XPENG IRON
Last reviewed
Apr 26, 2026
Sources
28 citations
Review status
Source-backed
Revision
v4 · 3,842 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Apr 26, 2026
Sources
28 citations
Review status
Source-backed
Revision
v4 · 3,842 words
Add missing citations, update stale details, or suggest a clearer explanation.
| XPENG IRON | |
|---|---|
 | |
| General information | |
| Manufacturer | XPENG Robotics (Pengxing Intelligence) |
| Country of origin | China |
| Year introduced | 2024 (first generation), 2025 (next generation) |
| Status | Pre-production; mass production targeted end of 2026 |
| Estimated price | ~$150,000 USD |
| Website | xpeng.com |
XPENG IRON is a humanoid robot developed by XPENG Robotics, the robotics division of Chinese electric vehicle manufacturer XPeng Inc. (NYSE: XPEV). First unveiled at XPENG's AI Day on November 6, 2024, Iron is the company's flagship humanoid platform, designed to bridge the gap between autonomous vehicle technology and general-purpose bipedal robotics. Standing 178 cm tall and weighing 70 kg, Iron is distinguished by its highly anthropomorphic appearance, 82 degrees of freedom, industry-first all-solid-state battery, and 2,250 TOPS of computing power delivered by three proprietary Turing AI chips. XPENG has described Iron as the culmination of a strategy to transfer roughly 70% of its EV technology stack, including autonomous driving AI, sensor systems, and electrical architecture, into a humanoid form factor.
XPENG's robotics ambitions date to early 2020, when the company began exploring embodied intelligence under the leadership of Liangchuan Mi. That same year, XPENG acquired Dogotix, a quadruped robot startup founded by Zhao Tongyang, in a deal reportedly worth approximately $100 million. Xiaomi founder Lei Jun is said to have helped facilitate the acquisition. The purchase led to the creation of Pengxing Intelligence (also rendered as XPENG Robotics), a subsidiary headquartered in Shenzhen, Guangdong, dedicated to developing intelligent bionic robots.[1][2]
The Pengxing team grew to over 300 employees at its peak, though internal restructuring and leadership changes caused the headcount to drop to around 70 before scaling back up to approximately 200 members. After Zhao Tongyang departed to found EngineAI, He Xiaopeng appointed Mi to lead the robotics division more directly, aligning the effort with XPENG's broader "Physical AI" strategy.[2]
XPENG publicly debuted its first humanoid robot, the PX5, at its annual Tech Day event on October 24, 2023. Standing 150 cm tall and weighing 109 kg, the PX5 featured 22 degrees of freedom, ultra-lightweight 430-gram hands with 11 DOF each, and 0.05 mm positioning accuracy on its robotic arms. The prototype demonstrated stable bipedal walking over grass, gravel, and uneven terrain for more than two hours on a single charge. It could also play football and ride Xiaomi's self-balancing scooter, showcasing mature balance control algorithms derived from XPENG's autonomous driving research.[3][4]
Though the PX5 served primarily as a proof of concept, it validated XPENG's thesis that technologies developed for self-driving EVs (perception algorithms, sensor fusion, control systems) could transfer effectively to humanoid platforms. XPENG internally iterated through several generations of robots between the PX5 and Iron, with Iron representing the fourth generation of the company's humanoid line.[2]
XPENG unveiled the first-generation Iron at its AI Day event on November 6, 2024, in Guangzhou. The robot stood 178 cm tall and weighed 70 kg. It featured more than 60 joints and was powered by the company's proprietary Turing AI chip, a 40-core processor capable of supporting AI models with up to 30 billion parameters. Technologies used in Iron were described as "homologous to AI cars," with the end-to-end large language model architecture enabling capabilities such as walking, grasping, holding, and placing objects.[5][6]
Following the 2024 unveiling, XPENG deployed early Iron units to its Guangzhou factory, where the robots began training on the production line for the XPENG P7+ electric sedan. In this setting, Iron handled sorting, transport, and basic assembly operations, reportedly installing up to 500 screws per day and saving over 300,000 yuan in annual labor costs.[7]
The next-generation Iron was revealed at XPENG's 2025 AI Day, themed "Emergence," held on November 5, 2025, at the XPENG Science Park in Guangzhou. This upgraded model featured significant advances in mechanical design, AI architecture, and battery technology. He Xiaopeng described the company's strategic positioning as an upgrade to "a mobility explorer in the physical AI world and a global embodied intelligence company."[8][9]
The next-gen Iron's movements proved so lifelike that online audiences expressed skepticism about whether a human was hidden inside. In response, He Xiaopeng returned to the stage the following day and had staff physically cut open Iron's leg to reveal its internal machinery, conclusively demonstrating that the robot was entirely mechanical.[10][11]
Iron stands 178 cm (5 feet 10 inches) tall and weighs 70 kg (154 pounds), dimensions chosen to closely approximate an average adult human. The robot's design philosophy centers on what XPENG calls "extreme anthropomorphism," an approach that prioritizes human resemblance over raw capability. Rather than building the strongest or fastest humanoid, XPENG has deliberately shaped Iron to soften the emotional distance between people and machines.[9][12]
The robot features a skeleton-muscle-skin layered structure. An internal endoskeleton provides structural support, while bionic muscles enable natural-looking motion across the torso and limbs. A humanoid spine allows for natural bending and posture changes, enabling the robot to stand, sit, lie down, and walk with fluid movements. Full-coverage flexible synthetic skin embedded with touch sensors covers the entire body, providing tactile feedback for interaction with people and objects. A 3D curved OLED display serves as the robot's "face," capable of showing expressions for social interaction.[8][9]
In a departure from the neutral, genderless designs common among competitors, XPENG has made Iron highly customizable. Customers can adjust the robot's height, physique (from slim to stocky), and even select male or female presentation. He Xiaopeng has compared this to how automobile buyers choose between vehicle models, describing the vision as making the robot "warmer and also more intimate" for eventual household deployment.[9][12]
The next-generation Iron features 82 degrees of freedom across its entire body, placing it among the most articulated humanoid platforms announced to date. The robot includes 62 active joints enabling natural movement patterns.[8][13]
Each hand contains 22 degrees of freedom, supported by what XPENG describes as the industry's smallest harmonic joints. These miniaturized actuators achieve a 1:1 ratio with human hand sizing while maintaining the torque and precision needed for dexterous manipulation. The hands can perform delicate tasks such as holding small objects and operating tools, as well as robust grasping of larger, heavier items. He Xiaopeng noted that the team "enabled Iron's light and gentle stride by adding passive degrees of freedom at the toes," contributing to the robot's natural gait.[8][14]
Iron walks at a maximum speed of approximately 6 km/h (about 1.67 m/s). The robot's foot joints are designed to absorb shock on hard surfaces, maintaining balance during locomotion. The combination of the humanoid spine, bionic muscles, and passive toe joints produces a walking gait that attracted widespread attention for its natural, human-like quality.[13][15]
Iron's intelligence is built on a tri-model AI architecture that combines three interconnected large models, each adapted from XPENG's autonomous driving technology stack:
| Model | Full name | Role |
|---|---|---|
| VLT | Vision-Language-Task | Core decision-making engine; enables autonomous reasoning and task planning in the physical world |
| VLA | Vision-Language-Action | Action generation model; translates sensory input into motor commands across all 82 joints |
| VLM | Vision-Language-Model | Multimodal perception and natural language interaction |
The VLT (Vision-Language-Task) model, announced for the first time at the 2025 AI Day, serves as Iron's "brain." It functions as the core engine for autonomous decision-making, enabling the robot to perform in-depth reasoning and plan complex multi-step tasks in real-world environments. The VLA (Vision-Language-Action) model, now in its second generation, was adapted directly from XPENG's autonomous driving division and handles the translation of perceived visual information into physical movements. The VLM (Vision-Language-Model) provides multimodal understanding and conversational capabilities.[8][9][16]
Working together, the three models enable Iron to simultaneously converse, walk, and interact with its environment. XPENG refers to this overall framework as its first-generation Physical-World Large Model, a deliberate convergence strategy in which "Physical AI" serves as the umbrella concept connecting self-driving cars and bipedal robots.[9][16]
Iron is powered by three of XPENG's self-developed Turing AI chips. Each chip features a 40-core processor and delivers 750 TOPS of effective computing power. Combined, the three chips provide 2,250 TOPS of total processing capability. XPENG claims this figure is three times higher than the second-best humanoid robot on the market and ten times higher than mainstream benchmarks in the industry.[8][9]
The Turing chip can support AI models with up to 30 billion parameters, enabling on-device inference without reliance on cloud computing for core decision-making. The same Turing chip architecture is used across XPENG's product ecosystem; the company's Robotaxi vehicles, for example, use four Turing chips for up to 3,000 TOPS of computing power.[6][17]
Iron runs XPENG's Tianji AIOS (AI Operating System), a real-time software platform designed for full-body coordination, balance control, task planning, and environmental awareness. Tianji AIOS manages the integration of perception, control, and voice interaction systems across the robot's sensor suite and actuator network.[18]
One of Iron's most significant technical distinctions is its use of an all-solid-state battery, which XPENG describes as the first such application in any humanoid robot. Traditional lithium-ion batteries rely on flammable liquid electrolytes, which pose safety risks in environments where robots operate near people. Solid-state batteries replace these liquids with ceramic or polymer electrolytes, fundamentally improving safety while also offering higher energy density and lighter weight.[8][13][19]
XPENG has stated that Iron's solid-state battery achieves an energy density exceeding 500 Wh/kg, roughly double the capacity of the battery used in Tesla Optimus within the same physical footprint. This high energy density has allowed XPENG to miniaturize other components and equip Iron with its humanoid spine and bionic muscle systems without excessive weight penalties. Major battery manufacturers such as CATL have noted that solid-state batteries are particularly well-suited for applications where energy density is paramount, such as drones and robots, rather than electric cars.[19]
The all-solid-state design also supports indoor AEB (Automatic Emergency Braking) functionality, adding another safety layer for deployment in enclosed public spaces and, eventually, homes.[19]
Iron's perception system, branded as "Eagle-Eye," provides a 720-degree field of view, encompassing full 360-degree horizontal and 360-degree vertical coverage. This system was adapted from XPENG's autonomous driving sensor suite, providing the robot with comprehensive spatial awareness comparable to the perception capabilities of a self-driving car.[15][20]
The vision system uses a combination of RGB cameras, 3D depth cameras, and LiDAR to build real-time three-dimensional maps of the robot's surroundings through Simultaneous Localization and Mapping (SLAM) algorithms. The system can identify objects, gauge distances, and, according to XPENG, even read micro-expressions on human faces for empathetic social interactions.[15][20]
Beyond the Eagle-Eye vision system, Iron integrates force/torque sensors, an inertial measurement unit (IMU), ultrasonic sensors, and the full-body touch sensors embedded in its synthetic skin. This multi-modal sensor array provides the data inputs required by the VLT + VLA + VLM AI stack to make real-time decisions about navigation, manipulation, and social interaction.[18]
XPENG has established its first embodied intelligence data factory in Guangzhou to address one of the humanoid robotics industry's most persistent challenges: the lack of high-quality training data. This facility is dedicated to collecting behavioral data from robot-environment interactions, enabling XPENG's AI models to continuously improve through extensive physical training sessions.[8]
In January 2026, XPENG announced the completion of its first "ET1" prototype robot, developed to automotive manufacturing standards. The ET1 milestone demonstrated the successful integration of XPENG's vehicle engineering expertise with its robotics capabilities, marking a key step toward mass production readiness. CEO He Xiaopeng personally announced the achievement.[21]
In February 2026, XPENG announced plans to break ground on a dedicated humanoid robot mass production facility at the Guangtang Sci-Tech Innovation City Embodied Intelligence Industrial Park in Guangzhou's Tianhe District. The 110,000-square-meter facility will function as the industry's first "full-chain" manufacturing hub, covering processes from R&D validation and small-batch trial production to large-scale manufacturing. Phase one construction includes robot factory buildings, power stations, and supporting infrastructure.[22][23]
XPENG has outlined an ambitious investment plan of approximately $13.8 billion over the next two decades to drive humanoid robotics development and scale Iron's production to meet anticipated market demand.[24]
Mass production preparation is scheduled to begin in April 2026, with the goal of achieving large-scale manufacturing by the end of 2026. If XPENG meets this timeline, Iron could become one of the first mass-produced next-generation humanoid robots, potentially reaching the market ahead of competitors such as Tesla's Optimus.[8][19][22]
Iron has already been deployed at XPENG's Guangzhou factory, where it assists with vehicle assembly tasks on the P7+ production line. Approximately hundreds of humanoid robots have been deployed on XPENG's manufacturing lines, handling sorting, transport, and screw installation operations. The company reports that a single Iron unit can install up to 500 screws per day, translating to annual labor cost savings exceeding 300,000 yuan per robot.[7]
XPENG's initial commercialization roadmap prioritizes controlled commercial environments rather than domestic or heavy industrial settings. The first deployment wave will position Iron units in people-facing roles:
| Application | Description |
|---|---|
| Tour guide | Leading visitors through facilities and exhibitions |
| Shopping guide | Assisting retail customers with product information |
| Reception | Greeting and directing guests at offices and hotels |
| Sales assistant | Supporting product demonstrations in XPENG showrooms |
| Security patrol | Monitoring facilities and performing inspection rounds |
XPENG plans to deploy Iron in its own retail showrooms first, where the robots will greet customers and assist with vehicle presentations.[8][9]
On January 31, 2026, Iron made its first public appearance at Shenzhen Bay MixC shopping center. After performing a smooth catwalk to the center of the audience, the robot lost its balance while standing still and fell backward. A staff member partially broke its fall. The incident went viral on Chinese social media, dominating online discussion. CEO He Xiaopeng responded on Weibo, saying: "It reminds me of how all toddlers learn to walk. After a fall, they will stand firm; the next step is to begin running, and to keep running." The following day, Iron returned to the event using a support frame while engaging audiences in educational exchanges and trivia.[25][26]
XPENG has announced a strategic partnership with Baoshan Iron & Steel (Baosteel), one of China's largest steel producers. Baosteel will serve as Iron's first industrial ecosystem partner, deploying humanoid robots at its facilities for real-world trials in inspection and other complex operational scenarios.[8]
To foster broader adoption and application development, XPENG has announced plans to release Iron's software development kit (SDK) to global developers. This open-ecosystem approach is intended to accelerate the creation of third-party applications and services for the Iron platform.[8]
A defining characteristic of XPENG's robotics strategy is the extensive reuse of technology originally developed for its electric vehicles. He Xiaopeng has stated that approximately 70% of Iron's technology stack derives from the company's EV platform. Shared components and systems include:[2][27]
| Technology area | EV origin | Robot application |
|---|---|---|
| AI chip | Turing chip for ADAS/autonomous driving | Core computing for VLT + VLA + VLM models |
| Vision system | Hawkeye/Eagle-Eye for XNGP driving | 720-degree spatial perception |
| VLA model | End-to-end autonomous driving model | Action generation from visual input |
| E/E architecture | Vehicle electrical/electronic systems | Robot control and communication backbone |
| Cloud AI | Shared cloud infrastructure for fleet learning | Robot training data processing |
| Control systems | Vehicle dynamics and stability control | Bipedal balance and locomotion |
This convergence strategy allows XPENG to amortize R&D costs across its vehicle and robotics businesses while accelerating the development of both product lines. He Xiaopeng has argued that "humanoid robotics and autonomous driving can accelerate one another when built on shared infrastructure," positioning robotics as the company's "third growth curve" alongside EVs and flying cars.[2][27]
Iron enters a rapidly growing market for humanoid robots, particularly in China, where government policy has targeted humanoid robotics as a key future industry. Beijing has announced plans to build a world-class humanoid robot industry by 2027.[28]
| Company | Robot | Notable features |
|---|---|---|
| Tesla | Optimus | Iterative development; factory deployment; global brand |
| Unitree Robotics | H1, H2, G1 | Low entry price (~$16,000 for G1); strong agility demos |
| UBTECH Robotics | Walker S | Early mover in humanoid service robots |
| Agibot | A2 | Backed by Shanghai AI Lab; modular design |
| EngineAI | SE01 | Founded by former XPENG Robotics team member |
| Figure | Figure 02 | BMW factory deployment; OpenAI partnership |
XPENG's differentiation rests on three pillars: the anthropomorphic design philosophy (prioritizing human likeness over industrial efficiency), the all-solid-state battery (a safety and energy density advantage), and the deep integration with an established automotive manufacturing and AI ecosystem. Where Tesla's Optimus has emphasized functionality and iterative hardware development, XPENG has pursued biomimicry so precise that audiences initially could not tell Iron apart from a human performer.[9][12][28]
| Category | Specification | Value |
|---|---|---|
| Physical | Height | 178 cm (5 ft 10 in) |
| Physical | Weight | 70 kg (154 lb) |
| Mobility | Total degrees of freedom | 82 |
| Mobility | Active joints | 62 |
| Mobility | DOF per hand | 22 |
| Mobility | Max walking speed | ~6 km/h (1.67 m/s) |
| Manipulation | Arm payload | 5 kg per arm |
| Manipulation | Max deadlift | 15 kg |
| Manipulation | Hand type | Dexterous, 5 fingers per hand |
| Manipulation | Harmonic joints | Industry's smallest (1:1 human hand scale) |
| Power | Battery type | All-solid-state |
| Power | Energy density | >500 Wh/kg |
| Computing | AI chips | 3x XPENG Turing (40-core each) |
| Computing | Total compute | 2,250 TOPS |
| Computing | Max model parameters | 30 billion |
| Computing | Operating system | Tianji AIOS |
| AI | Core AI models | VLT + VLA 2.0 + VLM |
| Sensors | Vision system | Eagle-Eye 720-degree (360 H + 360 V) |
| Sensors | Camera types | RGB cameras, 3D depth cameras |
| Sensors | LiDAR | Yes |
| Sensors | Additional sensors | Force/torque, IMU, ultrasonic, full-body touch |
| Design | Skin | Full-coverage flexible synthetic with touch sensors |
| Design | Display | 3D curved OLED (head-mounted) |
| Design | Spine | Humanoid biomimetic spine |
| Design | Customization | Adjustable height, physique, male/female presentation |
| Connectivity | Interfaces | WiFi, 5G |