After Plunging 40% This Year, XPeng Bets on Physical AI to Help Turn the Tide

Things haven’t been pleasant for investors in Chinese electric vehicle (EV) companies this year. Among the emerging new energy vehicle (NEV) companies, XPeng's (XPEV) stock is down over 41% for the year, which is far worse than rivals Nio (NIO) and Li Auto (LI). In my previous article, I noted that XPEV was attractive near its 52-week lows. It has, however, continued its downward trajectory and is now trading at new 52-week lows. Let's examine whether XPEV can recover in 2026 or if it’s time to give up on this underperforming stock.

Chinese Auto Market Is Witnessing Degrowth

The Chinese auto market is experiencing degrowth, with sales falling for 10 consecutive months. In the first seven months of this year, domestic passenger sales in the world’s biggest auto market shrunk by over a fifth. Sales of internal combustion engine (ICE) cars have been particularly weak. While new energy vehicle (NEV) penetration in the country rose to almost 65% in July, a record high, sales have fallen in all seven months this year.

Specifically, XPeng's deliveries in the first seven months of the year fell by nearly 13% compared to the corresponding period last year. The company’s monthly deliveries peaked above 42,000 vehicles in October 2025 and have since failed to breach that high bar. Its deliveries have failed to take off materially, and the average monthly deliveries in the first seven months of the year are below 30,000. Nio has fared much better and sold more cars than XPEV in the first seven months. Unsurprisingly, its stock has performed better even though it too is in the red.

XPeng Pivots to Physical AI

Meanwhile, XPeng is pivoting to physical artificial intelligence (AI), which is arguably the buzzword these days, and changed its name from XPeng Motors to XPeng Group earlier this year. We have similar stories in the U.S., where Lucid Motors and Tesla Motors previously dropped “Motors” from their names and are known as Lucid Group (LCID) and Tesla (TSLA), respectively. The name changes were meant to reflect their ambitions beyond the automotive industry. Remember, even Facebook renamed itself Meta Platforms (META). It’s a different thing altogether that the metaverse, which prompted the name change, has turned out to be a vanity project.

Meanwhile, XPeng has been making progress in AI and expects to showcase the next version of its Iron humanoid in Q3, which its CEO, He Xiaopeng, said during the Q1 2026 call will feature “multilingual communication, human-like full-body motion, and gradually autonomous execution of professional tasks.” It is targeting mass production of Iron by the end of this year and would initially deploy them in its showrooms.

XPeng’s autonomous driving system is considered the most advanced in China, and the company is testing its VLA 2.0 in Europe. It expects to get regulatory approvals for the system in several countries next year, which would help it bring the technology to more countries. The company plans to commence robotaxi operations in Guangzhou this quarter, which it then plans to roll out in more Chinese cities, followed by global expansion.

That said, the company needs to execute on the ambitious plans, which I have noted previously are quite “Tesla-like.” While XPeng has had its moments and showed good traction in deliveries last year, piggybacking on success stories like the Mona M03, it hasn’t been able to show sustainable growth. The management has its task cut out and has to execute on the physical AI initiatives while growing its EV business at the same time. While it did post a profit in Q4 2025, it needs to become a sustainably profitable business sooner rather than later. Delivering on all these aspects simultaneously won’t be easy for XPeng, if not outright impossible. It does not help that the Chinese EV industry continues to be plagued by overcapacity, tepid demand, and the resultant price war.

Since XPeng is not posting profits, we don’t have its price-to-earnings multiple, the most widely used valuation metric. However, the stock trades at a forward price-to-sales multiple of 0.71x, which looks quite reasonable. Overall, XPEV stock remains a high-risk bet, but even if the company can come anywhere near the ambitious growth roadmap it has outlined with products like robotaxis and humanoids, the stock could see stellar gains over the next couple of years.