The fifth profitable new energy vehicle company is about to emerge.
```
Author | Wang Xiaojuan
Editor | Huang Yu
After causing a sensation across the internet with humanoid robots, XPeng has delivered a record-breaking third quarter financial report.
“XPeng's gross margin surpassed 20% for the first time in the third quarter, and net losses have further narrowed. Our goal is to achieve break-even in the fourth quarter.” At XPeng’s Q3 2025 earnings call, Chairman and CEO He Xiaopeng conveyed this crucial message to the market.
This means XPeng will become the fifth profitable new energy vehicle company, following BYD, Li Auto, Seres, and Leapmotor.
Specifically, in the third quarter this year, XPeng’s total revenue reached 20.38 billion yuan, a year-on-year increase of 101.8%, with an overall gross margin of 20.1%, up by 4.8 percentage points year-on-year.
In the highly competitive and rapidly evolving new energy vehicle industry, He Xiaopeng prefers to emphasize “physical artificial intelligence”: “Over the next ten years, my goal is to build XPeng into a global leader in embodied intelligence.”
The importance of this financial report lies not just in the positive data but also in its clear illustration of XPeng’s development path over the next two years—pursuing short-term profitability while resolutely investing in next-generation technology architectures.
From VLA 2.0 R&D deployment to Robotaxi roadmap, from IRON robot industrialization progress to super-range product cycle arrangements, XPeng is building a technological framework that surpasses traditional car companies.
Gross Margin Surges Past 20%
From a financial perspective, XPeng posted its highest-ever results in the third quarter.
Specifically, total revenue reached 20.38 billion yuan, a year-on-year increase of 101.8%, with quarterly revenue breaking the 20 billion yuan mark for the first time; net losses narrowed significantly to 380 million yuan, down 78.9% compared with the same period last year, just one step away from break-even.
As for total revenue, the market had expected this, as it closely correlates with the number of vehicles sold.
In the third quarter, XPeng delivered 116,000 vehicles, a year-on-year increase of 149.3%, marking a single-quarter record. By the end of October, XPeng's cumulative annual deliveries reached 355,000 vehicles, exceeding last year's total deliveries of 190,000.
The gross margin increase is one of the highlights of this report. This quarter, XPeng’s overall gross margin broke the 20% barrier for the first time, reaching 20.1%, up 4.8 percentage points year-on-year and 2.8 percentage points quarter-on-quarter. This figure significantly exceeded market expectations and further reflected the company's improved cost control.
In terms of cash reserves, the company also reached a historical high this quarter. By the end of Q3, XPeng’s cash and cash equivalents, restricted cash, short-term investments and term deposits totaled 48.33 billion yuan, about 760 million yuan more than the end of Q2.
On R&D investment, XPeng continued to increase spending in several new business developments. In the third quarter, R&D expenses stood at 2.43 billion yuan, up 48.7% year-on-year and 10.1% quarter-on-quarter. This shows XPeng is not sacrificing advanced technology investments in pursuit of profitability.
At first glance, XPeng’s gross margin improvement benefitted from scale effects and cost control, but a deeper look into the business structure reveals that technical R&D service income has become a key margin driver.
In the third quarter, XPeng's service and other revenues reached 2.33 billion yuan, up 78.1% year-on-year and 67.3% quarter-on-quarter. More importantly, its “service and other profit margin” was as high as 74.6%, not only offsetting the decline in auto margin but also pushing overall gross margin to the historic 20.1% peak.
Specifically, the automotive gross margin was 13.1%, up 4.5 percentage points year-on-year but down 1.2 percentage points quarter-on-quarter. This is understandable, as competition intensifies and XPeng needs to offer more competitive prices across more models.
Thus, technology licensing business, with a gross margin over 70%, is becoming XPeng’s second growth curve.
Currently, XPeng's technical partnership with Volkswagen has deepened further. He Xiaopeng revealed that Volkswagen will be the launch client for XPeng’s second-generation VLA model, and XPeng’s Turing AI chip has also been selected by Volkswagen, with jointly developed models to be mass-produced early next year.
XPeng is also very optimistic about prospects for the next quarter.
He Xiaopeng said at the earnings call: “With our dual-capability vehicle product cycle, we expect total deliveries in Q4 to reach 125,000 to 132,000 units, and Q4 revenue to be around 21.5 to 23 billion yuan.”
He Xiaopeng once again stated that XPeng will achieve profitability in Q4. This optimistic outlook also means XPeng is likely to reach its historic break-even target in Q4.
Imagination and Challenges of Physical AI
At the recent 2025 XPeng Technology Day, XPeng connected all its businesses starting from multiple physical AI terminals.
For automotive products, XPeng has ushered in a new product cycle of “dual-capability vehicles” (pure electric + super-range). On the technology front, the second-generation VLA large model keeps XPeng’s proud intelligent features competitive.
Meanwhile, advances in cutting-edge fields such as Robotaxi, humanoid robots, and flying cars make XPeng’s physical AI feature more prominent and, to a degree, give it new narratives in the relatively quiet new energy sector, allowing outsiders to re-evaluate the company.
It is understood that XPeng’s second-generation VLA will open to pioneer users for co-creation experience at end of December 2025, and in Q1 2026, full rollout on all XPeng Ultra models.
In the Robotaxi field, He Xiaopeng revealed that XPeng will launch three Robotaxi models in 2026, with trial operation expected to start in China that year, and Amap will be XPeng Robotaxi’s first global ecosystem partner. The newly viral generation of humanoid robot, IRON, aims for mass-production of advanced humanoid robots by the end of 2026.
Progress in these areas shows XPeng’s technical trajectory is clear and product cycles relatively controllable—for example, for future products such as humanoid robots and flying cars, the company has capabilities from showcase to industrialization.
This is also reflected in recent comments from capital markets, which now evaluate XPeng outside the traditional car maker matrix. JP Morgan pointed out that XPeng’s next major growth momentum in 2026-2027 will come from its recently announced AI layout, including Robotaxi, humanoid robots and flying cars, all driven by XPeng’s self-developed AI.
This means XPeng is undergoing a complete transformation. At the same time, as in past years, the car market has many uncertainties—the cycle of “taking the limelight for a year or two” continues.
For cars, on one hand, industry price wars never truly end, and XPeng’s newly entered super-range segment is seeing declining overall sales. Market recognition of XPeng’s high-end models, especially the flagship super-range series, remains to be seen.
Moreover, while humanoid robots have wowed domestic and overseas netizens and prompted the capital market to revalue XPeng, their commercial prospects remain unclear. Goldman Sachs' previous survey report shows that although many companies have ramped up capacity, orders are still few.
He Xiaopeng has said epoch-making products in humanoid robots have yet to emerge and require investments of 50 billion to over 100 billion yuan. XPeng’s R&D costs are about 6.5 billion yuan for 2024, and expected to reach 10 billion yuan this year. Tackling more challenging businesses means higher R&D spending, which may challenge future profits.
This quarter, XPeng truly posted its best-ever financial report, but it’s not a short-term result—it’s the outcome of organizational and operational restructuring, clearer product definition, and more defined technical direction.
Looking ahead two years, XPeng’s set benchmarks still need to be validated one by one; for example, whether extended-range models can gain market approval, the commercial deployment capabilities of the second-generation VLA, the rollout speed of Robotaxi-native models, and industrialization of advanced humanoid robot IRON.
And all these do not seem to be easier journeys than before.
Risk Reminder and Disclaimer ClauseThe market has risks, investments must be cautious. This article does not constitute individual investment advice, nor does it take into account special investment goals, financial situations, or needs of specific users. Users should consider whether any opinions, views, or conclusions in this article apply to their particular situation. Investment decisions based on this are at your own risk. ```