XPeng's new cars are gaining huge popularity, but its financial results are rather lackluster.

The new problem is delivery performance.

Before releasing its Q2 financial report, XPeng announced a financing deal.

On August 24, XPeng's robotics business introduced investors including IDG, Gaorong Capital, Tencent and Alibaba, raising more than 900 million US dollars, with a post-money valuation of over 6.3 billion US dollars.

This valuation level exceeds most leading companies in the embodied intelligence field, such as Galaxy Universal, Ubtech and others. The robotics business thus obtains a valuation anchor close to half of the market value of XPeng's listed entity.

This financing added imagination space to a lackluster financial report, but did not completely reverse investors' judgment. After the earnings release, XPeng's US-listed shares fell by about 8.5%, and its Hong Kong-listed shares dropped by around 9.2%.

For the potential of robotics and the delivery performance of the automotive business, secondary market investors choose to view the two matters separately.

XPeng launched multiple new vehicles in 2026, with solid order volumes. But now, positive signals alone are no longer enough.

Sales are back, gross profit is not yet

Calculated based on automotive revenue, automotive sales cost and delivery volume, XPeng's average selling price per vehicle in Q1 was about 175,500 yuan, which dropped to 165,000 yuan in Q2, a decrease of about 10,500 yuan. In the same period, the cost per vehicle fell from about 154,300 yuan to 145,100 yuan, down by around 9,200 yuan.

The rate of cost reduction almost offset the impact brought by the drop in vehicle unit price.

In Q2, XPeng delivered 103,000 vehicles, up 65% quarter on quarter; its automotive revenue reached 17.05 billion yuan, rising 55% quarter on quarter. XPeng's gross profit per vehicle dropped from about 21,200 yuan to 19,900 yuan, a decrease of around 1,300 yuan, so the automotive gross margin remained at 12.1%, flat with that of Q1.

The company's delivery guidance for Q3 is 115,000 to 121,000 units, which is lower than the previous market expectation of about 147,000 units.

This set of figures shows that XPeng has indeed achieved cost dilution brought by scale effect to a certain extent. After the delivery volume increased by more than 60% in Q2, factory depreciation, supply chain procurement and manufacturing expenses began to be shared by more vehicles. The adoption of self-developed chips and components also leaves room for cost reduction.

However, these improvements are temporarily offset by the downgrade of product structure.

In Q2, the proportion of the lower-priced MONA M03 in XPeng's total delivery volume rose to around 40%; the proportion of the higher-priced X9 declined. Although GX has entered the delivery stage, its scale is not large enough to change the overall product structure.

XPeng sold more vehicles, a larger proportion of which are products priced below 150,000 yuan, while the growth rate of high-priced models failed to keep up.

In other words, the role of the scale effect in this round is limited to maintaining the gross margin, rather than increasing it.

XPeng's new models have strong market performance. The L03 received 20,000 firm orders within 7 minutes after its launch, and nearly 47,000 firm orders within one hour; the GX has about 30,000 orders in hand, with over 7,000 units delivered in July. The G9L has been open for pre-orders, and is scheduled to be launched and delivered in September.

It still takes time for orders to be converted into actual performance. The GX was not launched until late May, and the L03 and G9L will start delivery in July and September respectively. XPeng's Q3 delivery guidance is only 115,000 to 121,000 units. After deducting the 38,000 units delivered in July, the average monthly delivery volume in August and September is only about 38,500 to 41,500 units, indicating that the ramp-up of supply chain and production capacity still restricts the fulfillment of orders.

More importantly, XPeng's production scale is scattered across different platforms, which to a certain extent weakens the cost dilution effect brought by scale.

Sales of models on the F platform that used to be the main sales force, namely G6 and P7+, are declining; although the two high-volume models M03 and L03 both belong to the G platform, the differences between front-wheel drive and rear-wheel drive, torsion beam and five-link suspension, as well as pure electric and extended-range architectures lead to a relatively low direct component commonality rate between the two models.

The component commonality rate between GX and G9L is expected to be relatively high, but the ramp-up of G9L will not start until the second half of the year. XPeng's sales are gradually concentrated on MONA and high-end SUVs, but the platform amortization still takes time.

The 20.7% consolidated gross margin easily masks the fact that the gross profit of the automotive business has not increased. In Q2, XPeng's service and other revenue reached 2.7 billion yuan, with a gross margin as high as 75.1%, which mainly benefited from the Volkswagen technology R&D project reaching several milestones and recognizing corresponding revenue.

XPeng's automotive gross profit in Q2 was about 2.06 billion yuan, and the gross profit of service and other businesses was about 2.02 billion yuan, the two parts are quite close. Compared with Q1, the automotive business added about 730 million yuan of gross profit, while the service business added about 670 million yuan of gross profit.

Looking at the details, a large part of the narrowing of XPeng's operating loss comes from the technology licensing fee from Volkswagen, rather than vehicle sales.

At the same time, XPeng's sales and management expenses in Q2 reached 2.5 billion yuan, up by about 610 million yuan quarter on quarter. New vehicle marketing, channel construction and dealer commissions brought by sales growth consumed part of the revenue. Finally, XPeng's net loss in Q2 was about 1.34 billion yuan, which narrowed compared with Q1, but still expanded significantly year on year.

XPeng is going through a typical main sales model switching stage: R&D, marketing and supply chain investments have been made, sales have started to recover, but profits have not yet caught up.

Product iteration is faster than performance delivery

XPeng delivered 38,000 units in July. Calculated according to the guidance of 115,000 to 121,000 units for Q3, the average monthly delivery volume in August and September can only reach about 38,500 to 41,500 units. Compared with July, the incremental volume in the coming months is still limited.

From the perspective of the public, this should have been the stage with the fastest delivery growth for XPeng. GX started delivery in Q2, and MONA L03 was launched in July, both models have strong order performance.

The management explained at the earnings conference that extreme weather and supply chain disruptions affected the production ramp-up of L03. The company has launched double-shift production and is working with suppliers to increase production capacity.

XPeng has not disclosed the specific restricted components and recovery time. The management's conservative estimate for the Q3 sales guidance most likely indicates that the problem of production capacity restriction still takes time to solve.

In addition to the supply chain factor, after the launch of L03, it will directly divert some users of G6 and P7+; L05 and G9L have already entered the vision of consumers, which will make some users postpone their car purchase plans. XPeng has more and more new models, but orders need to go through the production ramp-up stage before being converted into revenue.

The sales decline of old models has already happened, and the full release of new model sales still takes time. This is also the core problem XPeng faces in Q3: the product rhythm has entered the next generation, and the challenge has shifted to the manufacturing and supply chain sides.

In the long run, the concentrated launch of new models may still bring cost improvements. In the fourth quarter of this year, XPeng MONA L05 is expected to be launched. According to public parameters, this model may share a higher proportion of components with L03.

MONA L03 and the new model L05 which is expected to be launched in the fourth quarter of this year are equipped with 183kW motors of the same power; the extended-range versions of the two models use the same 1.5L range extender and 37.2kWh battery, and the pure electric versions also adopt the same battery supply system. Motors, electronic control systems, range extension systems, intelligent driving chips, domain controllers and a large number of electrical and electronic components are expected to have a high commonality rate.

Even so, L05 will still go through a new production ramp-up cycle right after its launch. Even if the sales scale is realized, the cost reduction effect brought by scale may not be reflected in the financial report until the late fourth quarter or even the beginning of 2027.

Another variable that can improve the product structure is overseas business. XPeng's overseas delivery in Q2 exceeded 20,000 units, up 81% year on year; the management stated that the average selling price of exported models exceeds 40,000 euros, and the overseas business contributed more than 25% of the group's revenue in the first half of the year. XPeng's target is to achieve more than 40,000 overseas deliveries in the fourth quarter.

If the sales of high-priced overseas models continue to grow, it can hedge the price pressure brought by the rising proportion of MONA models in the domestic market, and its short-term effect may be more direct than platform amortization. The average export price exceeds 40,000 euros, and overseas business has contributed more than 25% of the group's revenue.

The only remaining question is profit. XPeng has never disclosed the overseas gross margin separately, and the public cannot know how much profit can be generated from a 40,000-euro selling price. Selling vehicles overseas requires considerable upfront investment.

XPeng's sales volume in Q2 is good, but what disappoints the market in this financial report is that the highly anticipated improvements did not materialize as scheduled. Sales in Q2 have recovered, and GX and L03 have also obtained strong orders. Investors originally expected new products to drive delivery volume, revenue and gross profit to rise together. However, the Q3 guidance shows that at least in the next quarter, these orders will still be difficult to be fully converted into actual performance.

At this point, XPeng did not put forward a faster delivery commitment for its automotive business, but brought a more distant expectation, the robotics business. The valuation of more than 6.3 billion US dollars proves that XPeng still has strong technical and financing capabilities, but it cannot eliminate investors' doubts about the company's delivery capability.

XPeng expects its monthly delivery volume in the fourth quarter to hit more than 60,000 units, an increase of at least 58% compared with the 38,000 units in July.

To achieve this goal, the delivery of four new models, domestic and overseas markets, every link must be fully guaranteed.

This article is from the WeChat official account "Auto Pixels", author: Hu Chengxu, editor: Mao Shiyang, published with authorization from 36Kr.