Gasgoo Munich- On June 10, XPENG announced a structural shake-up. Chairman and CEO He Xiaopeng is taking direct command of the robotics unit, assuming full control of the IRON business to drive mass production and commercial rollout.
The message behind this personnel move is unmistakable. After years of deepening its roots in intelligent vehicles, XPENG has elevated its humanoid robot division to a core priority—accelerating its pivot from a smart EV maker to a physical AI technology company. Industry observers are now watching closely to see if IRON, incubated through its automotive expertise, can deliver a second growth curve.
Taking the Helm: Leveraging Automotive Resources to Empower Robotics
He Xiaopeng's decision to take charge comes at a critical juncture as IRON approaches mass production.
In an internal memo, he likened this moment to the eve of the XPENG G3 launch eight years ago. The implication is clear: the humanoid robot has moved past early R&D and testing, standing just one step away from official delivery.
Reports indicate that over the past year, He Xiaopeng set aside one day each week to track robotics R&D and supply chain alignment. By assuming direct leadership, he aims to break down departmental silos, coordinate group-wide resources, and cut collaboration costs—all in a sprint toward mass production targets.
Leveraging years of automotive experience, XPENG has built a robust supply chain, precision manufacturing processes, rigorous quality control standards, and global operational capabilities. That mature industrial expertise can be directly transferred to the humanoid robot sector, significantly accelerating the path to mass production.
Image Source: XPENG
As early as late May, XPENG held an internal mobilization meeting, rallying nearly 1,000 employees from manufacturing, intelligent algorithms, powertrain development, and vehicle testing. The company has set a target to achieve scaled mass production of IRON by the end of 2026, positioning the product to join the first wave of mass-produced, high-end humanoid robots globally.
The humanoid robot race is crowded, and competition is intensifying. In He Xiaopeng's view, while the strategic direction is clear, operational execution and profitability remain formidable challenges—testing both judgment and the ability to deliver. That reality is a key reason he is stepping in to coordinate the business personally.
XPENG's foray into robotics is not a short-term experiment. After three years of refinement, IRON has completed multiple rounds of technical iteration, achieving capabilities such as autonomous walking, dynamic obstacle avoidance, human-machine interaction, and task execution in specific scenarios.
From self-developed chips, joint structures, and dexterous hands to AI large models shared with its automotive business, XPENG has achieved technological autonomy across the product's core components. By leveraging synergies across automotive, AI, supply chain, and smart manufacturing, the company has carved out a proprietary path for R&D and commercialization distinct from its peers.
Avoiding the Gimmick Route: Anchoring on Practical Physical AI
Most humanoid robots on the market today emphasize extreme athletic ability or stage presence—leaning toward the conceptual. XPENG's IRON, however, has charted a differentiated path focused on lifestyle and commercial applications.
He Xiaopeng stated that XPENG is not building a humanoid robot for display purposes. Instead, it is developing an AI partner capable of close companionship, suited for office and daily life scenarios, and scalable for mass deployment. Compared to industry offerings, IRON's core differentiation is distinct.
First, it is designed for close interaction with a gentle, humanoid aesthetic—shedding the cold, rigid look of industrial robots to fit indoor environments. Second, it uses local AI computing, relying on on-device models for dialogue and decision-making rather than cloud transmission, balancing response speed with data privacy. Third, the robot features a fully enclosed safety structure with lightweight, controlled movements to mitigate collision risks during interaction. Fourth, with full-chain self-development and production—including a dedicated data center and mass-production factory—XPENG can control costs at the source.
Image Source: XPENG
On the hardware intelligence front, IRON is equipped with three Turing AI chips delivering a total computing power of 2,250 TOPS. It runs XPENG's second-generation VLA large model—shared with its vehicles—repurposing the automaker's visual, perception, and decision-making algorithms. This allows the robot to handle dialogue, autonomous movement, and precision tasks, effectively transplanting automotive AI technology into a physical AI terminal.
Commercially, XPENG is adopting a prudent, step-by-step approach. IRON will initially roll out in XPENG stores and industrial parks to handle guided tours, inspections, and reception—controlled scenarios before expanding to retail support roles like cashiering and logistics. The business plan also accounts for overseas compliance. The company aims to achieve mass production of the IRON humanoid robot in the fourth quarter of this year and deliver to global commercial clients by 2027, creating a new lever for XPENG's global expansion.
Doubling down on the long-cycle, capital-intensive humanoid robot race is a calculated move for XPENG at this juncture.
Its automotive business is still in the early stages of profitability climbing. The ongoing price war and heavy R&D spending continue to drain cash flow, while the development and manufacturing of humanoid robots—along with supply chain setup—will further strain the group's financial resources.
Image Source: XPENG
Yet XPENG's advantages cannot be ignored. Unlike pure tech companies crossing into robotics, XPENG possesses closed-loop manufacturing capabilities, mass production control experience, and offline scenario channels—core strengths that internet tech firms lack.
Industry insiders believe automakers are naturally suited for the crossover into humanoid robots. Precision manufacturing, supply chain management, and intelligent algorithm systems can all be transferred to the robotics sector, significantly reducing the cost of trial and error in mass production.
The opportunity is clear, but the challenges of execution remain stark. Across the industry, humanoid robots generally suffer from high costs, limited use cases, and vague profit models. Most products remain in the pilot phase, with a viable profitability model yet to be proven.
For XPENG, He Xiaopeng's direct leadership addresses gaps in decision-making and resource coordination. IRON is more than just a hardware product; it is central to XPENG's strategy to shed its label as a mere automaker and transform into a physical AI technology enterprise.
The next few months of production refinement will determine whether IRON can launch on schedule. And whether XPENG can leverage its automotive strengths to successfully commercialize humanoid robots remains to be seen.
