Somite AI Raises $47M Series A

By Yuki Tahara, PhD, Ajinomoto Group Ventures

Somite AI has recently raised $47M from VC investors to build foundational models for human stem cells. The goal is to produce any cell types for any person and thus unlock new therapies derived from iPS cells. Why does it matter?    

iPS cells (induced pluripotent stem cells) are reprogrammed from skin or blood cells to revert into an embryonic-like pluripotent state. Their power lies in the fact that they can differentiate into any human cell type and can thus be harnessed for developing therapeutics for diseases with no other available treatment.  

iPS cell therapies will help unlock an emerging regenerative medicine segment projected to reach US $83B by 2030. However, as of mid-2025 no allogeneic iPS cell-derived therapy is approved in US. This highlights the regulatory hurdles for developing therapies in this new therapeutic modality.

The Cell Differentiation Bottlenecks

Of the thousands of known human cell types, scientists have successfully differentiated iPS cells into only a small number of well-characterized cell types, with best estimates coming in at less than 1% of known human cell types.

The first obstacle is the research and development phase. Identifying robust cell types that are meaningful for new therapies is not easy. 

Even after candidates are identified, developing them into therapeutics is hard. It involves determining the optimal signaling pathways to differentiate the cells, to ensure they are expressing the right genes, performing their biological function, and can integrate properly into tissues. Furthermore, to use such differentiated cells as pharmaceuticals, it is essential to establish a stable manufacturing process that consistently maintains the same purity and quality. This is difficult as cells are living systems and can behave very differently with just small variation in growth factors and nutrients in the culture conditions. Process factors such as CDMO transfer introduce further variables.

This complexity results in off-target differentiation, low yields, and high costs of candidate therapies. Today, optimizing the differentiation process of iPS cells into target cells is a manual, trial-and-error process that is time-consuming and expensive. Here is where data and machine learning can make a difference.  

Introducing Somite AI: Accelerating Cell Therapy with Foundational Models

Somite is a TechBio company building foundation models that use digital twin simulations to identify optimal manufacturing methods for iPS cell differentiation.

Somite’s key unlock is the ability to generate cell signaling data at 1,000x lower cost, which enables training large-scale models with unmatched predictive power, and using these models to discover, refine, and optimize high-impact applications in cell therapy.  

Starting with a high throughput screening process tailored to iPS cells, Somite can map single-cell gene expression across all stages of cell differentiation and trace and analyze the differentiation process at an unprecedented scale.

These proprietary data feed into its foundation model, DeltaStem, designed to accelerate the process of optimization for unparalleled purity, scalability, and consistency.

In parallel to its platform development, Somite is building its own therapeutic programs including beta cells for type 1 diabetes, articular cartilage for orthopedic applications, and satellite cells for muscle pathologies.  

The Investor Point of View

Somite’s platform technology targets the key roadblocks that have hindered the development of cell therapy. As this technology matures, it will become a game-changer in regenerative medicine and cell therapy, rendering today’s trial-and-error methods obsolete.  

This is surely no small feat, but we believe Somite’s team has a real shot. The CEO, Dr. Micha Breakstone, is a seasoned AI entrepreneur and led Chorus.ai through its $575 million acquisition. His cofounders are world-class professors including Dr. Jonathan Rosenfeld, Head of the Fundamental AI Group at MIT; National Academy of Science member Dr. Olivier Pourquié of Brigham and Women's Hospital and Harvard Medical School; James Prize winner Dr. Allon Klein of Harvard Medical School; NAS member Dr. Jay Shendure of University of Washington School of Medicine; and NAS member Dr. Cliff Tabin who is Chair of Genetics at Harvard Medical School.

AI alone may not solve humanity's challenges. But when paired with this exceptional level of multidisciplinary expertise, it becomes a powerful force for innovation and tackling disease challenges at scale.  

At the Ajinomoto Group, we have been supporting regenerative medicine since launching our cell culture medium product developed jointly with Nobel laureate Dr Shinya Yamanaka’s Center for iPS Cell Research and Application (CiRA) at Kyoto University. Our scientists have experienced first-hand the dire need for a process revolution in cell differentiation. We look forward to seeing how the cell therapy community will leverage this new tool to drive progress in regenerative medicine and treat diseases we can’t address today. 

‍