Dexmal, a company focused on embodied intelligence, announced the completion of its Series A and A+ funding rounds recently. The Series A round was led by NIO Capital. The proceeds will accelerate the company's R&D and deployment of integrated hardware and software for intelligent robots.

"We believe Physical AI, which extends intelligence from the digital space to the physical world, will reshape productivity structures and human life," said Yao Li, Partner at NIO Capital. "We are confident that the embodied intelligence sector will leverage China's structural strengths in supply chain and talent to drive real-world breakthroughs. The Dexmal team is a top-tier team that has navigated industry cycles while remaining at the forefront of AI development. They address the critical first step of embodied intelligence deployment with systematic thinking, continuously driving the dual flywheels of data and economic viability. The Best Model is Business Model. We look forward to Dexmal further accelerating the industry's growth trajectory."

Founded in March 2025, Dexmal is dedicated to the R&D and practical application of embodied AI hardware and software technologies, with a mission to “build intelligent, useful and trustworthy robots to make our life better.” The core team possesses a rare combination of top-tier AI academic credentials and over a decade of experience in scaling native AI products. They have extensive expertise in algorithm development, hardware R&D, data management, engineering innovation, and scenario deployment, alongside a unique integrated strength in "algorithms + hardware + scenarios." The company has also published over ten papers at top AI academic conferences and won multiple global championships in the CVPR 2025 and ICRA 2025 robotics challenges.

Notably, the team has substantial experience in deploying AI-powered logistics robots. Leveraging its expertise in smart logistics robotics technology and flexible warehouse automation products, the company has not only driven the intelligent upgrade of the logistics industry but also established first-mover advantages in software, hardware, and product development for its broader embodied AI initiatives.

Building on these strengths, Dexmal has independently developed an end-to-end multimodal embodied AI model, MMLA. This sophisticated system seamlessly integrates the capabilities of multiple components, including multi-sensor data, Large Language Models (LLMs), Vision-Language Models (VLMs), Vision-Language-Action Models (VLAs), and World Models, to enable intelligent generalization across different scenarios, tasks, and configurations.

Currently, Dexmal has open-sourced Dexbotic, a PyTorch-based VLA toolbox, providing a one-stop research service for practitioners in the embodied AI field. It has also launched DOS-W1 (Dexbotic Open Source-W1), an open-source robotics hardware product designed to significantly lower the barrier to robot usage and enhance the convenience of robot maintenance and modification. 

Furthermore, the company partnered with Hugging Face to launch RoboChallenge, the world's first large-scale real-robot evaluation platform for embodied intelligence. Dexbotic, DOS-W1, and RoboChallenge are forming a deeply synergistic ecosystem, actively advancing the embodied robotics industry from the perspectives of software, hardware, and standards.

Looking ahead, Dexmal will continue to promote the deployment of general artificial intelligence in the physical world through synergistic innovation in algorithm-driven development, hardware design, and achieving scenario closure within the field of embodied AI.