Low profile in publicity does not mean marginalization in investment.
If one were to map the active players in Shanghai's AI venture capital landscape in 2026, Xuhui Sci-Tech Investment would be an easily overlooked yet significant force.
Data does not lie.
From its first traceable public investment on May 16, 2024, to October 30, 2026, Xuhui Sci-Tech Investment completed a cumulative 99 investment deals over nearly 30 months, covering 98 enterprises.
Among these,69 investments were made in the first ten months of 2026 alone, accounting for 69.7% of the total, with an average deal frequency of once every 4.4 days. In August, the firm executed 13 deals, setting a historical peak.
Behind this frequent activity is a district-level state-owned platform established just over three years ago. The manager, Shanghai Xuhui Sci-Tech Investment Private Equity Fund Management Co., Ltd., was founded in May 2023, belongs to the Xuhui Capital system, and is backed by the Xuhui District State-owned Assets Supervision and Administration Commission (SASAC).
It has already secured a place on the investor lists of representative Shanghai AI companies such as MiniMax, StepFun, and Infinigence AI.
Heavy Bet on AI, Unfolding Along the Industrial Chain
In the 99 deals, artificial intelligence is the absolute main investment theme. Calculated directly by Qichacha industry categories, sectors containing "artificial intelligence" account for 42 deals. Observing the core businesses of these enterprises, this main line extends into inference chips, computing infrastructure, robotics, and industry applications.
AI software and large models constitute the most substantial portion.
At the general-purpose large model layer, Xuhui Sci-Tech Investment chose to enter top-tier projects. It secured shares in both MiniMax and StepFun, two first-tier projects. MiniMax listed on the Hong Kong Stock Exchange on January 9 of this year, taking only four years from inception to IPO, setting the fastest record for a global AI company from founding to listing. Meanwhile, StepFun completed a B+ round financing exceeding CNY 5 billion in January 2026, refreshing the record for the single largest financing in China's large model sector over the previous 12 months.
At the vertical application layer, the reach extends to social networking, education, healthcare, gaming, and smart hardware. Specifically, this includes Ouwo Technology in AI social networking, Moments AI in AI education, Tianyi Siwei and Weiyi Zhiyuan in AI healthcare, Coral Planet Technology in AI gaming, Yingyu Technology and MONTEE AI in AI hardware, as well as Yunlian Zhichong and Lingkun Hengchong in the AI pet direction.
Entering foundational models at one end and extending to various application scenarios at the other, the layout does not stop at large model companies themselves.
Embodied AI is the new track with the most concentrated activity, occurring almost entirely in 2026.
The highlight of this segment is not just which star projects appear on the list, but the emerging hierarchy in the layout between model algorithms, robot products, and core components.
On the model and algorithm side, Moushen Intelligence focuses on a "generative general-purpose embodied brain," while Jushen Technology concentrates on the robot "cerebellum" and motion control, targeting decision-making and action execution capabilities respectively. On the product side, TARS and Lingchu Intelligence combine embodied models, data, and robot R&D to explore implementation paths for general-purpose robots; Shuzhi Wanwu targets the emotional companionship needs of the elderly, bringing AI into more specific life scenarios.
On the core component side, Shenjie Technology develops robotic joint actuators, while Linjiedian Innovation Intelligent enters end-effectors such as dexterous hands. From the "brain" and "cerebellum" to joints and "hands," this map focuses not only on what robots look like but also on the critical links that enable them to truly move and work.
TARS is a representative project in this segment, and its large-scale financing reflects the capital heat around embodied AI. The company explores real-world implementations in industrial settings through its general-purpose embodied large model AWE. In 2025, TARS raised a total of USD 242 million across Angel and Angel+ rounds; in April 2026, it completed a Pre-A round of USD 455 million, setting a new industry record for single-round financing.
The semiconductor and chip segment pushes the layout further into the hardware and manufacturing behind AI, all through co-investment.
Yuanchuan Micro focuses on real-time inference chips, Xi Jian Technology enters brain-inspired visual perception, Simin Wei explores memory-computing fusion inference architectures, and Suzhou Xinshinuo provides semiconductor automatic material handling systems and software. Notably, Xi Jian Technology has completed three consecutive rounds of financing, totaling nearly CNY 500 million. Including brain-inspired visual chip companies adds depth to this AI layout—focusing not only on how models think but also on how agents perceive the world.
Beyond the AI main line, the map extends to life health, commercial aerospace, low-altitude economy, and quantum computing. These investments are fewer in number but expand the vision toward several still-growing technical routes.
There are 8 deals in biopharmaceuticals and medical devices, 3 in commercial aerospace and low-altitude economy, 3 in frontier directions such as quantum computing, synthetic biology, and new materials, and the remaining 4 distributed across smart mobility, energy, and industrial support fields.
Among these are enterprises that have secured large financings. Taiyi Liangsheng raised over CNY 100 million in its angel round, Apuqiang raised nearly CNY 1 billion in its Series B, and Tianze Yuntai entered gene and cell therapy. Different projects correspond to different R&D cycles, ensuring this map is not solely driven by short-term hotspots in AI applications.
Xiangdao Mobility extends this map to smart mobility. As a mobile出行 platform controlled by SAIC Motor, it connects ride-hailing operations with Robotaxi exploration, carrying more attributes of serving the local industrial system. The company has initiated its HK IPO.
Overall, AI determines the center of gravity of this map, while directions such as life health, quantum computing, and smart mobility broaden its boundaries. The main line is concentrated, but the vision is not singular.
Rooted in Xuhui, Gradually Radiating Nationwide
Looking at regional distribution, the investment layout shows clear gradient characteristics. Local Shanghai projects account for approximately 82%, with most core AI projects clustered in Xuhui District.
This distribution relates to the positioning of the district-level state-owned platform to serve local industries, and is inseparable from Xuhui's own fertile soil for the AI industry.
Xuhui District itself is the core carrier area for Shanghai's AI industry. Research institutions such as the Shanghai Artificial Intelligence Laboratory and QiZhi Research Institute are located here, and Caohejing Development Zone is an important carrier for the construction of the Shanghai International Science and Technology Innovation Center. The aggregation of research, talent, and enterprises gives investment institutions early access to teams and makes it easier for early-stage projects to find R&D and industrial resources.
Beyond capital, incubation spaces are also absorbing these startup projects. The "MoSu Space" in Xuhui Riverside Center covers a total area of about 22,000 square meters. According to public disclosures in August 2026, it has cumulatively incubated nearly 100 upstream and downstream AI supply chain enterprises and assisted in landing 21 Qingmiao Fund projects. The connection between investment and incubation gives capital support a more concrete landing point.
Being rooted locally does not imply a closed layout. Beyond deepening efforts in Shanghai, Xuhui Sci-Tech Investment selectively invested in 18 out-of-region projects, outlining an investment contour of "centered on Shanghai, extending to the Yangtze River Delta and nationwide."
Although these out-of-region projects are distributed across different cities, their business directions echo the aforementioned industrial layout, including inference chips, semiconductor equipment, and industrial software, as well as large models, low-altitude economy, and synthetic biology. While geography expands, the technical main line does not scatter accordingly.
Yuanchuan Micro in Hangzhou works on AI inference chips, addressing the inference computing demand behind large model applications; Suzhou Xinshinuo works on semiconductor AMHS equipment, entering the automatic material handling link in chip manufacturing; Jiushao Intelligence in Anhui works on high-end industrial software, facing basic needs in industrial R&D design. These few investments fall within the Yangtze River Delta radius and connect respectively to Shanghai's AI, integrated circuit, and advanced manufacturing industries.
Zizhi Engine in Beijing works on general multimodal large models, and Jisi Intelligence enters digital human live streaming, extending to both the model and application ends respectively. Boundary Smart Control in Guangdong works on flight control systems, a key link in the low-altitude economy. Oupait Optoelectronics in Shandong works on high-precision optical devices, with applications in laser communication, semiconductors, and other fields. Wuhan Hesheng Technology in Hubei works on synthetic biology, further extending the layout to the forefront of life sciences.
Putting these projects together reveals not simple geographic expansion, but the search for technologies and enterprises outside Shanghai that resonate with the existing industrial layout.
"Rooted in Xuhui" is the base color, while "radiating to the Yangtze River Delta and selecting nationwide" is the supplement. The investment focus remains local, but the vision for finding projects extends further. Serving regional industries does not require limiting choices to within the region.
Over Half Sole Investments, Behind Two Strategies
More notable is its betting approach. Of the 99 deals, 56 were sole investments, resulting in a sole investment rate of 56.6%.
In the primary market where top projects often attract multiple institutions and co-investments are common, a district-level state-owned platform acted independently in more than half of its investment events. This contrast is more worth dissecting than mere investment numbers.
Among the 56 sole investments, 42 came from the Qingmiao Fund, accounting for three-quarters of the total sole investments. This is directly related to the positioning of the Qingmiao Fund.
Established by Xuhui District in September 2025, the full name of the Qingmiao Fund is Shanghai Xuhui Artificial Intelligence Youth Entrepreneurship Fund, with a total scale of CNY 2 billion, clearly oriented towards "investing early, investing small, investing hard tech, and investing in youth." As a fund focused on early-stage incubation, it aims to identify potential teams early, support product launches, and accumulate a batch of AI industrial resources for Xuhui District.
For such early-stage projects, the starting point of investment is not necessarily a massive co-investment list; it can also involve one institution making judgments and providing support first. The record of numerous independent moves by the Qingmiao Fund echoes its positioning of extending to early stages. By the end of July 2026, the Qingmiao Fund had received over 600 project applications, made decisions on over 100, and invested over CNY 140 million.
Another easily overlooked point is that sole investments do not remain limited to seed and angel stages.
Of the 56 deals, 18 fell in the Pre-A round or later, including11 Series A rounds, as well as Wuhan Hesheng Technology in the A+ round and Jindu Biotech in the Pre-A+ round. Yingyu Technology, Jisi Intelligence, Chaosu Intelligence, and Coral Planet Technology are also among them. The boundary of independent betting is not limited to the initial stage of entrepreneurship but extends to phases where some enterprises continue to polish products and expand markets.
In contrast, the investment records of Guangqi Huichan Phase I and Guangqi Huizhi Phase I show more distinct co-investment characteristics. Of the 17 investments, only 1 was a sole contribution, while the other 16 involved co-investors.
The division of labor between the two strategies becomes apparent. The Qingmiao Fund extends its reach to earlier startup origins, while the Guangqi series participates in key projects primarily through co-investment. One end expands coverage of early-stage projects, while the other connects diverse capital and industrial resources, without needing to bet using the same method.
When institutions with different backgrounds participate in the same round, the significance lies beyond just funds. Industrial capital, financial investors, and local state-owned assets can bring different judgments and resources. For enterprises, these resources have the opportunity to form complements; for Xuhui Sci-Tech Investment, co-investment also allows it to establish connections with a broader investment network.
Placing the investor lists of projects like MiniMax, TARS, and StepFun together reveals the intersection of local state-owned capital, industrial capital, and top-tier market-oriented VCs. Xuhui Sci-Tech Investment stands within this, forming a connection between regional industrial cultivation and key project investment in this map.
This also creates a paradox in its presence. Investment records are becoming increasingly dense, but its name may not always be under the spotlight of financing news. In star projects involving multiple institutions, external attention often falls more easily on lead investors and top-tier institutions. The continuous layout of a district-level state-owned platform actually requires putting multiple records together to see clearly.
But low profile in publicity does not mean marginalization in investment. Its weight depends not only on whether it stands in the lead investor seat, but also on its ability to continuously discover projects and connect investment with space, incubation, and industrial resources.
Nearly 30 months, 99 investments, 98 enterprises. What is noteworthy is not just which star projects appear on the list, but that models, chips, robots, and applications are being placed into an interconnected investment layout. From early-stage teams to key projects, from underlying technologies to scenario applications, the contours of this map are becoming clearer.
The weight of a district-level state-owned platform may be hidden in these sustained investments. It appears not only behind the names of Shanghai AI that have already been seen, but also reserves positions in advance for the next batch of enterprises that have not yet been fully recognized.
Text | Zhou Yu Editor | Yi Da