Samsung Electronics has partnered with French artificial intelligence (AI) company Mistral AI to develop a specialized AI model for semiconductor design and manufacturing. This initiative moves away from simply adopting external general-purpose AI, focusing instead on creating a dedicated model that learns from and utilizes Samsung's semiconductor data internally, applying it directly in production.
On September 8, Samsung Electronics announced the strategic partnership with Mistral AI. The two companies will combine Samsung's Device Solutions (DS) division's semiconductor technology and manufacturing data with Mistral AI's AI models to jointly develop an optimized model and operational platform for semiconductor tasks.
The areas of application include data analysis, defect prediction, and process optimization. Samsung plans to adapt Mistral AI's large language model, Mistral Large, to fit the DS division's data, aiming to shorten design timelines and enhance manufacturing efficiency and quality.
A key aspect of this collaboration is the operational approach. The companies will implement an on-premises model, running AI on their own servers and data centers. This allows for the analysis of critical technical data generated during semiconductor design and manufacturing without transferring it to external cloud services, thereby reducing the risk of technology leakage.
On the same day, Mistral AI announced it raised €3 billion in a funding round that included Samsung Electronics, valuing the company at €21 billion. This investment marks one of the largest equity raises for a private technology firm in Europe within just three years of its founding, with Samsung participating as a new investor.
The collaboration between the two companies will extend across the entire DS division, including memory, foundry, and logic sectors. They plan to build an ecosystem that applies AI from semiconductor design to manufacturing, recognizing that the ability to quickly analyze production data and reflect it back into design and processes is becoming a key competitive advantage as semiconductor processes become more refined and manufacturing variables increase.
Mistral AI, founded in 2023 by researchers from Google DeepMind and Meta, focuses on industries with high security requirements, such as finance, manufacturing, and energy, leveraging its technology to operate models internally.
Jung Hyun-joon, head of Samsung's DS division, stated, "As the complexity of semiconductor design and manufacturing rapidly increases, the ability to safely and precisely utilize vast amounts of data is emerging as a core competitive advantage. We will build AI models and platforms specialized for the semiconductor industry."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.