XPeng's Robotaxi Pilot Meets Its Paris Reckoning as the Stock Hunts for a Bottom

XPeng opens robotaxi rides in Guangzhou and preps an October 12 Paris debut for the G9L, with JPMorgan cutting its target as shares sit near a 52-week low.

XPeng has picked an awkward moment to ask investors for patience on two fronts at once. With its shares down roughly 53% year-to-date and changing hands near EUR 8.56 — barely 4.9% above the 52-week low of EUR 8.16 touched on 2 October — the Chinese automaker is simultaneously opening a robotaxi pilot to outside riders in Guangzhou and preparing a pivotal European debut in Paris.

The two initiatives target the same problem from opposite ends: a core business selling electric cars into a brutal price war, and a valuation that needs a story with better margins.

Guangzhou Pilot Shifts From Lab to Street

XPeng has begun accepting bookings for autonomous rides in designated districts of Guangzhou through a mini-program, though only passengers holding an invitation code can request a trip. The company has branded the unit XPENG YOYO, formally ending a phase of pure technology validation and starting live user testing.

The groundwork was laid quickly. A first series-production vehicle based on the GX model rolled off the line in May, followed by more than 2,000 internal test runs over the summer. Since August, XPeng has held approval for remote testing without a safety driver on specified Guangzhou roads. Opening the service to outsiders pushes the company onto the operational proving ground where fleet scalability under real conditions gets tested.

Management has a number in mind for when the economics might work. CEO He Xiaopeng targets break-even per vehicle in Guangzhou no earlier than the second half of 2027. Whether that holds depends heavily on local utilization and vehicle density. In smaller cities, fleets of just a few hundred units could be enough to clear the threshold, according to the company, while major metros demand a far larger footprint.

XPeng is not positioning itself as a fleet operator. The model casts the company as a pure technology supplier, earning from hardware sales, technical services and recurring commission on rides, while partners handle local fleet management. That shifts the capital burden of running vehicles onto outside providers — but it also means partner efficiency sets the pace at which scale economies arrive.

A Hardware Bet That Skips LiDAR

The vehicle relies on four in-house Turing chips delivering 3,000 TOPS of compute, paired with a second-generation AI model. Crucially, the system does without expensive LiDAR sensors and high-precision HD maps, which keeps per-unit production costs well below conventional Level 4 concepts from rivals.

If that cost advantage holds, XPeng plans to launch a dedicated robotaxi successor model in 2027 and push into additional cities and international markets. A high-margin software and licensing stream would sharply reduce reliance on cutthroat passenger car sales — and for a stock trading just above its yearly low, timely monetization of the platform would underpin a genuine fundamental re-rating.

The downside case is equally clear. Dropping LiDAR could become an operational hurdle when seeking regulatory clearance for fully driverless operation, since a camera-only approach demands a seamless data and remote-monitoring network to handle complex traffic. Sixfold redundancy across steering, braking and power supply does allow switchover within 100 milliseconds, yet any incident during testing could prompt city authorities to slow the expansion of test zones.

Concentration in Guangzhou adds a timing risk. Other key markets such as Beijing and Shanghai have not yet been cleared for the service. If regulatory approval for passenger operations without a safety driver slips beyond 2027, development costs keep weighing on margins without the targeted commission revenue arriving — and XPeng loses valuable ground to competitors while its auto business remains under pressure.

Paris Sets the Next Hard Deadline

While Guangzhou tests the robotaxi thesis, Europe tests the core business. On 12 October, XPeng plans the global premiere of its G9L SUV at the Paris Motor Show, alongside a showcase of its physical AI technology stack. European pricing for the G9L will be announced that same day, with order books opening.

The operational backdrop is solid. XPeng delivered 118,390 vehicles in the third quarter of 2026, up 15% from the prior quarter, with September volume alone reaching 41,256 units. In China, cheaper models carry the load — the MONA L03 topped 10,000 monthly deliveries in September — but volume models tend to dilute margins, making a profitable European price point essential once transport costs and tariffs are factored in.

Local production is meant to help. Initial G9L production runs have been completed at Magna's Graz facility in Austria, where the G6, G9 and P7+ have already been built. An established European manufacturing base cuts logistical friction and softens the impact of trade barriers.

JPMorgan Steps to the Sidelines

Not everyone is convinced the timing works. On 29 September, JPMorgan analyst Nick Lai downgraded the stock from Overweight to Neutral and cut his price target to $11.50 from $24, citing persistent weakness in China's auto sector through the second half of 2026 and structural challenges stretching into 2027. Soft domestic demand, rising procurement costs and policy uncertainty topped his list of concerns, with trade barriers further complicating overseas expansion.

That leaves investors with a clearly defined risk profile. As long as the recent low holds and delivery momentum persists, there is room for a recovery. Should Chinese demand weaken further or the European launch fall flat, fresh selling pressure looks likely.

The 12 October showcase will settle part of the question, deciding whether international expansion can stabilize the stock. The parallel milestone sits in Guangzhou, where the shift to regular passenger service without a safety driver — targeted for 2027 — and the dedicated robotaxi model planned for the same period will determine how fast the partner ecosystem scales beyond its current pilot zone. Until hard data on fleet economics emerges, the shares stay tethered to execution on both fronts.