With New Material, MISTRAL wants to master the transition from model to product

An AI model can analyze a document or write code. For a company to use it on a daily basis, it must still build the tool adapted to its employees and connect it to its software. By acquiring New Material, Mistral strengthens the teams responsible for this work. OpenAI and Anthropic are also investing in the same direction.

A company can test an artificial intelligence model in minutes. Integrating it into the work of a legal department, a sales team or a design office requires more. You must choose the tasks to entrust to him, give him access to the right information and organize the verification of his results.

It is on these questions that New Material must strengthen Mistral, the studio has just announced its acquisition. His team will form a new entity, Solutions Design, to design and develop tools with clients. The amount of the operation and the staff involved have not been communicated.

From model to application, there remains a tool to build

In a legal department, if a model can summarize a contract, to make it a useful tool, several questions remain to be resolved: Where are the documents located? Who has the right to consult them? How to find the clause that justifies a response? How to compare two versions and validate a modification?

The model provides analysis capacity when the application organizes its use: accessible documents, possible actions, displayed results and controls. If the lawyer has to move from one software to another and copy everything, the tool may lose some of its interest.

Building an AI product means understanding what users need to accomplish, then building the software that helps them do it. This concerns the screens, but also the functioning of the tool and its integration into the company. The role of New Material consists precisely in connecting these two dimensions.

Mistral wants to learn from each deployment

The company has developed strong expertise in the field, New Material boasts six years of experience in sectors such as automotive, commerce, law and health. At Mistral, Solutions Design must work with customers and convey their needs to the teams who develop the products.

A difficulty encountered in one project can thus lead to a useful improvement elsewhere. If several companies need to find the documents on which a response is based, Mistral can develop a common function, rather than rebuilding it for each client.

The company already offers services to choose uses, adapt models and put them into production.

After calculation and industry, strengthen tool design

New Material constitutes Mistral’s fourth known acquisition in 2026. The three previous ones brought it different skills: Koyeb for running applications on its infrastructure, Emmi AI for simulating physical phenomena and Pimento for its experience with creative tools using AI.

These operations expand what Mistral can provide to a business, namely a model, the means to make it work, and the help needed to build an application. New Material reinforces this last step.

OpenAI and Anthropic are also investing in the creation

In August 2023, OpenAI acquired the Global Illumination team, which was building creative tools and digital experiences using AI. Its collaborators have joined the teams working in particular on ChatGPT. As with New Material, the operation brings experience in product design and development.

OpenAI also announced, in May 2026, an agreement to acquire Tomoro, an engineering company specializing in applied AI. Around 150 engineers and specialists were to join OpenAI Deployment Company, a structure responsible for helping companies build and deploy their systems.

At Anthropic, a services company created with Blackstone and Hellman & Friedman in particular, announced in May the acquisition of Fractional AI. This team must help companies integrate Claude into their activities.

These acquisitions meet the same need: to have people capable of understanding how a company works and building the applications that will be useful to it. Thus, laboratories are strengthening their presence in carrying out projects, alongside integrators and specialized firms.

A more direct customer relationship, but additional work

Especially since for Mistral, participating in the construction of a tool can open up a broader commercial relationship. The company better understands the customer’s needs and can offer software, calculations and support. A successful first application can also facilitate the launch of other projects.

This strategy involves defining the place of partners. Integrators and business editors are already doing part of this work. Mistral will have to decide which projects to deal with directly and which to entrust to them. OpenAI and Anthropic also involve partners in their systems: strengthening internal teams does not mean being able to carry out all deployments.

The economic risk is concrete, if each new client requires months of specific development, Mistral will have to increase its workforce at the pace of the projects. Revenues will be able to increase without the cost of production decreasing.

The challenge will therefore be to reuse what has been built: a connection to software, a verification function or a test method. The commissioning time, the integration cost and the effective use of the tools will make it possible to judge the result of this acquisition.

New Material should help Mistral deliver applications that find their place in daily work. Its contribution will be all the more valuable if what it builds for one company also simplifies the project of the next.

Founded in 2023 by Arthur Mensch, Guillaume Lample and Timothée Lacroix, Mistral AI announced in September a Series D of 3 billion euros, for a valuation greater than 21 billion euros. Samsung Electronics led the round, with Scaleup Europe Fund, managed by EQT, and PSG Equity as co-lead managers. The amount spent on the acquisition of New Material has not been made public.