Liang Wenfeng's junior fellow student, with a valuation of 10 billion
Post-2000s talent from Zhejiang University has secured another round of financing.
According to disclosures from multiple media outlets, Moxin Technology, a 4D world model enterprise founded by Chen Tianrun, a post-2000s PhD candidate at Zhejiang University, is about to close a new round of financing. The total financing amount may reach the level of 10 billion yuan, pushing its valuation up to nearly 100 billion yuan.
This marks the fifth round of financing completed by Moxin Technology in 2026.
Back in August, the company just finished a 500 million yuan Series A financing, which attracted institutional investors including state-owned capital institutions, industrial capital and financial investors to participate in the subscription. Its total accumulated financing this year has reached nearly 10 billion yuan, with its valuation already jumping to nearly 40 billion yuan.
This means that in just one month, the scale of the company's new round of financing has caught up with the total financing amount of the past 8 months, and its valuation has risen from nearly 40 billion yuan to nearly 100 billion yuan, a 1.5-fold increase, successfully making it a unicorn enterprise.
Behind the accelerated financing progress of Moxin Technology, there are two key points that cannot be underestimated.
First, the world model track that the company has bet on is heating up rapidly, and this round of financing is advancing against the favorable market trend.
As foundational large models gradually mature, the industry focus has shifted to how AI understands the physical world. The world model is regarded as the underlying threshold that must be crossed before the large-scale implementation of embodied intelligence and humanoid robots, and has also become a new investment hotspot in the primary market.
Second, Chen Tianrun studied at the College of Information and Electronic Engineering of Zhejiang University for his bachelor's degree, making him a fellow alumnus of Liang Wenfeng, the founder of DeepSeek. This label has also added a lot of highlights to the development story of Moxin Technology.
Post-2000s talent from Zhejiang University bets on 4D world model, valuation surges 1.5 times in one month
The entrepreneurial starting point of Moxin Technology was a manually assembled 3D printer on the campus of Zhejiang University.
Chen Tianrun graduated from the College of Information and Electronic Engineering of Zhejiang University with a bachelor's degree, and was later recommended for direct doctoral study at the School of Computer Science of Zhejiang University, under the supervision of Pan Yunhe, academician of the Chinese Academy of Engineering and former president of Zhejiang University.
He has long been deeply engaged in the field of 3D vision, and has published more than 50 papers in journals including *Science*, *Nature Photonics*, as well as top conferences including ICCV and CVPR. His research directions cover 3D visual reconstruction, spatial intelligence and world model algorithms, making him one of the young domestic researchers who focused on the 3D spatial intelligence direction at an early stage.
Back in his undergraduate years, Chen Tianrun developed a strong interest in 3D printing.
At that time, household 3D printing devices were generally expensive, and the modeling operation threshold was high, making it difficult for ordinary users to use. In 2019, Chen Tianrun, who was still an undergraduate, manually assembled his first 3D printer in his dormitory at Zhejiang University, trying to lower the usage threshold of household 3D printing.
In June 2021, Moxin Technology was officially established, launching KOKONI, a consumer-grade intelligent 3D printing product equipped with AI-assisted modeling capabilities. The product went global and gained a large number of overseas C-end users, completing the first round of commercial verification.
It is also the years of accumulation in the 3D field that allowed Chen Tianrun to see greater opportunities.
During the R&D process of KOKONI products, Moxin continued to accumulate technologies related to 3D reconstruction and 3D scene data generation, and built up native 3D data assets.
Chen Tianrun judged that both embodied intelligent robots and autonomous driving require underlying models that can simulate and predict the operation rules of the real physical world, namely world models.
Therefore, at the end of 2024, Moxin Technology made a strategic adjustment, shifting the company's core business to the spatial intelligence track, focusing on tackling key problems of 4D world models.
At the end of 2025, Moxin released the KOKONI-World world model prototype; in July 2026, the company jointly released the MoWorld 4D world model with the National Artificial Intelligence Application Pilot Test Base and Huawei Cloud.
According to the company's technical report, this model has about 14 billion parameters, adopts the MoE Mixture of Experts architecture, and completes training, compression and deployment on Huawei Ascend NPU super nodes, achieving a maximum real-time inference speed of about 50 frames per second, with the inference cost being only 30% of that of the imported GPU solution of the same scale.
At present, MoWorld has been applied to fields including industrial digital twin, embodied intelligence training, cultural tourism and entertainment.
Prior to this round of financing, Moxin Technology had successively completed 4 rounds of financing within 8 months, with a total accumulated financing of nearly 10 billion yuan. Its investor structure is diversified, covering state-owned capital funds, industrial capital and technology VCs.
Among them, the state-owned capital side includes Shenzhen Venture Capital and Zhejiang Venture Capital; industrial capital includes Huawei Hubble, Legend Holdings, Fullhan Microelectronics and SMIC Pulead; technology VCs include institutions such as CDH Baifu, Loongson Venture Capital, Tunlan Investment and Xingrui Capital.
According to media disclosures, the new round of financing of Moxin Technology continues the previous feature of intensive participation of industrial capital, with state-owned capital and market capital further increasing their investment intensity. The investor lineup covers multiple fields such as chips, computing power, intelligent terminals, digital content and advanced manufacturing.
This full-dimensional capital matrix support is essentially a collective bet by core players in the upstream and downstream of the industrial chain on the industrialization value of world models, which means that the multi-party joint layout of world model industrialization has entered a critical stage, and at the same time confirms the scarce value of Moxin Technology in the track.
Post-2000s entrepreneurs are spawning a group of AI unicorns
In the AI track, unicorns led by post-2000s talents that rise rapidly like Moxin Technology are not unique cases.
For a long time in the past, the main force of domestic AI entrepreneurship was senior technical backbones from large factories or scientific researchers with many years of industrial experience.
However, in the past two years, as the focus of the AI industry has extended from foundational large models to cutting-edge segmented directions such as world models, verifiable reasoning and embodied intelligence, a group of post-2000s scientific and entrepreneurial talents have grown rapidly, and some enterprises have taken the lead in entering the 10-billion-yuan valuation range.
Among them, Axiom founded by Hong Letong, the "post-2000s genius girl", is one of the typical representatives.
She dropped out of her Stanford PhD program to start a business, founded the artificial intelligence company Axiom, which focuses on formal verification, uses mathematical methods to solve the problem of large model reasoning hallucinations, and realizes that every step of AI reasoning can be verified.
In March this year, the company completed a 200 million USD Series A financing, led by Menlo Ventures, with follow-on investments from old shareholders including Greycroft, Madrona Venture, B Capital and Toyota Ventures. Its post-investment valuation reached 1.6 billion USD, making it a unicorn in just over one year after its establishment.
In the embodied intelligence track, Jia Xiaoyou, co-founder of Lingxin Qiaoshou, is another post-2000s representative.
She studied environmental design at Beijing Forestry University for her bachelor's degree, and participated in entrepreneurship during her studies. In 2023, she founded Lingxin Qiaoshou together with Zhou Yong, leading the appearance and structural design of the dexterous hands for humanoid robots.
In February this year, Lingxin Qiaoshou completed a nearly 1.5 billion yuan Series B financing, with its valuation exceeding 10 billion yuan. It is already the enterprise with the highest valuation in the dexterous hand field, and its products have now entered enterprises including Huawei, Xiaomi and Foxconn.
Behind these newly emerging 10-billion-yuan valuation unicorns is the supporting trend that a post-2000s entrepreneurial legion is emerging in the AI track.
Cao Xi from Lisi Capital once admitted at the 20th China Investment Annual Conference that the generational shift in the AI industry has been completed: "In the investment portfolio of our fund, the age of founders has rapidly decreased from post-1995s to post-2000s."
And this has been corroborated by relevant data.
According to the report from early-stage venture capital institution Antler, the average age of AI unicorn founders has dropped sharply from 40 in 2021 to 29 in 2024. In Q1 2026, the proportion of founders under 30 years old in AI financing projects has reached 34%, an increase of 22 percentage points compared with 2024.
The industrial logic behind this trend is clearly visible.
On the one hand, world models, embodied intelligence and verifiable reasoning are brand new tracks without heavy historical industrial burdens. The research results accumulated by young researchers in cutting-edge laboratories can be transformed into products faster, seizing the first-mover advantage in new tracks.
On the other hand, funds in the primary market are shifting out of the red ocean of foundational large models, and a large amount of capital flows to new directions of AI applications and spatial intelligence, providing a capital soil for young scientific and entrepreneurial talents.
In other words, as the "AI-native generation", post-2000s will grow and co-evolve with the AI era.
This article is from the WeChat official account "Lieyun Select", written by Sun Yuan, and published by 36Kr with authorization.