BigHat Biosciences designs protein medicines for cancer and immune diseases by feeding AI-generated antibody designs into an automated lab, then using the test results to improve the next batch. The company said it closed a $75 million Series C to expand that system and advance two cancer candidates, including one already being tested in people.

DFJ Growth and Premji Invest co-led the round. BigHat said the financing lifted its total capital raised to $223 million and will support the clinical-stage pipeline as well as the system that generates experimental data for its protein-design models.

Inside that workflow, software designs millions of molecules, and a cell-free lab produces and characterizes more than 2,000 a week across at least 20 assays for binding, function and developability, according to BigHat. Scientists use a control layer to send samples through instruments and tie each result to the right molecule; those results then feed the next round of model training.

The platform serves both BigHat’s own drug pipeline and outside collaborations. The company names Amgen, Merck, Johnson & Johnson, AbbVie and Lilly as partners, but the financing announcement doesn’t identify them as customers or disclose commercial terms. BigHat’s disclosed pipeline combines wholly owned and partnered antibody programs, rather than a pure software product handed off at the design stage.

BHB810, the lead program, is an antibody-drug conjugate directed at CDH17-positive gastrointestinal tumors. An antibody carries a drug payload toward cells with that marker. BigHat announced on September 1 that the first patient had been dosed in a Phase 1 escalation study of advanced gastric and gastroesophageal cancers, which will assess safety, tolerability and preliminary tumor activity.

The other program named for financing is BHB299, an avidity-driven T-cell engager designed to target CEACAM6-expressing solid tumors. BigHat says it is nearing completion of preclinical development and plans to begin human trials in 2027. Fast design cycles can narrow candidates; they cannot replace the patient data those studies are meant to produce.