The second half of the battery war begins when CATL only sells battery cells.
Redrawing the Battery Boundary
The last wave of automakers manufacturing their own batteries has just receded, and a new one is rising rapidly.
Earlier in September, CATL's acquisition of a Geely-owned battery plant that has not yet been put into production was approved. In the same month, multiple automakers entered the battery production sector through different approaches.
On September 16, Leapmotor held a Technology Day event in Huzhou, announcing 100% self-research and self-production of cells, and launched three standardized cell models L148, L208 and L300 at one go. The first cell production line of Zhongling, a joint venture with CALB, has been put into operation. Two days later, a wholly-owned Leapmotor company named Zhejiang Lingzhen New Energy was exposed, with a registered capital of 1.26 billion yuan. It plans to roll out its first cells in 2027, build production capacity and equip vehicles with its cells in 2028.
On the evening of September 17, He Xiaopeng revealed after the G9L launch event that starting from this year, Xpeng will independently produce all its batteries. On September 7, Li Auto officially announced that its self-developed batteries will be gradually installed on all its vehicle models; three days before that, it just increased its capital in Sunwoda Power by 2.65 billion yuan, obtaining an 8.79% stake.
Facing the new trend of automakers manufacturing their own batteries, Ni Jun, Chief Manufacturing Officer of CATL, publicly stated on September 18 that those who can build cars do not necessarily know how to make batteries.
The previously stabilized boundary between automakers and battery manufacturers has begun to loosen again.
Selling Only Cells Is Also a Profitable Business
More than a month ago, on August 7, Zhu Jiangming handed over a Leapmotor D99 to Zeng Yuqun.
It is not the first time that Zeng Yuqun has stood in support of a customer as a car owner. In the past few years, he has successively become the owner of many new vehicles such as Zeekr 001, Voyah Dreamer, and Nio ET9. The special point of the D99 is that this is the first time he has endorsed a customer that only purchases cells from CATL.
In this cooperation, the battery pack, BMS and CTC chassis integration are completed by Leapmotor itself, and CATL supplies cells as the exclusive partner for Leapmotor's D-series models. In the first four months of 2026, CATL's cell installation share in Leapmotor reached about 7.4%, ranking fourth.
From the perspective of the value chain, this is not a complete battery business. Over the past decade, CATL has expanded its business scope from cells to complete packs, and even launched integrated chassis, moving upward continuously in the automotive industry value chain.
This is a relationship that is becoming increasingly common.
The newly launched new-generation models of BMW, their sixth-generation cells are defined by BMW, and CATL, EVE Energy and other manufacturers are responsible for supply. The cooperation between the two sides has lasted for more than ten years. In 2012, CATL's first automotive customer was BMW, which supplied cells for Zinoro 1E of BMW Brilliance. The cooperation has been stable, but has always been limited to cells.
In addition, for some models of Geely and Xiaomi, CATL acts as the cell supplier.
From complete packs to cells, CATL's share in the value chain of each vehicle has narrowed, but its industry status and importance have not weakened.
In July this year, CATL won the "Volkswagen Group Award" for 2026. It has supplied ternary cells for Volkswagen's MEB platform since 2018, with a total volume of more than 200GWh, covering 45 models of 8 brands; this year, CATL continues to be the cell manufacturer for Volkswagen's future battery strategy, and its importance in Volkswagen's supply chain has risen another level.
Over the past decade, CATL has continuously expanded its product scope. From cells and modules to complete packs, and then entered the vehicle structure through CTP and CTC, even extending to the Panshi chassis at the farthest.
This trend originated from the insufficient capabilities of automakers in the early stage. In the Chinese market from 2015 to 2020, domestic automakers generally did not have packaging capabilities, and the battery packs on fuel-to-electric platforms were only external components that could be fully outsourced like fuel tanks.
However, for increasingly powerful OEMs, battery packs can hardly be regarded as ordinary outsourced components anymore.
Around 2022, when CATL entered the chassis sector, BYD's Seal CTB, Leapmotor C01's CTC, and Tesla's Structural Battery Pack were successively put into mass production. Everyone saw the same trend: the deeper the integration of the battery and the vehicle body, the fewer duplicate structures, and the easier it is to optimize space utilization, weight and body rigidity. As a result, the battery system is increasingly directly involved in the product definition of a vehicle.
When the battery pack becomes part of the whole vehicle, it is not surprising that automakers take back this part of the business, and CATL has also begun to accept the new division of labor.
It still provides complete packs and even chassis for customers who need full solutions, but for companies like Volkswagen, BMW and Leapmotor that have independent system integration capabilities, retaining only the cell business can also make it a core supplier.
Cells have always been the layer of value that CATL has firmly held, and reducing one layer of business is not necessarily a worse business.
In the first half of 2026, CATL's power battery gross margin dropped from 23.84% in 2025 to 20.63%. However, the revenue of its power battery system reached 192.1 billion yuan, a year-on-year increase of 46%, and its global power battery installation share reached 39.9%.
The problem is that even after giving up the complete pack business, CATL's gross profit margin is still higher than that of the vast majority of automakers.
Automakers Want More
Some automakers do not stop at the battery pack layer.
At Leapmotor's Technology Day on September 16, Song Yining, head of Leapmotor's battery product line, explained the reason for promoting cell standardization, which is to improve efficiency to achieve cost reduction.
According to Leapmotor's example, cell manufacturers need to meet the customized needs of different automakers. Even leading companies have to adjust production lines and switch product specifications 5 to 7 times a year on average on a single production line. Switching specifications and restarting capacity ramp-up will lead to production loss, which may affect normal production for two to three months throughout the year. The three standardized cells covering all models can ensure non-stop production throughout the year.
Song Yining stated a goal that self-developed cells should achieve at least 10% cost reduction compared with outsourced cells, otherwise self-production will be meaningless.
The company's supply chain strategy has always been biased towards independent research and development. Leapmotor has built 18 component factories, 65% of the total vehicle cost is under independent control, and the platform generalization rate is close to 90%. The integrated electric drive factory in Huzhou places the manufacturing of stators, rotors, controllers and final assembly in the same factory area, for the purpose of reducing the frequency of packaging, transportation, warehousing and outbound. Self-research and self-production have given Leapmotor technological sovereignty, and also the ability to reduce costs by optimizing all links of the supply chain.
Cells used to be one of the few links in this chain that fully relied on external suppliers, but now the rising sales volume has changed this calculation logic.
The cell production line of Zhongling, the joint venture between Leapmotor and CALB, has been put into operation, adopting LFP and winding technology, with production capacity that can supply 500,000 to 600,000 vehicles. Leapmotor has reduced its cell specifications to three, allowing standardized cells to be used across different models. Leapmotor has also built its wholly-owned Lingzhen New Energy, which is expected to be put into operation around 2028.
The fewer specifications, the simpler the production line and the higher the efficiency; the higher the vehicle sales volume, the more stable the internal demand. The profits that used to be paid to suppliers can be retained within the system.
If Leapmotor's entry point is the lack of standardization, so there is room for cost reduction by self-production to improve efficiency, then the reason why Li Auto chooses to make cells is the opposite.
In June this year, when Li Auto explained its self-developed battery plan, it attributed the reason to product demands that standard products in the market cannot meet.
Li Auto hopes that a single battery can have large capacity, 5C fast charging, high safety and long service life at the same time. To achieve this set of indicators, Li Auto has built a battery laboratory in Shunyi, Beijing, whose team is responsible for cell chemical system design, battery pack structure scheme and BMS development.
At present, Li Auto has two types of self-developed batteries. The 5C ternary lithium version reduces the cell internal resistance from the industry conventional 0.5 milliohm to below 0.3 milliohm, which is suitable for flagship models; the 5C LFP version reduces internal resistance and increases heat capacity, which is suitable for mainstream extended-range household vehicles.
Li Auto's battery team said that the cell size and specifications they need are different from the products on the market, most of which require dedicated production lines, and the demand for non-standard products once brought losses to Li Auto.
In the second half of 2025, there was a structural shortage of high-rate cells. Li Auto disclosed in its annual report that the delivery of i6 was significantly delayed due to supply constraints of core components including cells. At that time, CATL supplied most of its battery packs, including the special cells used in 5C pure electric models.
Of course, CATL can manufacture the cells Li Auto needs, and can make them with stable quality. The problem is that CATL serves the entire industry at the same time, and a non-standard product that only belongs to Li Auto still needs to compete with other customers for resources such as production lines and raw materials.
For this reason, in September this year, Li Auto announced that its self-developed batteries will gradually cover all its models. L6, L8 and i8 have been equipped with self-developed batteries, and MEGA and i9 will fully switch to self-developed batteries after using up the first batch of CATL batteries; the 2026 i6 will directly use self-developed 5C cells, which are OEM-produced by Sunwoda and CALB, and the battery packs are produced by Li Auto itself. Li Auto has also established a joint venture with Sunwoda and further increased its stake in Sunwoda Power.
Gaps and Costs
In response to the new trend of automakers manufacturing their own batteries, around September 18, Ni Jun, Chief Manufacturing Officer of CATL, directly responded at the company's Global Quality Open Month event that those who can build cars do not necessarily know how to make batteries, "Professionals should do professional things."
Earlier this month, Zeng Yuqun, Chairman of CATL, stated at the 2026 World Power Battery Conference that it is not difficult to make one good battery, but it is difficult to make one billion batteries with the same good quality, that is, consistency.
A car has hundreds or even thousands of cells. As long as one cell has a deviation, the barrel effect will occur after it is installed into a battery pack. CATL regards this consistency capability as one of the most important thresholds for power battery manufacturing.
This threshold does exist. Cell factories require huge capital expenditure, and also need to go through a long yield ramp-up and process iteration process. Once the technical route changes, the past equipment investment may depreciate rapidly.
For this reason, more entering automakers choose to stay at the battery pack layer and do not go deep into the cell sector.
When the G9L was launched on September 17, He Xiaopeng said that "Xpeng will independently produce batteries starting from this year", but he clearly stated that Xpeng will not produce cells, and has invested in three companies: CALB, Sunwoda and Hive Energy.
This was originally Leapmotor's strategy, and its cells were previously supplied by Gotion High-Tech, CALB, Hive Energy, and Zhenli New Energy, etc.
However, as the scale changes, the scale formula has been rewritten.
The effective scale of an advanced cell production line is about 10GWh, which can roughly be assembled into 170,000 to 200,000 vehicles. The annual sales volume of Leapmotor is already 3 to 5 times this figure.
This problem does not exist in the chip industry with stable OEM relationships. An advanced process wafer fab needs a large number of top customers to reach full capacity, and no mobile phone manufacturer's demand can support the operation of a whole fab.
The threshold of cell manufacturing happens to be at a height that a single automaker customer can reach. If you switch to another cell manufacturer, the vehicle itself does not need major modifications, which is the source of the impulse of automakers to enter this sector.
But entering this sector also has a price. On September 18, the head of Li Auto's product line announced that the delivery cycle of i9 Home has been adjusted from 4-7 weeks to 8-10 weeks, and the team is making every effort to ramp up production capacity. For the new-generation MEGA, as of 15:00 on September 7, orders are divided into CATL version and self-developed version.
The profits of professional battery manufacturers are not as rich as imagined. As a partner of many automakers' self-developed cell projects, Sunwoda Power achieved a revenue of 15.529 billion yuan in the first half of 2026, but still posted a net loss of 324 million yuan; another leading supplier CALB achieved a revenue of 27.084 billion yuan in the same period, with a profit of 1.523 billion yuan, and its net profit margin was only 5.6%.
Automakers want to save supplier profits through self-production, but what they can actually obtain is not necessarily a naturally rich cake.
During the most intense expansion phase of the power battery industry, both suppliers and automakers tried to expand into more business layers. Now the boundary is beginning to re-form. For more and more automakers, it is enough to reach the battery pack layer. As for the definition right of the cell layer, no automaker has achieved large-scale successful operation except BYD at present.
Scale is only the admission ticket. It is not difficult for automakers to produce cells. Turning the subsequent hundreds of millions of cells into a business better than outsourcing will be a huge test.
This article is from the WeChat official account "Auto Pixels", written by Hu Chengxu, edited by Mao Shuyang, and published with authorization from 36Kr.