DeepSeek Founder Expands Into Physical AI
DeepSeek founder Liang Wenfeng is reportedly planning to lead an investment round in robotics AI startup Winverse AI, following a prior investment of over 100 million yuan in Unitree Technology. The moves signal that the large language model company is extending its technology and capital reach beyond digital intelligence into the physical world.
From Language Models to Embodied Intelligence
DeepSeek has built its reputation on advanced large language models, but the next frontier for AI lies in embodied systems—machines that can perceive, reason, and act in physical environments. Winverse AI focuses on bridging AI models with robotic hardware, a field that demands massive computing resources, sophisticated simulation, and real-world data. Liang’s personal and corporate investments suggest a strategic pivot toward integrating AI with robotics, potentially creating a feedback loop where language models enhance robotic control and physical data improves AI reasoning.
Industry Implications
This shift reflects a broader trend in the AI industry: the convergence of digital intelligence and physical automation. Companies like Tesla, Boston Dynamics, and Figure are racing to commercialize humanoid robots, while Chinese firms such as Unitree and Winverse AI are advancing rapidly. For DeepSeek, investing in robotics could diversify its revenue streams and secure a foothold in the emerging embodied AI market. It also positions the company to compete with global tech giants that are vertically integrating AI and hardware.
Forward-Looking Perspective
If successful, DeepSeek’s foray into embodied AI could accelerate the deployment of intelligent robots in logistics, manufacturing, and services. However, challenges remain: hardware costs, safety regulations, and the need for robust AI models that can operate reliably in unstructured environments. As Liang Wenfeng continues to deploy capital, the market will watch whether DeepSeek can translate its language model expertise into tangible physical-world applications—and whether this strategy yields a new wave of AI-driven automation.
