XPENG Raises Over $900M to Industrialize IRON

XPENG finances its humanoid platform IRON with over $900 million, achieving a valuation of $6.3 billion. The company aims for mass production by the end of 2026, followed by deliveries in 2027, with dedicated resources for R&D, data, and industrial capabilities. This funding round, led by IDG Capital with participation from Tencent and Alibaba, comes as XPENG's stock has seen a significant decline over the past year.

Mass Production by End of 2026 and Initial Deliveries in 2027

XPENG plans to industrialize IRON by the end of 2026, with an initial rollout in its own stores and campuses before expanding further. Deliveries to customers in China and abroad are expected in 2027. To support this ramp-up, funds are allocated for the construction of end-to-end mass production facilities. Part of the capital will also support commercial expansion outside of China, with XPENG targeting production and delivery volumes on an automotive scale.

IRON Platform: 2,250 TOPS Embedded and 76+21 Degrees of Freedom

The IRON robot features 76 degrees of freedom for the body and 21 for each hand, within a fully enclosed flexible lattice structure designed in-house, which XPENG describes as ensuring both mobility and dexterity. The hardware platform, designed by XPENG, includes chips, motion controllers, distinct modules, and specific devices for the hands that allow for precise manipulation. In terms of computing, 2,250 TOPS provided by three internally designed Turing AI chips enable the execution of the physical foundation AI model on the device, with low inference latency and local data processing. IRON is central to the physical AI strategy, and its manufacturing adheres to automotive quality standards and processes.

Funding Round: IDG Leads, Tencent and Alibaba Support

The funding round is led by IDG Capital, with participation from Gaorong Ventures and the addition of Tencent and Alibaba as strategic investors. The physical AI unit has secured over $900 million, achieving a valuation of $6.3 billion through stock purchase agreements. Following the closure, XPENG plans to retain control of the business and maintain the consolidation of the unit in its accounts. XPENG presents this fundraising as a record in physical AI in China, both in terms of amount and valuation. The funds are intended to support long-term investments in the comprehensive development of physical AI, enhance team incentives, finance software and hardware R&D, model training and iteration, high-quality data generation, industrial facility construction, and commercial expansion outside of China.

Market Context: -7.22% on the Session and -51.24% Year-on-Year

At the time of the announcement, XPENG's stock was down 7.22%, bringing the decline to 51.24% over twelve months. The group's automotive business is facing intensified competition in China from local manufacturers and Tesla in both domestic and foreign markets. In this context, investors still value the robotics unit at $6.3 billion, even as the parent company's stock continues to trend downward.

What Investors Are Betting On and XPENG's Stated Ambition

IDG Capital believes that physical AI is entering a phase of large-scale manufacturing and commercial deployment, which would favor XPENG, praising its combination of edge processors, foundation models, and complete robotic systems, as well as synergies with automotive and autonomous driving. Gaorong Ventures sees humanoids moving beyond demonstrations of mobility and dexterity towards reliable production and tangible value creation, highlighting XPENG's automotive-level standards and over ten years of industrial experience. He Xiaopeng recalls twelve years of internal R&D, a physical world foundation model, Turing AI chips, and AI infrastructure, asserting that a new phase towards production and deployment has been launched. He believes that support from leading investors should accelerate growth and talent attraction, with the ambition of creating an IRON with a very human design, built to high standards, and envisioned as a daily partner. XPENG also argues that humanoid hardware suited for human environments facilitates the collection of behavioral data and, ultimately, a data-models-applications cycle intended to accelerate learning and the acquisition of new tasks.