XPeng Targets Production of More Than 1,000 IRON Humanoid Robots per Month

GUANGZHOU — When XPeng walked its next-generation Iron humanoid robot down a runway late last year, the machine’s movement was fluid enough to spark immediate skepticism. Rumors circulated online that the automaker had hidden a human actor inside the chassis, forcing CEO He Xiaopeng to publicly display schematics of the internal framework to prove the hardware was real.

Months later, that theatrical presentation is hardening into a highly capitalized manufacturing reality.

According to internal timelines, the Chinese electric vehicle manufacturer is scaling up its production capacity to exceed 1,000 Iron robots per month by the end of 2026. The aggressive expansion at its new 110,000-square-meter facility in Guangzhou marks a deliberate strategic shift. XPeng is attempting to pivot from a traditional smart-EV builder into a diversified physical AI enterprise.

Rewiring the Automotive Supply Chain

The logic behind an automaker entering the robotics space rests on technological overlap. The core systems required to operate an autonomous vehicle—spatial mapping, neural networks, real-time edge computing, and complex sensor suites—are fundamentally the same systems required to operate a bionic machine.

XPeng is betting that its established automotive supply chain and manufacturing scale will allow it to outpace pure-play robotics startups that lack factory infrastructure. The Iron humanoid functions as a showcase for the company’s in-house engineering, featuring several components adapted directly from its vehicle programs.

The Hardware Blueprint

• Processing Power: The robot is equipped with three proprietary Turing AI chips, yielding a combined computing power of 2,250 TOPS.
• Neural Architecture: The hardware runs on XPeng’s second-generation Vision-Language-Action (VLA) large model, the same framework driving its autonomous vehicle updates.
• Physical Kinematics: Standing 1.78 meters tall and weighing 70 kilograms, Iron features more than 60 joints and 22 degrees of freedom in its hands.
• Energy Density: Power is supplied by an all-solid-state battery pack, a technology automakers have raced to commercialize for vehicles but found easier to package first in a humanoid frame.

Showrooms Before Factory Floors

Unlike industrial competitors that design robots strictly for heavy manufacturing or logistics, XPeng is taking a consumer-facing path. The company plans to deploy the first production units of Iron as retail sales assistants across its Chinese showrooms in the first quarter of 2027, followed by an international store rollout later that year.

This aggressive commercial timeline places XPeng in direct competition with Tesla’s Optimus program and emerging agile humanoids like the Unitree G1. While Tesla has kept its specific Optimus production metrics and delivery channels closely guarded, global robotics indices are already tracking how these platforms stack up in terms of commercial readiness. Industry analysts point to comprehensive resources like the Unitree G1 vs Tesla Optimus analysis on Anton Robots as evidence that the commercial battleground is moving from conceptual tech specs to real-world pricing and availability.

“Progress in physical AI will likely reach commercial maturity in robotics before it achieves full autonomy in cars,” He Xiaopeng told employees at a recent internal meeting. The executive reasoned that static retail environments present far lower safety risks than public roads, where a minor software anomaly can result in a high-speed collision.

The Macro Race for Embodied AI

The commercialization of humanoid robotics in China is moving past the pilot phase. Shenzhen-based Ubtech has already deployed its Walker S1 humanoids into Zeekr’s EV assembly plants to handle materials, while domestic component suppliers are expanding production lines to meet rising international research orders. The broader market momentum is supported by significant capital deployment, with XPeng raising its AI-related research and development budget to 7 billion yuan ($1.03 billion) this year.

With the Guangzhou factory floor now transitioning into active assembly, the metrics for AI dominance are shifting. The technology is no longer confined to server racks and cloud computing; it is being measured by the monthly output of assembly lines. For commercial buyers and tech enterprises, navigating this rapidly expanding hardware landscape requires structured data.