> ## Content Index
> Fetch the complete content index at: https://bioweek.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# BigHat Closes $75 Million Series C as Its AI-Designed ADC Enters the Clinic
- URL: https://bioweek.com/bighat-closes-75-million-series-c-as-its-ai-designed-adc-enters-the-clinic/
- Published: 2026-09-25T05:03:51.000Z
- Updated: 2026-09-25T05:03:51.000Z
- Description: BigHat Biosciences raised $75 million as BHB810, a CDH17 VHH-Fc ADC, entered a Phase 1/2 gastric cancer trial. The round tests whether wet-lab data can fix AI drug discovery's translation problem.
- Author: BioWeek
- Tags: Markets, Pipeline, Science

BigHat Biosciences has raised $75 million in Series C [financing](https://www.biospace.com/press-releases/bighat-biosciences-announces-75-million-series-c-financing-to-advance-ai-designed-therapeutics-and-leading-agentic-data-platform-for-rapid-protein-design?ref=bioweek.com) as its first wholly owned antibody-drug conjugate reached patients. The San Mateo, California company announced the round on September 24, 2026, bringing total equity funding since its 2019 founding to $223 million. In the same month, BigHat dosed the first patient in a Phase 1/2 study of BHB810 for advanced gastric and gastroesophageal junction adenocarcinoma. Partnerships and platform metrics no longer set the story. Human data will.

### The CDH17 bet, and why size is the argument

BHB810 is a VHH-Fc antibody-drug conjugate directed at cadherin-17, a cell-surface adhesion molecule expressed across 50% to 60% of gastric cancers and also present on colorectal and pancreatic tumors. The antigen has no overlap with HER2, PD-L1, or Claudin 18.2 (Vyloy). That orthogonality is the commercial case in gastric cancer, a market already carved up by those biomarkers.

The construct is a compact single-domain antibody fusion that BigHat describes as about 50% smaller than a conventional full-length IgG, a design choice aimed at solid tumor penetration rather than another increment in binding affinity. The payload is monomethyl auristatin E, attached through a site-specific, enzyme-cleavable linker using GlycoConnect and toxSYN technology licensed from Synaffix (a Lonza company) at a drug-to-antibody ratio of 4\. Preclinical work in AACR 2026 abstract 6922 showed complete or near-complete tumor clearance across nearly 30 patient-derived and cell-derived models, with no detectable payload deconjugation or retro-Michael payload loss in human serum.

Those results are still xenografts and serum tubes. The first-in-human [trial](https://clinicaltrials.gov/study/NCT07529808?ref=bioweek.com) (NCT07529808) is where the smaller scaffold, the DAR 4 MMAE load, and CDH17 selectivity have to survive in patients with advanced gastric and GEJ disease.

### Binding was never the bottleneck

First-generation computational biotechs spent years optimizing in silico affinity, then ran into efficacy, tolerability, and pharmacokinetic developability in humans. Recursion, Exscientia, and BenevolentAI became the cautionary set: clinical failures and pipeline reprioritizations that left investors treating AI-designed as a slogan until a molecule cleared CMC and Phase 1\. Computational binding affinity did not reliably become in vivo activity.

BigHat's answer is Milliner, a platform that pairs synthetic biology cell-free protein synthesis and a high-speed automated wet lab with active learning and Bayesian optimization. The loop designs, synthesizes, and tests thousands of variants weekly, generating proprietary experimental data rather than mining public structural sets. The company was founded in 2019 by Peyton Greenside, now CEO, and Mark DePristo, the former CEO who transferred leadership to Greenside in March 2025\. Greenside holds a PhD from Stanford. The Series C is capital against the claim that fit-for-purpose wet-lab data, collected at industrial throughput, can front-load manufacturability and biophysical developability instead of discovering them in the clinic.

### Pharma paid for discovery. The clinic will price ownership.

BigHat already sold the platform. Eli Lilly has collaborated on antibody developability, including Lilly TuneLab foundation models and Catalyze360 support. AbbVie paid $30 million upfront in December 2023 for oncology and neuroscience discovery, with up to $325 million in milestones. Merck has three project collaborations. Johnson & Johnson and Amgen are additional partners, and Amgen Ventures, Eli Lilly and Company, and Merck Global Health Innovation Fund all returned as investors.

DFJ Growth and Premji Invest co-led the round. New investors include Catalio Capital Management, LG Technology Ventures, and Sigmas Group. Returning backers also include 8VC, Alexandria Venture Investments, Andreessen Horowitz, Discovery Ventures, GRIDS Capital, Intermountain Ventures, Quadrille Capital, and Section 32, which led the $75 million Series B in 2022\. Prior rounds were a $5 million seed, a $19 million Series A, and a $44 million Series B extension in 2025.

A second wholly owned program, BHB299, is an avidity-driven T-cell engager selectively targeting CEACAM6-expressing solid tumors. It is completing preclinical development, with clinical entry planned for 2027\. That molecule will face a different translation problem than an ADC, and it is still a year away from patients.

### What the round actually funds

Seventy-five million dollars does not buy a shortcut around gastric ADC toxicology. BHB810 still has to show that a VHH-Fc conjugate at DAR 4 can deliver MMAE into CDH17-positive tumors without failing on payload release, exposure, or off-tumor binding. Milliner still has to show that thousands of weekly variants produce molecules that scale, not just assay well. BHB299 still has to leave the preclinical stack in 2027.

If the gastric study produces a usable safety profile and any signal of activity in a CDH17-selected population, BigHat will have moved AI protein design out of the partnership slide deck and into a setting where developability is measured in patients. If it does not, the $223 million raised to date will join a familiar ledger: computation found a binder, and biology kept the rest.