HomeArticle

36Kr Exclusive | Shangtang Guoxiang Invested in a Consumer-Grade Spatial Camera Company for Collecting Real-World Data for Embodied Intelligence

欧雪2026-05-25 10:25
Centered around Camera + AI.

Author: Ou Xue

Editor: Yuan Silai

Yingke learned that Zhuma Innovation, a consumer-grade spatial camera company, recently completed tens of millions of yuan in Series Angel+ financing. This round was led by SenseTime Guoxiang Capital, followed by CDH VGC and Fengrui Capital. Shendu Capital served as the exclusive financial advisor for subsequent financing. The funds will be mainly used for product R & D, mass production preparation, and overseas market expansion.

Zhuma Innovation was founded in November 2025. It is a spatial intelligence company with Camera + AI at its core. The company hopes to define a new category of AI hardware - the spatial camera: it is not only a consumer-grade 3D content creation tool but also a real-world 3D data entry point for embodied intelligence and world models.

In terms of the team, Zhang Ji, the founder, once served as the vice president of Qunhe Technology and was deeply involved in large-scale spatial data processing and commercialization. Guo Jie, the chief scientist, is a tenured associate professor at Nanjing University and a top scientist in the fields of 3DGS and spatial simulation. The team has full-stack capabilities from 3D data collection to cloud computing, from algorithms to hardware.

According to Zhang Ji, the company's first product, Pebble, encapsulates 3D Gaussian Splatting, multi-sensor fusion, edge-cloud collaboration, and AI-native interaction into a consumer-grade device, allowing users to capture the real 3D world as easily as shooting a video, achieving "what you see is what you scan, and what you scan is what you get." Currently, the product is still in the R & D stage and has not been mass-produced.

With the rise of embodied intelligence and VLA large models, the industry's demand for real-world 3D data is undergoing a structural change. In the past, robot training relied heavily on teleoperation data, which was costly and difficult to scale. More and more people believe that first-person human videos, spatial sensor data, and continuous behavior data in real environments will become the new "staple food" for model training.

This is precisely the core positioning of Zhuma Innovation: to become the data collection infrastructure in the robot era by collecting data with first-person perspective, 3D structure, semantic information, and real scale through the spatial camera.

This type of data can be used to build more realistic spatial world models, help robots understand the environment layout, object relationships, and task context, and can also feed back into embodied intelligence simulation and real-time spatial perception training.

The following is an excerpt from the dialogue between Yingke and Zhang Ji:

Yingke: How will the introduction of SenseTime and CDH in this round help the company's industrial layout?

Zhang Ji: We currently see two clear opportunity windows: one is that 3DGS technology is moving from the laboratory to consumer-grade applications, and the other is that the demand for real spatial data in the embodied intelligence industry is about to explode.

The entry of SenseTime and CDH is not just a financial investment. SenseTime has strong talent and algorithm accumulation in the AI field, and our capabilities in spatial intelligent reconstruction can form a strong alliance. CDH has a deep layout in the robot field. They hope that we, as the spatial intelligence infrastructure, can connect with more robot companies. These two investors can form a very strong complementary relationship with the company, which is what we value most.

Yingke: The consumer-grade 3D reconstruction market has never really taken off in the past. Do you think the inflection point has arrived?

Zhang Ji: We are more optimistic than before. After the last round of financing was announced, a large number of people from different industries came to communicate with us and put forward many real needs. Previously, people didn't realize that 3D reconstruction could be done like this, but now the perception is obviously changing.

In addition, industrial-grade 3D equipment companies have realized the existence of the market but have not chosen to compete directly. Many are discussing the possibility of in-depth cooperation with us. Some startup teams are just starting to enter. Although there is competitive pressure, I believe that more people entering the market will definitely contribute to the rapid development of the consumer-grade market.

The real competition in the current industry is about speed and the solidity of technology - whoever can clearly define the product and make the user experience work first will gain a first-mover advantage. We have obvious product progress every month, and this rhythm is very important.

Yingke: What role will the company play in the embodied intelligence ecosystem in the future?

Zhang Ji: Our core strategy remains unchanged. We still aim at professional groups such as spatial designers and consumer-grade users, and based on our capabilities in spatial intelligence and spatial computing, we provide consumer-grade spatial cameras. However, in this process, we have naturally accumulated something very scarce in the embodied intelligence industry - 3D data of the real physical world.

Currently, most embodied intelligence companies are training in simulated environments, and the authenticity of the synthesized data is greatly reduced. The real spatial data we collect is photoreal, with 3D structure, semantic information, and physical correctness, which is a good infrastructure for them. This is not a deliberately planned second growth curve but a natural overflow of the technology stack in the process of completing our main business. So we won't do embodied intelligence at the product level, but we will make some arrangements and openness at the technical research level.

Views from investors:

Yu Jun, a partner at SenseTime Guoxiang, said: The computability, comprehensibility, and editability of the physical world are the foundation and base for future application development, and each individual will also become a contributor and user of this base. Mr. Zhang's team is particularly suitable for providing users with solid and accessible ultimate product solutions. The team also has a deep understanding of the early user group. We look forward to Zhuma becoming the Insta360 for perceiving and computing the "4D world."

Mou Liushan, the vice president of CDH VGC, said: The integration of the physical and digital worlds is an irreversible trend, and the 3DGS camera is the bridge connecting the two. Currently, the breakthrough in 3DGS technology and the sharp decline in lidar costs have brought about a market inflection point for the rapid development of 3DGS cameras. With a sharp eye and deep accumulation, the Zhuma team entered the market at the right time with a precise product definition. In the future, it can grow into the data infrastructure for embodied intelligence and has more possibilities. We look forward to witnessing the future where the real world is recaptured and understood together with the Zhuma team.

Ma Rui, a partner at Fengrui Capital, said: Congratulations to the company for completing the new financing in just one month. Fengrui continues to increase its investment because we are very optimistic about the founder Zhang Ji and the team's execution ability. We look forward to Zhuma launching a consumer-grade spatial camera and becoming a new data entry point for the world model.