Mistral Robotics Chief Seeks €200M for New AI Startup

Mistral's robotics chief is raising €200 million for a new startup. The venture aims to build AI models for robotics trained on more comprehensive data than competitors.

This fundraising effort follows a recent pre-seed round. Another robotics AI startup secured €8 million in funding.

Mistral itself raised €2 billion earlier this year. That funding was linked with a valuation of approximately €6 billion.

A Mistral spokesperson commented on the new venture: "We are excited to support the development of a new generation of AI."

Details about the new startup's specific technology remain under wraps. The focus is reportedly on creating foundational models for physical systems.

The Competitive Landscape

The new venture enters a field crowded with major players. Companies like Google's DeepMind are developing robotics-specific foundation models (RFMs). These models are trained on diverse datasets, including video and simulation data, to perform complex tasks.

Nvidia is also a key player with its GR00T project, a general-purpose foundation model for humanoid robots. Announced in 2024, GR00T is designed to understand natural language and emulate movements by observing human actions. The model's performance is reportedly benchmarked against other systems.

A technical paper on GR00T states: "GR00T shows the potential to learn from human demonstration."

OpenAI's investment activity highlights the sector's momentum. The company's venture fund backed 1X Technologies, a 776% increase from its €0.3 million investment in 2019.

OpenAI is also reportedly involved in a deal with robotics firm Figure. The agreement focuses on developing AI models for humanoid robots, specifically models for vision, language, and movement, and includes an equity component.

Funding and Model Developments

Recent months have seen significant capital flow into AI robotics. In the US, Sanctuary AI unveiled its Phoenix humanoid robot, while Tesla continues development on its Optimus bot. Other notable projects include Boston Dynamics' Atlas, Apptronik's Apollo, and Agility Robotics' Digit.

In China, ByteDance's AI lab, ByteDance Research, is working on a "robot brain" for general-purpose AI models. DeepSeek's founder, who raised $0.5 billion from Alibaba, is behind the open-source "DeepSeek Coder" model. This model is designed to handle complex coding tasks across multiple programming languages.

Cognition, an Applied AI-lab focused on reasoning, is another contender. It's building an LLM, named Devin, aimed at autonomously handling software development tasks. The company claims its approach is different, focusing on long-term reasoning and planning over short-term accuracy.