What? DeepSeek is starting to invest in robot companies again?!!
After investing approximately 141 million yuan in Unitree Technology in August this year, Liang Wenfeng has now turned his attention to a newly established robot AI startup.
Winverse AI.
According to Bloomberg, DeepSeek plans to receive a seed funding round of tens of millions of dollars.
The post-investment valuation is approximately 400 million US dollars, which is equivalent to about 2.8 billion RMB.
Winverse AI was just established this year. Its founder, Hu Han, was in charge of research on visual language and foundational models for embodied intelligence at Tencent.
It is worth noting that this is the second time DeepSeek has invested in the robotics field within just two months.
In August this year, DeepSeek just invested approximately 141 million yuan in YuShu Technology. The two parties also plan to cooperate on embodied intelligence, robots, and large models.
Today, DeepSeek has turned its investment focus towards the AI model team behind robots.
If the financing is completed, it will be one of the largest investments in startups since DeepSeek was established.
A former Tencent scientist started a business, with DeepSeek investing tens of millions of dollars
Let’s first look at the company that DeepSeek is willing to pay for.
Winverse AI was established in 2026, and it is still in the early research and development stage. No specific products or technical approaches have been disclosed publicly yet.
In contrast, its founder Han Hu's academic and industrial experience has already produced many achievements.
According to Hu Han's personal academic homepage, he graduated from Tsinghua University with a bachelor’s degree in 2008.
Obtained a Ph.D. in 2014, and later worked on research at Baidu Deep Learning Research Institute and Microsoft Asia Research.
In December 2023, Hu Han joined Tencent Hunyuan as an outstanding scientist, responsible for the Visual Language Model (VLM) and Embodied Intelligence Base Model team.
During his time at Tencent, he also served as the technical leader for Hunyuan language large model technology and formed the Hunyuan voice team.
In academic research, one of Hu Han’s most representative achievements is the Swin Transformer in the field of computer vision.
In 2021, Hu Han, as one of the co-author of the paper, and his team proposed a hierarchical visual Transformer architecture based on shift windows.
This research improved the computational efficiency of Transformer when processing high-resolution images, enabling it to be more effectively applied to various visual tasks such as image classification, object detection, and semantic segmentation.
With this achievement, the team received the ICCV 2021 Best Paper Award, which is also one of the representative academic honors in the field of computer vision:
In addition, Hu Han was also included in the 2022 MIT Technology Review China “35 Innovators Under 35” list.
Looking back at his research experience, a relatively clear technical path can be seen.
In the early stage, research mainly focused on computer vision and visual basic models. Later, it gradually moved towards multi-modal large models, and during Tencent’s period, it further participated in the development of embodied intelligence basic models.
These research directions are closely related to the visual perception, environmental understanding, and task execution capabilities required by robot AI. They also provide the technical expertise necessary for the establishment of Winverse AI.
As for financing, according to Bloomberg, DeepSeek plans to lead investment in Winverse AITens of millions of US dollars seed fundingMeituan’s Longzhu Capital has also made an investment commitment.
If the transaction goes smoothly, this will be a significant investment by DeepSeek in a startup of this size since it was established over three years ago.
This financing also has a background that is worth noting—
It is understood that Tencent is also one of the potential investors in DeepSeek’s new round of financing. Now, the robot AI company founded by a former Tencent scientist has received attention from DeepSeek and Meituan-related capital.
From large model technology research and development to startup ventures in embodied intelligence, the talent and capital connections among leading tech companies are also beginning to extend more into the robotics field.
And for DeepSeek, this investment continues its layout in the robotics field since this year.
From financing to investment, DeepSeek increases investment in robots
Since this year, DeepSeek has shown a significant increase in activities in the capital market.
In June this year, DeepSeek completed its first round of external financing since its establishment.
According to reports, its first round of external financing amounted to approximately $7.4 billion, ending the phase where it had long relied on the funding provided by Huanfang Quantification.
Two months later, DeepSeek appeared on Yu Shu Technology' IPO strategic placement list.
According to the strategic placement results announced by YuShu Technology on August 6, Deep Seek was allocated approximately 933,400 shares. At the issuance price of 150.80 yuan per share, the subscription amount is approximately 141 million yuan, and the share lock-up period is 36 months.
In this placement, DeepSeek participated as an industrial strategy investor, and signed a strategic cooperation memorandum with YuShu Technology.
The two parties plan to carry out joint research and development on AI large models and embodied intelligence. The focus will be on exploring how to improve robots’ ability to understand complex environments, as well as their generalization capabilities when facing different tasks and scenarios.
At the business level —
When DeepSeek conducts exploration related to embodied intelligence, robot products and technical solutions from Yushi will be given priority under the same conditions. When Yushi carries out model adaptation, training, and integration development, its technical services from DeepSeek will also be given priority.
In addition, DeepSeek will also provide support in areas such as model architecture, AI computing cluster construction, and data center operations, based on actual needs.
In October, there were new developments regarding DeepSeek’s own financing and listing plans.
According to a report by Bloomberg on October 6, DeepSeek is raising a new round of financing of at least 80 billion RMB, with the final scale possibly approaching 100 billion RMB, and a target valuation of approximately 500 billion RMB.
Companies such as Tencent and Envision Energy are also involved. An important reason for this financing is that DeepSeek is preparing to launch an IPO in early 2027, and the listing venue may be the Sci-Tech Innovation Board; however, this has not yet been officially determined.
Now, two months later, DeepSeek has turned its attention to Winverse AI, which specializes in basic robot models.
The technical directions involved in the two investments are also different.
Yu Shu's technical expertise mainly focuses on robot bodies, motion control, and hardware platforms, providing a real-world platform for embodied intelligence technology.
Winverse AI focuses its research and development on basic robot models, aiming to improve robots’ environmental perception, understanding, and decision-making capabilities.
One focuses on the “body” of a robot, while the other targets the “brain” of a robot.
At present, DeepSeek has not disclosed whether there is a technical cooperation plan between the two companies, nor has it announced any self-developed robot products.
However, with consecutive investments in robot body enterprise and embodied intelligence model teams, DeepSeek’s layout in the robotics field has already begun to take shape.
Reference link:
[1] https://ancientmooner.github.io/
[2] https://www.bloomberg.com/news/articles/2026-10-12/deepseek-backs-ex-tencent-ai-scientist-s-400-million-robot-lab