Shimadzu and Axio BioPharma to validate AI media optimization

Shimadzu Corporation has begun a collaboration with Axio BioPharma Inc., a biotechnology company based in Wisconsin, to optimize cell culture media using AI, the company announced on November 5, 2026. The work pairs Shimadzu's CellTune Ver. 2 cell culture optimization support software and its Media Blending Kit with Axio's antibody expression and evaluation technologies. What the two companies have signed up for is a joint validation project, not a product launch: culture conditions the software proposes will be tested and evaluated at Axio's laboratory, and the results fed back into the design of the next set of conditions.

The software proposes the formulation, the kit prepares it

CellTune Ver. 2 uses AI to propose optimal conditions, including media formulations and culture environments, based on experimental data, according to Shimadzu. The company's Japanese release names the input more specifically: analysis data from culture supernatant obtained by liquid chromatography mass spectrometry (LC-MS). The Media Blending Kit packages a base medium together with several supplements, and Shimadzu says it lets a proposed formulation be prepared quickly.

Shimadzu's reasoning for that pairing rests on the number of options. Media contain many components, among them amino acids and vitamins, so the number of possible combinations is enormous, and media development has leaned heavily on individual researchers' experience. Selection and formulation of the medium, the company says, carries into productivity, quality and cost.

Axio's laboratory runs the experiments

The experimental work sits on the Axio side. Conditions proposed with Shimadzu's technologies will be tested and evaluated in Axio's laboratory, and the results will go back into the design of subsequent conditions, which is the data-driven, iterative loop the two companies set out to validate. Shimadzu describes Axio as a company that supports drug development with an experimental platform for evaluating antibody expression and validating culture conditions.

According to Axio's website, its antibody production lab Forge sits in Madison, Wisconsin, and runs transient expression in CHO and HEK293 cells, taking a program from sequence or plasmid through purified antibody and an analytical package covering purity, aggregation and yield. That is how Axio describes the platform it already operates, and not an outcome of this collaboration.

What separates this from MediTune is the division of work

Shimadzu already offers a media optimization service of its own. MediTune, its AI custom media development service, combines Shimadzu Diagnostics' media manufacturing know-how with Shimadzu's AI-based design and delivers the whole path in one package, from designing a formulation to supplying the custom medium. Shimadzu says MediTune covers uses including regenerative medicine and cell therapy, and that it can supply material from research grade through GMP-compliant grade.

The Axio project is arranged differently. Shimadzu provides the software and the kit, and the testing and evaluation happen in a partner's lab. That is the line between the two offerings: who does which part, rather than which modality is being worked on. Shimadzu says it will assess the practical applicability of the media optimization workflow together with Axio, and that it will keep working with partners in Japan and overseas to contribute to the biopharmaceutical industry, particularly in the North American market.

What to settle before putting an AI proposal into your own program

Start with the data that feeds the proposal. Shimadzu's account has the AI working from LC-MS data on culture supernatant, so the first question for a team considering this route is whether the components it already measures, the timepoints it samples and the analytical method it runs line up with what Shimadzu assumes.

The second question is whether a proposed formulation can be reproduced in house. That means checking whether the base medium and supplements in the blending kit can stand in for an existing medium and feed, and whether the same formulation can still be prepared after scale-up. Screening in transient expression and production in fed-batch differ in both the conditions that get varied and the metrics used to judge them, so decide which stage of your own process this is meant to serve before going further.

For contract work or joint development, settle the rights to the resulting media formulation and the ownership of data generated during the validation. Where AI proposals start to inform CMC decisions, the records behind each proposal, the input data and the settings used, need to be traceable afterward. The delivery form of the software and the location where experimental data is held are worth checking against what Shimadzu stated when it launched CellTune Ver. 2 worldwide on July 2, 2026.