates.
What to watch next
The clinical gate is HXN-1001. Phase IIa ulcerative colitis is a randomized, placebo-controlled study of safety, pharmacokinetics and preliminary efficacy. Until that readout exists, the 8-week half-life and the company's preclinical potency claims versus unnamed TL1A benchmarks are not a commercial argument. Watch whether a Crohn's study is added, whether the company ever posts a trial ID on ClinicalTrials.gov, and whether Sanofi starts human studies of HXN-1002 and HXN-1003. HXN-1011's Phase 1 is the second internal clock; additional INDs the company has promised for 2026 and 2027, including TSLP/IL-33 and OX40L programs Zhu has discussed publicly, are how the platform either looks industrial or looks like two lucky antibodies.
The corporate gate is the listing. A Hong Kong IPO at a reported $5 billion would be one of the larger AI-biotech debuts in the city after Insilico Medicine, but the August 2026 CSRC red-chip objection is a real block, not a rumor to ignore. Watch whether Earendil refiles, restructures, or stays private on the $787 million. Watch also whether the WuXi manufacturing alliances become a BIOSECURE-Act problem for any future US development. None of those questions is answered by the fundraising headline. They are answered by a Phase IIa table, a posted prospectus, or a Sanofi trial start.
Sources
- Earendil Labs - $787 million financing (March 20, 2026)
- bioRxiv - OmegaFold preprint (Wu, Peng et al., July 2022)
- Earendil Labs - technology page
- Earendil Labs - HXN-1001 Phase IIa first patient (April 2, 2026)
- WSJ - Hong Kong IPO red-chip regulatory hurdle (August 5, 2026)
- Earendil Labs - HXN-1011 Phase 1 first cohort (May 18, 2026)
- BioPharma Dive - $787 million raise and Sanofi partnerships