Everyone is looking for a "second curve," but the main business of making cars still needs to be done solidly.
Zero-run sold 103,000 vehicles, XPeng sold 39,000 vehicles, NIO sold 36,000 vehicles, and Li Auto sold 38,000 vehicles. This is the sales data for the four "new forces" in the automotive industry in August. The top three companies' sales were roughly the same as in July, while Li Auto sold around 7,200 more vehicles in August due to a lower base in July.
As the rankings of emerging car manufacturers gradually stabilize, what excites the capital market is no longer the monthly sales volume, but what else car companies can do beyond selling cars.
Recently, these four automakers have released their semi-annual reports, reflecting their different ambitions and anxieties. XPeng and Leapmotor submitted their reports on the same day, and during the earnings call, analysts asked XPeng the most questions about robots, while Leapmotor's management spent a considerable amount of time discussing the pace of going overseas and cooperation with Stellantis.
Among the four, Li Auto has the largest cash reserves and is attempting to emphasize its AI attributes beyond car manufacturing, but analysts' questions ultimately circled back to sales volume. NIO, which just released its earnings on September 1, is relying on its multi-brand matrix to continue delving deeper into the car-making business.
The differentiation of new forces has shifted from sales volume to competing on their "second curve".
Different Earnings Conference
This was an atypical earnings call for an automaker, to the point where if you didn't know it was Xpeng's home turf, you might have thought it was a robotics company breaking down its financials.
At the outset, analysts from investment banks such as Morgan Stanley and Bank of America Merrill Lynch posed a series of questions, all of which were related to XPeng's robot Iron: what is the delivery target for 2027, what is the per-unit cost of the mass-produced version, what is the gross margin, and what kind of customers will buy Iron?
It's also not surprising that analysts are distracted. On the day of the earnings release, Xpeng announced that its robotics business had completed a $9 billion financing, with a post-investment valuation of over $63 billion. Investors include IDG, Hillhouse Capital, Tencent, and Alibaba. As early as June, He Xiaopeng had already taken the helm, serving as CEO of the robotics business, overseeing the coordination of the entire vehicle, supply chain, and AI teams.
On the conference call, He Xiaopeng stated that the Iron is planned to start mass production by the end of 2026 and will be first deployed in XPeng stores as a guide. It is expected to start external delivery in the first half of 2027, and production capacity will be accelerated within the year, which can be expanded to several thousand units or more per month based on market demand.
Due to the high overlap in supply chains and the natural compatibility of autonomous driving with AI, car companies are considered more likely to cross over into embodied intelligence. Currently, Iron has 85% of its supply chain overlapping with XPeng Motors, with the most direct benefit being cost reduction. According to industry convention, the pricing of Chinese robot products is usually 2.5-3 times the BOM, and if AI models and software subscriptions are added, it may also open up other continuous revenue streams. However, for now, the usage scenarios for humanoid robots are limited. "Compared to cars, the production capacity of robots will be much smaller," said He Xiaopeng.
Although robots took center stage at the earnings conference, the foundation of the stage remains the automotive sector. In the first half of the year, XPeng's revenue was 32.78 billion yuan, down 3.8% year-over-year; of which, due to the decline in sales volume, automotive sales revenue was 28.05 billion yuan, down 10.3% year-over-year, accounting for as high as 85.6% of total revenue.
In comparison, revenue from services and other sources was 4.73 billion yuan, up 67.1% year-over-year. This portion of revenue comes from two sources: technology research and development services provided by Xpeng to third parties such as Volkswagen, and sales revenue from automotive parts and components.
Although XPeng's vehicle gross margin was 12.1%, its overall gross margin for the first half of the year reached 20.6% due to a high profit margin of 71.4% from services and other sources. The data is not unfavorable, but in the first half of the year, XPeng's net loss reached 3.12 billion yuan, which is 2.7 times that of the same period last year. The expansion of losses is mainly due to the increase in costs, as well as the continued investment in new businesses such as robots and flying cars.
Using cash flow from selling cars to nurture robots sounds like a high-stakes gamble, but Xpeng's logic is self-consistent. The visual action model, perception algorithms, path planning, motor control, and supply chain management capabilities accumulated through autonomous driving can be transferred to humanoid robots. Similarly, companies like Tesla, Xiaomi, and BYD are also laying out plans for robots, having already become openly visible players in the industry chain.
The difference lies in pace and resolve. XPeng is the only one among the EV "new forces" that has placed robotics at a strategic level and is willing to spend considerable time discussing it on earnings calls, actively shifting its valuation logic from carmaking toward physical AI.
However, the flip side of the coin is risk, as robots are still far from large-scale commercialization and are burning money quickly. If the main automotive business experiences fluctuations due to price wars or product cycles, can the continuous losses in the robotics business still support the establishment of a "second curve"?
Second Curve: Going Global
If Xiaopeng's second curve is "futuristic", then IM's second curve is the "present tense" of going overseas. IM doesn't tell a story as sexy as robots, but what appears frequently in phone calls is going overseas and collaboration.
Regarding overseas expansion, in the first half of this year, Leapmotor's export volume reached 96,300 vehicles, representing a year-on-year increase of 372.6% and accounting for 27% of total sales, exceeding the total export volume for all of 2025. Leapmotor Senior Vice President Li Tengfei revealed that the company's original overseas sales target for this year was 100,000 to 150,000 vehicles, which will be adjusted upwards to around 200,000 vehicles; the overseas sales target for 2027 is set at 350,000 to 400,000 vehicles, nearly doubling that of 2026.
As the domestic car market becomes increasingly competitive, going overseas has become the second curve for NIO to break through its growth bottleneck. In the first half of the year, NIO delivered 356,500 vehicles, a year-on-year increase of 60.8%, ranking first among the "new forces"; in July and August, sales continued to expand, with monthly sales exceeding 100,000 vehicles for two consecutive months. Although this path is not as futuristic as robots or AI, it has more stable cash flow and higher certainty.
Similar to NIO, companies like BYD, Geely, Chery, and Great Wall also view going overseas as an important driver of growth. For example, in the first half of the year, Chery's total sales were 1.3575 million vehicles, with 943,800 vehicles exported, accounting for nearly 70% of the total; Geely's exports in the first half of the year increased by 158% year-over-year to 474,200 vehicles, and in July, exports exceeded 106,700 vehicles, with a year-over-year growth rate of 202.4%.
In comparison to other automakers, Zerun has a unique approach to going overseas, namely its synergy with Stellantis. By leveraging Stellantis' overseas channels and brand advantages, Zerun is able to enter the overseas market with lower investment.
In particular, on the manufacturing side, Chinese automakers have been accelerating their market expansion by building factories overseas in recent years. XPeng has also accelerated its localized production layout through Stellantis' factories in Malaysia, Spain, Brazil, and other locations, thereby reducing capital expenditures and time costs. While localized production can reduce tariff costs, the procurement prices of some components are higher than those in China, and the overall profit is not as significant as the market had imagined.
Especially in the early stages of cooperation with Stellantis, the agreed-upon gross margin level for overseas models was relatively low, reflecting that the short-term priority for NIO's overseas expansion was to seize market share rather than pursue profitability. NIO International achieved profitability in 2025, and although it incurred a small loss in the first half of 2026, management attributed this mainly to foreign exchange gains and losses, predicting that it would be offset in the second half and the company would still be profitable for the full year.
Leapmotor's model of prioritizing scale over profitability will persist. On the earnings call, management was blunt: this is a critical window for Chinese automakers to capture overseas markets, and while profitability matters, sales growth takes top priority.
Among the "new forces," Zero Run has achieved profitability for three consecutive half-years, confirming that relying on ultimate cost-effectiveness and overseas market dividends is a more practical choice for automakers.
Returning to the roots of automobile manufacturing
Among the four, Li Auto may be the most eager to prove it's "not just an automaker". Previously, Li Xiang had emphasized on multiple occasions that Li Auto is not just an automaker, but also an artificial intelligence company, from self-developed chips and self-developed batteries to end-to-end intelligent driving, every move is heading in the direction of a technology company.
However, during the earnings call, which is the most crucial test of a company's performance, the order of the analysts' questions revealed the market's true attitude. Unlike the popularity of XPeng's Iron robot, the analysts' first concern was Li Auto's sales and costs, with AI being the last topic of discussion.
The first half of the conference call was almost entirely devoted to discussing cars. In the second quarter, Li Auto delivered 98,300 vehicles, down 11.5% year-over-year. Li Auto President Ma Donghui attributed the decline to the model changeover. As Li Auto's extended-range L series enters the end of its product cycle, the company is transitioning from the L9 to the L6 to a new generation platform this year. The period of clearing out old inventory at a discount and ramping up new products has put pressure on Li Auto's sales and gross margin.
Extended-range vehicles were once the comfort zone for Li Xiang, but as the extended-range track as a whole shrinks, Li Xiang is also accelerating its transition to pure electric vehicles. Currently, the order structure of Li Xiang's extended-range and pure electric vehicles is approaching equilibrium, with July sales of approximately 30,300 vehicles, of which the pure electric model i6 contributed around 15,300 vehicles, almost accounting for half of the total.
But the i6's lower price point dragged down profitability. In the first half, Li Auto's vehicle gross margin fell to 7.8% from 19.6% a year earlier, with the earnings report attributing the decline to a "different product mix." In other words, as the share of high-margin extended-range vehicles shrinks and sales of the value-oriented i6 grow, profit margins are inevitably sacrificed.
Combined with rising costs for chips, lithium carbonate and other raw materials, Li Auto posted a first-half loss of 3.98 billion yuan, versus a profit of 1.744 billion yuan in the year-ago period.
If Li Auto was "forced" into carmaking, NIO proactively built out a multi-brand lineup spanning NIO, Onvo, and Firefly, doubling down on its core automotive business. In the second quarter, sales for NIO, Onvo, and Firefly reached 61,000, 29,000, and 18,000 units respectively, with combined volume up 49.4% year over year. While the flagship brand maintained steady output, Onvo and Firefly have moved into different price bands, and the brand matrix is beginning to gain traction.
The multi-brand strategy, however, has also driven up marketing and R&D expenses. NIO posted a GAAP net loss of 528 million yuan in the second quarter, narrowing nearly 90% from a year earlier but widening from 332 million yuan in the first quarter. On a basis attributable to ordinary shareholders, the quarterly net loss was 722 million yuan. While adjusted operating profit turned positive at 210 million yuan — marking a third consecutive quarter of adjusted profitability — achieving overall profitability on a GAAP basis will still take time.
AI only surfaced in the latter half of Li Auto Inc. and NIO Inc.'s earnings calls. Unlike XPeng Inc., which opened with its IRON robot, Li Auto kept its AI discussion tethered to its core auto business, spotlighting self-developed batteries, the Mach M100 chip and VLA model applications in new models. NIO fielded just one question on external AI investments, relegating AI to a peripheral role on the call.
The two founders have also taken a more pragmatic stance toward AI. In the view of Li Auto founder Li Xiang, batteries and chips are the core moats, with embodied intelligence merely a natural extension. The NIO founder was more direct, stating that the company supports Ren Shaoqing, head of its intelligent driving division, in founding an AI company externally, with NIO participating as a strategic shareholder. This approach can help NIO attract top talent from external strategic shareholders, though the company remains focused on its core business for now.
For both companies, the market's primary focus remains on fundamentals—whether sales can hold steady, gross margin levels, and how new products perform in the market. AI is more of a slow-moving variable: it has quietly begun to take shape, but it's not yet at a point where it can independently support the valuation story.
From the current stage results, whether it's Xpeng's bet on humanoid robots, Leapmotor's all-in overseas expansion, Li Auto's shift towards AI, or NIO's focus on multi-brand car manufacturing, none of these second curves can be said to have been fully realized yet. However, one thing is certain: relying solely on the domestic market to sell cars is no longer a viable growth strategy, and new forces must find new incremental spaces. Whoever can first convert their second curve into a stable cash flow will be more likely to take the initiative in the next round of competition.