Sage Intelligent completes Pre-B round financing of hundreds of millions of yuan, six-arm robot integrates industrial world model

36Kr has learned that the global industrial embodied intelligence robotics company — Shanghai SAGE Intelligent Technology Co., Ltd. (hereinafter "SAGE Intelligent") — recently completed a Pre-B round financing of several hundred million yuan. The investors include Jiantou Investment Co., Ltd., Nantong Investment Management, and Yunqi Capital, three institutions in total. This round of funds will mainly be used for large model R&D, iteration of the multi-arm robot body, production line expansion, and deep expansion of the global market.

SAGE Intelligent was founded in 2017, and its core team combines nearly 20 years of robot R&D experience with top-tier artificial intelligence algorithm expertise. Founder and CEO Zhang Jianzheng holds a Doctor of Engineering degree in Mechanical Engineering (Intelligent Robotics direction) from Shanghai Jiao Tong University, and previously worked at STEP and FANUC.

As industrial manufacturing evolves toward flexibility and complexity, traditional industrial robots are limited by fixed operations and single tasks, while most general-purpose embodied robots struggle to meet industrial-grade requirements for high precision and determinism. SAGE Intelligent is precisely targeting this gap in complex industrial scenarios — such as precision assembly, dexterous packaging, and long-sequence testing.

On the product side, SAGE Intelligent has built a full-form body matrix covering single-arm, dual-arm, four-arm, and six-arm configurations. According to Zhang Jianzheng, the new-generation four-arm robot adopts a spatially symmetric configuration, controlling four robotic arms through one brain, and with advanced coordination algorithms, the efficiency of multi-arm parallel operation has doubled compared with the previous generation. The six-arm robot "Xuanjia" adopts a spatially heterogeneous design with two arms each distributed at the front waist, rear waist, and shoulders, enabling all-around 360-degree coverage of multiple surrounding production lines and achieving dual-workstation parallel operation and four-arm alternation.

"More arms is not always better — just as a table can be stable with at least three legs, and four or five are also acceptable, but more than that is meaningless." Zhang Jianzheng told 36Kr that, supported by spatial configuration theory, six arms is currently the most reasonable layout.

Compared with four arms, the six-arm configuration increases degrees of freedom by 60%. In confined spaces, ensuring dynamic isolation of safe space, avoiding collisions, and achieving coordination of complex processes places extremely high demands on computing power and algorithm effectiveness. At present, this spatially heterogeneous six-arm configuration is unique globally, and its operational complexity has increased exponentially compared with four arms, with efficiency improved by 50%.

At the level of the "brain," the core of embodied intelligence, the VLA (vision-language-action) models widely adopted in the industry currently have the limitation of end-to-end reflexive imitation and lack physical causal understanding, leading to failed scenario generalization and lack of interpretability. SAGE Intelligent's solution is the self-developed industrial world model released at this year's WAIC — XuanShu·SageNexus (WASM).

The core breakthrough of SageNexus lies in internalizing physical causality, achieving a leap from "passive response" to "active reasoning." Architecturally, it uses an Encoder to map heterogeneous data such as vision, point clouds, force feedback, and process documents into a unified physical semantic space, and then a Decoder translates high-level task intentions into executable action sequences and performs physical feasibility checks, thereby effectively suppressing large model hallucinations and avoiding the generation of actions that are unsafe or violate process constraints.

"What we are talking about is the closed loop of five elements — data, body, control, model, and scenario — putting the data flywheel into motion." Zhang Jianzheng emphasized. Based on 360TB of multimodal real industrial data and three-stage training, the model has accumulated a library of 2,000+ scenario skills. Under the extremely stringent determinism requirements of industrial scenarios, the model can achieve reliability of over 99.9%.

In addition, through real-time cloud-edge-device collaboration, SAGE Intelligent has built a three-loop architecture consisting of a cloud-side "slow loop," an edge-side "fast loop," and a safety bottom-line "loop," ensuring stable output of robots in long-horizon complex tasks.

Against the backdrop of computing power anxiety sweeping the industry, SAGE Intelligent has proposed a differentiated path: using algorithm effectiveness to address dependence on computing power. Zhang Jianzheng believes that continuously stacking computing resources will drive up customer investment costs, while improving algorithm efficiency and reducing reliance on brute-force computation is the true measure of embodied intelligence being implemented.

Relying on the maturity of its technology foundation and real application validation in lighthouse factories, SAGE Intelligent's commercialization has seen rapid growth, covering four major industries — 3C semiconductors, automotive, energy, and food and pharmaceuticals — with more than 50 customers that are Global Fortune 500 companies or listed companies.

"In the first half of this year, we have already completed last year's full-year task, and the full-year business has basically doubled, with a growth rate of 100% to 120%." Zhang Jianzheng revealed.

In addition, overseas markets are becoming SAGE Intelligent's second growth curve. The company had considerable business in Mexico last year, and this year began fully developing Southeast Asian markets (Vietnam, Indonesia, Malaysia, Thailand, Singapore), while gradually winning large orders in the European market. It is reported that its overseas intended orders exceed 500 million yuan. To accommodate surging demand, SAGE Intelligent's factory in the Shanghai Maqiao Artificial Intelligence Experimental Zone is currently at full production capacity, and the company plans to expand to about four times the size of the existing site within the next three years.

The following is an excerpt from the interview between 36Kr and Zhang Jianzheng, founder and CEO of SAGE Intelligent (edited):

36Kr: In industrial scenarios, why do customers generally prefer multi-arm robots rather than combinations of multiple single-arm or dual-arm robots?

Zhang Jianzheng: This is mainly determined by customers' real-world scenarios. Especially in the existing production line stock market, space is often very limited, yet efficient production is still required. If multiple dual-arm robots are used, they not only take up space but also make it difficult to achieve a high degree of coordination. Our six-arm robot adopts a spatially heterogeneous design, enabling a single robot to serve four surrounding production lines. For example, when the three arms on the left are working, the two lower arms each hold different workpieces for assembly, while the upper arm holds an electric screwdriver or glue gun for fastening or dispensing. This kind of efficient coordination in a confined space cannot be achieved by stacking single-arm or dual-arm robots. In addition, in packaging scenarios—bagging, inserting manuals, sealing, boxing, and lidding—multiple hands need to work simultaneously, and the six-arm robot can complete these smoothly in one pass. This is a highly challenging process breakthrough.

36Kr: You mentioned "using algorithmic effectiveness to solve dependence on computing power." How is this achieved at the model level?

Zhang Jianzheng: Right now, everyone in the industry is used to desperately stacking GPUs, but customers often cannot accept the high cost. Our large model (XuanShu·SageNexus) does not rely entirely on brute-force computation. First, our data is very pure—all of it comes from actual data in industrial scenarios, which reduces ineffective consumption during model training. Second, we have made extensive optimizations to the algorithm architecture, so that after the model is trained, some parts can be deployed directly on edge devices to execute the trained Skills. By improving algorithm efficiency, we enable the robot to meet 99.9% industrial-grade determinism while greatly reducing dependence on computing power. This is the true core competitiveness of an embodied intelligence company.

36Kr: At present, the embodied intelligence industry is exploring swarm intelligence. What are SAGE Intelligent's plans in this regard?

Zhang Jianzheng: From single-agent intelligence to swarm intelligence, and then to ecosystem intelligence—this is our strategic path from 1.0 to 3.0. We have already entered the swarm intelligence stage. We have now initially achieved swarm control algorithms that enable hundreds of robots to work jointly in a single factory. Next, we will further expand our R&D team through financing, iterate on swarm intelligence, and strengthen multi-agent linkage to achieve tighter coordination and more precise control. Further ahead, what we want to do is ecosystem intelligence—not only controlling hundreds or thousands of robots, but also unifying control with the hundreds or thousands of other devices in the factory.

Investor perspectives:

Nantong Investment Management: The SAGE team has a deep understanding of industrial scenarios and embodied intelligence, and also possesses strong implementation capabilities. Through years of R&D accumulation and business practice, SAGE has built solid core competitiveness at multiple levels, including high-quality industrial data, a large model oriented toward the physical world, and an industrial on-site encoding and decoding system. Its embodied robots have achieved large-scale commercial deployment in multiple industrial production fields such as 3C semiconductors, the automotive industry chain, the energy industry chain, high-end food, and pharmaceuticals, and have won repeat purchases from multiple leading benchmark customers. We believe that "machines replacing humans" in the industrial sector is an inevitable trend. With multimodal perception and autonomous decision-making capabilities, embodied intelligence is accelerating the breakthrough of the rigid bottlenecks of traditional industrial automation. The industrial sector is becoming the core scenario for the near-term deployment of embodied robots. SAGE focuses on industrial production scenarios, has obvious leading advantages, and its business has entered a period of rapid growth.

Wang Feng, Investment Director of Jiantou Investment: The lack of real and effective industrial scenario deployment and related data is the core bottleneck and challenge currently facing embodied intelligence companies. SAGE Intelligent's products have achieved large-scale industrial scenario deployment at more than 100 large state-owned and central enterprises and Fortune Global 500 companies, and have accumulated massive amounts of relevant data, forming a solid technical barrier. At the same time, its core team has demonstrated top-tier industry strength in R&D and engineering implementation. Based on this, we remain optimistic about SAGE Intelligent's development prospects in the industrial embodied intelligence track.

Yunqi Capital stated: Robots on industrial sites must not only adapt to limited production line space and complex processes, but also ensure stability during long-term operation. SAGE has been deeply engaged in industrial scenarios for many years and has continuously iterated its products around multi-arm coordination, forming a product layout covering single-arm, dual-arm, four-arm, and six-arm robots, while accumulating solid technical capabilities in robot bodies, control, and models. The team has achieved batch delivery in multiple manufacturing industries and accumulated rich experience in engineering implementation. We are optimistic about SAGE's ability to combine technological innovation with real industrial needs, and we look forward to the team promoting the large-scale application of industrial embodied intelligence in more production scenarios.