Xpeng Raises $900M for Robotics Unit to Take On Tesla's Embodied AI with Alibaba Backing

The Quick Take

• Xpeng's robotics unit raised US$900 million from Alibaba and IDG Capital to challenge Tesla's embodied AI push
• Xpeng is pivoting from EV-only to an embodied AI strategy that combines humanoid robots with its vehicle platform
• The $900M raise comes despite Xpeng reporting a widening net loss, reflecting investor confidence in the AI thesis

Chinese EV maker Xpeng has secured US$900 million in new funding for its robotics subsidiary, with investors including Alibaba Group and IDG Capital backing the company's ambition to challenge Tesla's embodied artificial intelligence strategy. Xpeng's pivot positions its vehicle AI software stack and autonomous driving technology as a platform for humanoid robotics, mirroring Tesla's approach of leveraging vehicle-trained neural networks for physical AI applications. The fundraise is one of the largest disclosed rounds in Chinese robotics this year.

For the broader EV and AI hardware sector, Xpeng's success in raising capital despite a widening net loss signals that investors are willing to accept near-term losses in exchange for optionality on the embodied AI market, which Morgan Stanley and Goldman Sachs have separately estimated could grow to trillions of dollars in addressable value over the next decade. Alibaba's participation is strategically significant: Alibaba's cloud infrastructure and AI model capabilities could provide Xpeng's robots with the compute backbone they need for real-world task execution.

The key technical and commercial milestone to watch is whether Xpeng can demonstrate a commercially viable humanoid robot SKU within 18-24 months, consistent with its stated timeline. The competitive landscape includes BYD, DJI, and a wave of well-funded startups like Unitree Robotics and Agibot. The macro variable that determines Xpeng's robotics trajectory is China's government manufacturing-automation subsidy policy: additional subsidies for factory humanoids would dramatically shorten the path to positive-unit economics for the division.

Synthesized from 1 source.