XPeng heads into its third-quarter report with a story that has rarely looked bigger and a stock that has rarely looked less convinced. Last quarter, management layered a physical-AI pitch, anchored by a record humanoid robotics financing, on top of a sharp rebound in car deliveries. Yet the shares have dropped 13.4% since that report, roughly 9.6 percentage points worse than the S&P 500. That gap suggests the market cares less about robots and robotaxis than about one plainer question: can XPeng turn rising volume back into profit?

The revenue bar is unusually clear. Management guided third-quarter revenue to RMB21.7 billion to RMB23.4 billion, or about $3.23 billion to $3.48 billion, on deliveries of 115,000 to 121,000 vehicles. The prevailing consensus of roughly $3.85 billion sits well above even the top of that range. That makes the guidance look like a reset rather than a sandbag, and it means the company could deliver exactly what it promised and still technically fall short of the Street. On a year-over-year basis, the guided range implies revenue growth of roughly 13% to 22% from the $2.86 billion posted a year ago, and sequential growth of about 11% to 20% from the second quarter's $2.91 billion. That is solid progress, but not the kind of acceleration that justifies a number near $3.85 billion. A usable per-share consensus is not available, so the profit line will be judged against recent history: a loss of $0.21 per share last quarter versus a much narrower $0.06 loss a year earlier.

That widening loss is the heart of the matter. Vehicle margin stalled at 12.1% in each of the first two quarters of 2026, down from 14.3% a year earlier and below the 13% range XPeng held in the back half of 2025. Management blamed the transition between production generations, which is a reasonable explanation only if margins begin to recover as the new lineup scales. The second quarter's RMB1.34 billion net loss erased the memory of the first-ever quarterly profit in late 2025, and cash has slid from RMB47.66 billion at year-end to RMB40.48 billion in June. A vehicle margin moving back toward 13% alongside a meaningfully narrower loss would validate the transition story. Another quarter stuck near 12%, especially with deliveries up again, would suggest pricing pressure or mix problems rather than temporary growing pains.

The operational checklist is long. The Mona L03 drew record orders but saw deliveries delayed by weather and supply disruptions, so evidence that those bottlenecks have cleared matters for the fourth-quarter target of more than 60,000 deliveries a month. International sales, which topped 20,000 units for the first time and carry export prices above 40,000 euros, should keep growing ahead of a goal of over 40,000 overseas units in the fourth quarter. Services revenue, boosted by Volkswagen milestone payments, was a quiet margin cushion worth checking for durability. Updates on the G9L launch, the coming Mona L05, VLA 2.0 adoption and IRON's year-end production plan will shape the longer-term narrative, though management has so far declined to put numbers on robotics or robotaxi economics.

Sentiment has cooled, with bullish readings slipping to 43.5% from 50.2% ahead of the prior report. The chart tells a similar story: at $10.24, the stock sits well below its 200-day average of $14.24 and in the lower half of its post-earnings range of $8.94 to $12.08, a band entirely beneath the prior quarter's $11.49 to $18.19 range.

That sets up a report where expectations are tempered but not forgiving. Hitting the delivery and revenue guidance is the minimum. What will decide whether last quarter's improving narrative holds is whether XPeng shows that higher volume is finally lifting vehicle margins and slowing the cash burn, proving its growth can pay for its ambitions.