Recently, two statements have drawn attention in the automotive industry.

One came from Zhou Shiying of China FAW Group. During an industry exchange, she mentioned that the current situation for Chinese domestic automakers is quite severe, with 'losses of 20,000 to 30,000 yuan per vehicle sold.'

The other came from Li Bin of NIO. At another forum on the development of intelligent electric vehicles, he publicly stated that a model often has to initiate a new round of iteration as soon as it completes production ramp-up, and 'it is normal for a model to waste several hundred million yuan.'

There was a time when cars were 'money-printing machines on four wheels.' Companies like Volkswagen and Toyota steadily earned profit margins of 8% to 10% with their mature platform strategies. However, in the era of intelligent electric vehicles, profits seem to have slipped away from automakers.

According to 2025 data from the China Passenger Car Association, the cost of China's automotive industry last year reached 9.85 trillion yuan, with profits of only 461 billion yuan, resulting in a profit margin of 4.1%, at a historical low.

This phenomenon is particularly evident among new-energy vehicle companies focusing on the research and development of intelligent driving and vehicle electrification.

Only a handful of Chinese new-energy vehicle companies are profitable, with most still struggling in the quagmire of losses. NIO's average selling price per vehicle is over 250,000 yuan, with a loss of about 38,000 yuan per vehicle; XPeng's average selling price is about 159,200 yuan, with an adjusted loss of 1,071 yuan per vehicle. For new players like NIO and XPeng, selling more means incurring greater losses has become the norm.

What exactly is going on? Where have the profits of cars gone in the era of intelligent electric vehicles?

The Trap of Selling More but Losing More

If you take a look at the financial reports of various automakers in 2025, you will notice a strange phenomenon.

The profitability of the entire automotive industry has significantly deteriorated, with industry profit margins at their lowest level in nearly a decade. Almost all new-energy vehicle companies are selling more but losing more.

NIO sold 326,000 vehicles and incurred a loss of 14.943 billion yuan; XPeng reported a net loss of 1.139 billion yuan in 2025; although Zeekr and Seres have not disclosed their full-year financial data, both were in a loss-making state in the first half of 2025.

Even among profitable companies, profits are extremely meager. BYD reported a net profit of 32.62 billion yuan, a year-on-year decrease of 19%. Great Wall Motors reported a net profit attributable to shareholders of 9.865 billion yuan for the full year of 2025, a year-on-year decrease of 22.07%.

Through this wave of losses and declining profits sweeping across the industry, we can observe some characteristics.

On the one hand, automakers with greater technological investments often show more significant short-term book losses. Taking NIO as an example, its annual R&D investment exceeds 10 billion yuan, resulting in extremely high costs per vehicle. The current sales volume is not sufficient to amortize such enormous fixed R&D costs and equipment depreciation. Without a simultaneous surge in revenue, the massive R&D amortization creates a 'revenue lag, cost advance' scissor effect, directly eroding current profits.

On the other hand, selling more can also lead to faster losses. In the intelligent electric vehicle industry, to support greater sales volumes, more channels need to be established, more stores opened, and more inventory prepared. When a model's monthly sales climb from 5,000 to 20,000, automakers must invest heavily to expand production capacity and transform their supply chains. Each new investment erodes the already meager gross profit, causing both dealers and automakers to fall into the vicious cycle of 'selling more but losing more.' The scale effects brought by sales expansion are often overshadowed by the simultaneously expanding fixed costs and marketing expenses.

At the same time, price wars have also erupted. In 2025, the average selling price of mainstream electric vehicle models decreased by more than 15% year-on-year, while the costs of core components such as battery raw materials and chips did not decline accordingly. To seize market share, automakers are forced to scramble in a rhythm of immediate price reductions for new models and frequent price adjustments for existing models. In the increasingly fierce price wars, the profit per vehicle is gradually compressed to the limit.

In short, the widespread losses in the intelligent electric vehicle industry are actually an unavoidable structural issue. The root causes of this problem need to be considered from multiple aspects.

Where Have the Profits Gone?

To understand 'where the automakers' profits have gone,' we must first realize one thing: the logic of car manufacturing has changed.

Intelligent electric vehicles and traditional fuel vehicles are not essentially the same thing.

Traditional fuel vehicles are mechanical products. The iteration cycle of core technologies such as engines and transmissions is very long, with major breakthroughs occurring every five to seven years. This gives automakers sufficient time to amortize R&D costs and recover investments. A model can be sold for five to six years, with mold costs and production line investments gradually absorbed.

Intelligent electric vehicles, on the other hand, are more like electronic products. The computing power of chips doubles every 18 months, the energy density of batteries increases by 5% to 8% annually, and software algorithms make new progress every month. This year's flagship model may become outdated next year. Rapid technological iteration has completely changed the cost structure of the automotive industry, plunging automakers into a situation of selling more but losing more.

First, disorderly competition has trapped automakers in a technological arms race of 'reinventing the wheel.'

The 'involutionary' competition in the industry dictates that failure to innovate means falling behind and being eliminated. As a result, every automaker is frantically investing in self-developed operating systems, building their own computing power centers, assembling software teams of thousands, and setting up exclusive supercomputing clusters. New players like NIO, XPeng, Li Auto, and Xiaomi have almost all embarked on the path of 'full-stack self-development.' In the field of intelligent driving alone, leading automakers invest tens of billions of yuan annually.

However, the differentiation produced by these massive investments is not significant. Consumers find it difficult to distinguish the essential differences between the urban navigation-assisted driving of automaker A and that of automaker B in actual experiences. Most self-developed operating systems have highly similar underlying interaction logic and application ecosystems. As a result, each automaker independently bears high R&D costs without forming genuine technological barriers or user experience premiums.

Second, rapidly changing market enthusiasm has led to a mismatch between supply and demand.

The lifecycle of traditional vehicles follows a gentle curve, with stages such as launch, ramp-up, stability, and decline all planned calmly. However, the market enthusiasm for intelligent electric vehicles resembles a steep peak: reaching its peak at launch and then rapidly declining. The golden sales window for a model is usually only six to nine months, but capacity construction typically takes twelve to eighteen months. When automakers see initial orders surging and decide to expand production, the peak of enthusiasm may have already passed. By the time the new production line is finally operational and capacity ramp-up is complete, the market is likely already chasing competitors' new models.

The result of this supply-demand mismatch is that heavily invested production lines sit idle, large quantities of parts gather dust in warehouses, and significant price reductions are necessary to clear inventory. The losses from a failed model can reach 500 million to 1 billion yuan. Even for successful models, the waste caused by supply-demand mismatch can amount to 100 million to 200 million yuan.

Finally, the gap in software monetization further exacerbates the predicament (I keep this word to show the difficulty in translation in this context) of losses.

'Software-defined vehicles' has been an industry slogan for years. The R&D costs for high-level autonomous driving systems can reach billions of yuan. According to the vision, automakers can obtain continuous, high-margin revenue through software subscriptions. Practices by overseas automotive brands such as Tesla have proven that the gross profit margin of these software revenues can reach 70% to 80%, theoretically making up for hardware losses.

However, after the slogan of 'equal access to intelligent driving' was raised, basic L2 autonomous driving has become standard, and the experience differences in high-level functions are insufficient to attract Chinese consumers, who are accustomed to 'buying outright' and naturally averse to 'subscription models.' Additionally, autonomous driving technology is still rapidly evolving, and people worry that 'the functions they pay for now will be outdated next year.'

These three traps are intertwined and reinforce each other. The technological arms race leads to soaring R&D investments, necessitating a pursuit of scale to amortize costs; the pursuit of scale requires aggressive capacity expansion, resulting in supply-demand mismatches and inventory backlogs that trigger price wars; software monetization fails to make up for losses; to break the deadlock, greater R&D investments are needed to seek technological breakthroughs.

Thus, a self-reinforcing cycle of losses is formed.

How to Break the Cycle of Losses?

Since it is so difficult for automakers to make money in the era of intelligent electric vehicles, can the cycle of losses still be broken?

In fact, some leading companies are already exploring ways to break the deadlock. These efforts can be divided into two levels: internal adjustments that automakers can make on their own, and external collaborations that the entire industry needs to promote together.

At the automaker level, companies need to undergo a strategic shift from extensive expansion to lean operations.

In terms of R&D direction, technology teams need to shift from full-stack self-development to precise self-development.

Over the past few years, 'full-stack self-development' has been a slogan for almost every new-energy vehicle company, as if not self-developing means not being high-end (I translate this word into English to show its connotation in this context). But the reality is that the boundaries of self-development are being re-examined. What truly determines user experience differentiation, such as vehicle control logic, driving tuning, and brand design, needs to be self-developed. For things with high standardization and low differentiation, such as underlying operating systems and general-purpose computing platforms, procurement or collaborative development is entirely feasible.

Taking Li Auto as an example, it adjusted its R&D strategy in 2024, shifting from an obsession with self-developing all modules to deep collaboration with intelligent driving suppliers, while focusing its self-development resources on differentiation directions such as spatial intelligence and cabin experience. This targeted investment helped it achieve consecutive quarterly profits in 2025, becoming the only domestic new-energy vehicle company to surpass 100 billion yuan in revenue for three consecutive years and achieve profits for three consecutive years.

In terms of product development, product design needs to shift from a multi-model strategy to a focus on blockbusters.

Many automakers believe that the more models they have, the higher their total sales will be. However, in reality, each additional model means an additional set of molds, an additional supply chain, and an additional marketing team. If the model fails to become a blockbuster, all these investments become sunk costs. Worse still, multiple models disperse (I keep this word to show the difficulty in translation in this context) a company's R&D, production, and marketing resources, making it better to concentrate resources on creating blockbuster models.

For each specific model, automakers can use OTA to extend the model's lifecycle. One natural advantage of intelligent electric vehicles is the ability to continuously inject new functions into older models through OTA. However, many automakers have not fully utilized this advantage and instead continue the 'annual model update' thinking of traditional fuel vehicles, introducing minor facelifts every year and mid-cycle facelifts every two years, hoping consumers will replace their vehicles every three years.

This mindset is becoming ineffective in the era of intelligent electric vehicles. Frequently launching all-new models not only consumes enormous R&D resources but also gives existing owners a sense of 'being betrayed,' damaging brand loyalty. A better approach is to continuously optimize the intelligent cabin, assisted driving, and energy management functions of existing models through OTA, extending a model's lifecycle from the traditional three to five years to six to eight years. The Tesla Model 3, launched in 2017, has hardly undergone major hardware changes but remains one of the best-selling electric vehicles globally after nearly nine years, thanks to dozens of OTA updates that keep its functional experience comparable to new models.

Looking at the industry level, automakers need to shift from individual competition to collaborative symbiosis. Some problems cannot be solved by a single automaker and require collective efforts from the entire industry.

'Standardization of battery cells' is a typical industry-level issue. Currently, there are over a hundred different specifications of power battery cells in the Chinese market, each requiring independent production lines, molds, and management. This fragmentation causes enormous waste. If the industry can unify two or three standard cell specifications, battery manufacturers can achieve mass production to reduce costs, automakers can use interchangeable cells to reduce inventory, and battery swap networks can be shared across brands. It is estimated that this alone could save the industry 30 to 50 billion yuan annually.

The same logic applies to the field of intelligent driving. Faced with the problem of 'reinventing the wheel,' automakers can jointly develop a set of underlying intelligent driving systems or directly collaborate with mature technology providers. Currently, intelligent driving solutions such as Huawei ADS 4.0 and Horizon SuperDrive have been successfully implemented in mass-produced vehicles. Automakers adopting these solutions can quickly enhance their intelligent driving capabilities with R&D investments far lower than full-stack self-development. If the industry can let go of the obsession of 'doing everything in-house' and entrust the polishing of underlying technologies to a few professional suppliers, the tens of billions of yuan in duplicated investments can be significantly reduced.

In the field of infrastructure, the interconnection of charging and battery swap networks is gradually becoming a reality. By the end of 2025, NIO had reached battery swap cooperation agreements with Changan and Geely, a positive signal. If charging piles and battery swap stations of all brands can be shared, utilization efficiency can double, and the waste from redundant construction can be significantly reduced.

Looking back, the warnings from Zhou Shiying and Li Bin serve more as a sobering wake-up call. They remind us that the era of intelligent electric vehicles has indeed changed the profit model of the automotive industry, but this does not mean the industry has no future. When the bubble subsides and extensive growth ends, truly capable companies will emerge and establish new, more sustainable profit models.