WeiTai represents one of the earliest strategic moves by Xiaomi into the field of embodied intelligence.
Recently, WeiTai Robotics, established just two years ago, announced the completion of its Pre-A round of financing. This round was led by the Red Bird Qihang Fund, with participation from industrial investors including Lingyi Investment, Yongxin Fangzhou, Weihao Chuangxin, and Ningbo Jide, alongside co-investment from Shanghai Angel Club. Several existing shareholders also continued to provide capital support. Concurrently, the next funding round is nearing completion, bringing the cumulative financing amount to several hundred million yuan.
This financing round is primarily composed of industrial capital, including funds related to Xiaomi's ecosystem companies and leading manufacturing enterprises. Notably, Weihao Chuangxin is backed by Will Semiconductor and had previously focused exclusively on the chip sector; this marks its first foray into embodied intelligence. Ningbo Jide is a key enterprise within the Xiaomi ecosystem, and this also represents its first investment in the embodied AI domain.
Last year, WeiTai completed its angel round and angel+ round, led by Xiaomi and Vertex Ventures respectively. Even earlier, its seed round secured investment from a leading AI figure.
WeiTai is indeed one of Xiaomi's earliest forays into embodied intelligence. Earlier this year, Xiaomi also led a funding round for Zibian Robotics. More recently, it has participated in the initial funding round for the physical AI company SynapX and the Pre-A round for dexterous hand company Xinuo Future, continuously strengthening its position in the embodied intelligence industry chain.
The Shift from Software AI to Physical AI
"Over the past year, embodied intelligence and physical AI have indeed entered a critical window of opportunity," stated Li Rui. "The essence of this window is that AI is transitioning from a purely virtual world into the real physical world."
He further explained, "In the past two years, capital was more focused on the robot's body or its 'brain.' But as robots need to perform real work, there's a growing realization that they must possess human-like tactile sensing and hand-eye coordination capabilities."
This shift is evident in both capital markets and client demand. Li Rui revealed, "From the second half of 2025 to now, over half a year, we have experienced an unprecedented surge in attention and a dramatic increase in client demand. The orders we have secured in recent months, along with those currently being signed, far exceed the total for the entire previous year."
Expanding Product Portfolio
In terms of its product matrix, WeiTai launched four new vision-tactile products this year, covering the complete chain from perception to actuators and data acquisition.
The most notable product is the GF515—a vision-tactile biomimetic fingertip designed for five-finger dexterous hands. This product achieves high-resolution tactile imaging, high-frequency perception, and multi-dimensional force information acquisition within a 15mm wide artificial fingertip space. "A dexterous hand without tactile sense is more like a mere moving mechanical hand. With vision-tactile capability, the dexterous hand truly begins to possess fine manipulation abilities," Li Rui described. "The significance of GF515 is that it gives dexterous hands a sense of touch akin to human hands."
Additionally, the GF220 is designed for two-finger gripper scenarios, the Gripper 90 is an integrated vision-tactile gripper actuator, and the UMI85 is a vision-tactile data acquisition device used to provide real physical interaction data for embodied intelligence models.
Dual Commercialization Pathways
For commercialization, WeiTai primarily follows two main paths. The first is entering real intelligent manufacturing scenarios. A typical example is the deep collaboration with Xiaomi's smart manufacturing factory, applying vision-tactile hand-eye coordination capabilities to fine manipulation tasks in mobile phone assembly. "Such tasks are difficult to solve with vision alone. The real challenge occurs at the moment of contact: whether contact is made, if the force is appropriate, if there is slippage, or if damage might occur. All of these require tactile feedback," Li Rui stated.
Beyond Xiaomi, WeiTai is also deploying its technology in various scenarios including 3C electronics, automotive, logistics, new energy, and food, engaging in deep collaborations with multiple leading clients.
The second path involves ecosystem partnerships with robotics companies. Recently, a new generation product from a dexterous hand company valued in the tens of billions also adopted WeiTai's vision-tactile technology. The fingertip of this product integrates vision-tactile sensors, capable of sensing deformations as small as 130 micrometers. "We are not in competition; it's an enabling relationship. They solve the hand's structure, degrees of freedom, drive, and execution capabilities. We solve the tactile perception, force feedback, and control loop when that hand contacts the physical world. We are building the tactile version of NVIDIA," Li Rui explained.
Regarding current high-demand application scenarios, Li Rui categorized them into three major types: adaptive grasping, precise placement, and flexible object manipulation. He used a vivid analogy: "These three types of skills correspond to the three stages of a child's growth—from grabbing anything, to learning to assemble Lego blocks, to folding clothes and cooking."
A Pivotal Year for Scaling
As the embodied intelligence sector heats up, more entrepreneurs are entering the vision-tactile field. Simultaneously, some well-funded leading embodied AI companies also possess the capability for in-house vision-tactile R&D.
When asked about industry competition and the possibility of being overtaken, Li Rui responded with certainty, "Not likely. Vision-tactile is a highly systematic capability involving hardware, software, algorithms, and extensive engineering know-how. Even for a team like ours with over a decade of accumulated experience, we still encountered many pitfalls during product development. These pitfalls cannot be quickly resolved simply by throwing more people or money at them."
It was revealed that a leading embodied intelligence company attempted in-house R&D of vision-tactile technology but ultimately abandoned the effort, choosing instead to collaborate with WeiTai. "During this window of opportunity, a better choice is division of labor and collaboration, with each party focusing on what they excel at."
The founding team's background provides confidence. During his PhD at MIT, Li Rui, along with his advisor, pioneered the research direction of robotic vision-tactile sensing and invented the world's first vision-based tactile sensor with resolution surpassing that of the human finger. "From the initial innovation to subsequent continuous iterations, we have been leading the development direction in this field," Li Rui said.
WeiTai has defined 2026 as a pivotal year.
"If 2024 was the starting point for the company's establishment and technology productization, and 2025 was a year of customer validation and accelerated commercialization, then 2026 will be the critical year for WeiTai to transition from product delivery to scaled implementation," Li Rui added. "The most important change in 2026 will be that the industry shifts from 'seeing who can talk a bigger game' to 'seeing who can deliver more solidly.' This is an opportunity for WeiTai because our technology was born for real-world operations."
Global Trends and Consensus
Globally, the US and China are exhibiting different development paths in embodied intelligence. The US is strong in the AI 'brain,' while China excels in industrial 'hands and feet.' Vision-tactile technology serves as the neural interface connecting intelligence to the physical world.
From WeiTai's perspective, regardless of the differing paths, a consensus is becoming clear: for robots to enter the physical world, they must have better perception and manipulation capabilities. "Vision is important, but it cannot solve all contact problems. In scenarios involving in-situ operations, flexible objects, precision assembly, and force-controlled inspection, tactile sensing is indispensable."
Recently, international giants like Meta, Google, and Amazon have identified vision-tactile sensing as a key component of next-generation embodied intelligence. Furthermore, the new generation dexterous hand from the global unicorn Sharpa comes standard with integrated vision-tactile sensors. Domestically, Tashi Zhihang recently launched the OmniVTA vision-tactile world model.
It is evident that top global physical AI products have completed technological iterations, moving decisively away from pure visual, open-loop operation modes.
Strategic Financing Philosophy
Reflecting on this funding round, Li Rui, founder and CEO of WeiTai, noted the most significant change was the rapid evolution in investor understanding.
"During this financing process, many investors asked very in-depth questions," he recalled. "They didn't just ask about market size and sensor parameters; they also asked many questions related to mass production—customer validation, algorithm closed loops, data acquisition, and the path for future integration with large models."
This contrasts sharply with the situation two years ago. "Two years ago, we had to educate many investors about what tactile sensing is and its potential applications. But in the past year and a half, we've clearly felt that investors already have a certain conceptual understanding of this field, and some even proactively sought us out after conducting a holistic analysis of embodied intelligence."
What impressed Li Rui even more was that some industrial investors, when persuading them to accept funding, would specifically explain, "I have such-and-such a scenario where your sensors and algorithms can be applied, and this technology can be used in my products or production lines."
From WeiTai's perspective, this indicates that capital's understanding of embodied intelligence is deepening, moving beyond mere concepts. More and more investors recognize that vision-tactile is not just a sensor accessory on a robot, but an entry point for perception, control, and data closed loops in the era of physical AI.
Regarding the use of this new funding, WeiTai plans to focus investments in three key areas: iteration of vision-tactile perception technology and core components, actuator implementation, and vision-tactile data collection and model closed loops. "Future robots must possess generalized operation capabilities, which requires learning from real contact data."
Maintaining Discipline Amidst a Hot Market
While financing in the embodied intelligence sector is booming, with phenomena like "the next round being finalized before the current one ends" frequently appearing in the market, WeiTai maintains a rare sense of clarity.
Li Rui admitted, "This phenomenon does exist in the market, and it also indicates that the embodied intelligence and physical AI sectors are very hot. During our financing process, we also clearly felt the high level of capital attention. Many investors hoped to participate early and continue to follow up."
However, he emphasized, "Internally, we have always remained quite clear-headed. Financing cannot become the company's main focus. The company's main focus will always be technology, product, customers, and delivery. A financing pace that is too slow might cause us to miss the industrial window, but a pace that is too fast might also cause the company to be led by valuation."
The most distinct difference in hard-tech entrepreneurship compared to other industries is that it's not about raising the valuation as high as possible, but rather whether the company has the capability, after each funding round, to use new milestones to support its value in the next stage.
Li Rui drew an analogy, "Financing is just refueling. The real competition is in the factories, on the production lines, in the hands of the customers." Simultaneously, securing funding quickly does not mean spending it quickly, and a high valuation does not grant permission to make fundamental mistakes. In the hard-tech industry, the greatest fear is not a lack of funds, but that money causes the company to lose its discipline.
Reflecting on the entire two-year entrepreneurial journey, Li Rui concluded, "The goal of a general-purpose robot is somewhat like climbing Mount Everest, but we are laying eggs along the way. We are not tackling the most difficult problems head-on from the start, but rather advancing towards this goal in a rhythmic, phased manner. Truly powerful robots in the future will not be the victory of a single technological path, but the systematic fusion of the brain, eyes, hands, tactile sense, control, and data."
Comments