Deepcell and Gilead Sciences entered a partnership to co-develop a custom AI foundation model for Chinese hamster ovary (CHO) cell line development using morphology data. Deepcell will image Gilead’s CHO lines on its REM-I high-dimensional cell morphology analysis and sorting platform, then train the model using self-supervised learning. The companies said the approach aims to assess cell-line morphology structure without manual labeling, with a goal of giving cell development teams earlier signal on attributes tied to manufacturing performance such as titer, stability, aggregation, and metabolic profile. Deepcell also said the initial model will be evaluated on held-out Gilead cell lines and that the partners expect to publish an open-weight version. The collaboration reflects continued investment in applying AI foundation models to biologics manufacturing workflows, where morphology-based readouts can help reduce cycle time in selecting production cell lines.