XPeng's Robot Walks Off the Line — Its Share Price Can Barely Crawl

XPeng's humanoid robot IRON rolls off the line, but shares slide 9% as EV deliveries grow just 4% and losses widen.

The factory floor in Guangzhou told one story on a recent morning: XPeng's humanoid robot IRON strode autonomously off its production line, a machine with 76 degrees of freedom in its body, 21 in each hand, and three Turing AI chips delivering 2,250 TOPS of computing power. The trading floor told quite another. Shares in Hong Kong and the US slid as much as 9 percent and 8.5 percent respectively on the day of the announcement, leaving the stock at €9.42 in Europe — barely 2.3 percent above the 52-week low of €9.21 touched on September 3.

The disconnect between the robotics showcase and the market's response captures the central tension surrounding XPeng right now. Management is pitching a future beyond electric vehicles, but investors are fixated on a present that still isn't profitable.

The Robotics Bet Is Real — and Expensive

The production milestone itself is no small feat. XPeng's humanoid robot assembly line is now officially operational with an automation rate exceeding 80 percent, and the company bills IRON as the world's first advanced general-purpose humanoid robot. Initial deployments are planned for its own stores and corporate campuses, with mass production targeted by the end of 2026 and commercial deliveries to customers in China and abroad slated to begin in 2027.

The capital markets have taken notice of the ambition, even if the equity price hasn't. A funding round in late August raised over $900 million for the robotics division at a valuation north of $6.3 billion, led by IDG Capital with participation from Gaorong Ventures, Tencent, and Alibaba. The strategic logic is straightforward: much of the supply chain overlaps with XPeng's existing EV business, and management has signaled that robotics margins could ultimately exceed those of the car-making operation.

Competitors are moving quickly too. Figure and AGIBOT are already in mass production, while Tesla's Optimus has yet to cross that threshold — a reminder that the humanoid robotics field is condensing fast.

The Core Business Tells a Different Story

Yet for all the futurism, the numbers that matter most to the market remain stubbornly grounded. August deliveries came in at 39,107 vehicles, up just 4 percent year over year — a modest result by the standards XPeng itself has set. To hit its third-quarter guidance of 115,000 to 121,000 vehicles, the company needs September deliveries between 37,866 and 43,866, which would require meaningful acceleration from August's pace.

Second-quarter revenue reached 19.74 billion renminbi, but the net loss of 1.34 billion renminbi widened year over year, weighed down by currency effects and heavy research spending. The company has guided to Q3 revenue of 21.7 to 23.4 billion renminbi, which looks solid on paper but does little to resolve the fundamental friction between growth promises and earnings reality.

There are bright spots worth acknowledging. The G9L model made its official debut on August 11 with pre-sales opening in mainland China, and a Philippine market entry with the X9 and L03 is slated for September — though notably without an accompanying corporate announcement. The VLA 2.0 intelligent driving system is reportedly due for a major update in the coming weeks, again without primary confirmation from the company. International expansion and technological development are clearly in motion, but much of it remains announcement rather than verified fact.

A Market That's Stopped Giving Credit

The share price chart tells the story of a trust deficit that has built over months. The stock sits roughly 62 percent below its 52-week high of €24.40, reached only last November, and has shed 48 percent since the start of the year. It trails its 200-day moving average by 33 percent. Spectacular robotics announcements have so far failed to reverse any of that.

The explanation isn't hard to find. Companies building humanoid robots aren't selling them in meaningful volumes yet — the revenue lies in the future while the risk sits in the present. Investors looking at 2026 EV sales figures and margin pressure aren't inclined to reward a factory-floor demonstration with a higher share price, no matter how impressive the choreography.

The central question isn't whether IRON is impressive. It clearly is. The question is whether the market is willing to reprice an automaker as a robotics play before the first commercial deliveries in 2027 have actually happened. So far, the answer has been a consistent no. Between a robot learning to walk and a stock that keeps falling lies the gap where XPeng must prove that a striking production-line spectacle can become a sustainable business model — and until the core EV operation scales profitably, that gap is likely to persist.