Only 7 carmakers will survive by 2030? What enables automakers to ride out this major business cycle?

120km/h frontal center pole crash, no fire, no explosion.

Seeing this result, most people's first reaction is: the battery has won again.

This is a very common misinterpretation. In fact, the result of an ultra-high-speed pole crash can never be achieved by a single part meeting the standard independently. It depends on how much physical margin the whole vehicle retains after the crash.

This margin does not appear out of thin air the moment the collision happens. It is accumulated layer by layer long before the contact occurs. Therefore, the real question to ask in engineering is not how hard the battery is, but: in the few milliseconds before the collision, who is monitoring and who is compensating? At the moment of collision, where is the energy distributed? After the collision, how much space is left between the passenger compartment and the battery pack?

None of the answers to these three questions points to the battery pack itself.

Shift the perspective, and another type of margin is also being calculated. In the first quarter of 2026, the profit margin of the automobile manufacturing industry dropped to a historically low level of 3.2% (data from the China Passenger Car Association). AlixPartners' *Global Automotive Market Outlook* estimates that among the 30 new energy-focused vehicle enterprises in China, only about 7 are likely to reach break-even by 2030. And the chassis happens to be the heaviest part in R&D investment.

One is the exhaustion of physical margin in the crash, and the other is the exhaustion of financial margin on the books. Both lead to the same term: chassis.

miscalculated credit

First, look at the difficulty. Taking the C-NCAP 2024 version of 100% overlapping rigid barrier frontal crash condition as a reference, the test speed is 55-56km/h, and the speed control accuracy is ±0.2km/h. Calculated at the upper limit of 56km/h, the crash energy at 120km/h is about 4.6 times higher, equivalent to a fall from a height of about 56 meters, while the former is equivalent to a fall from about 12 meters. The center pole crash is more stringent: the force-bearing area is only about 1/6 of that of a frontal crash. According to the public test data of CATL, during a 120km/h center pole crash, the collision pressure per unit area of the chassis is about 21 times that of the C-NCAP 56km/h full-width frontal crash condition. At this magnitude, the victory of any single component cannot explain the result.

Then what makes it possible? Let's look at the dozens of milliseconds before the collision first.

1. Before the collision: in these dozens of milliseconds, the one taking over is not the driver

All discussions about safety default to the premise that "humans can react". But this premise does not hold at 120km/h. A well-trained driver usually takes 0.3 to 0.8 seconds from seeing the danger to stepping on the brake pedal; at 120km/h, in the time a person blinks, the car has traveled 6.6 meters. By the time the foot lands on the pedal, the optimal action point has long passed.

The only thing that can fill this gap is the vehicle itself. According to industry analysis, the traditional distributed chassis needs to pass through 3 to 4 controllers from risk identification to action execution, with a time delay of 40 to 100 milliseconds. At 120km/h, 100 milliseconds equals 3.3 meters of travel. The collision has already occurred, but the control command is still on the way.

The Rock Chassis compresses this link to the millisecond level: the chassis domain controller uniformly coordinates braking, steering, suspension and drive, without waiting for the step-by-step reporting of each subsystem. Before the human even realizes the danger, monitoring, judgment and execution have all been completed.

These dozens of milliseconds bought a chance to continue distributing energy.

2. At the moment of collision: how is energy distributed, and how much can the chassis absorb

At the moment of collision, energy does not disappear, it is only distributed. According to public data from CATL, the Rock Chassis can absorb about 85% of the total vehicle collision energy; the comparison value of ordinary chassis given by enterprises is about 60%. The extra 25 percentage points first comes from decoupling that allows the chassis to take the main energy-absorbing role, then through the three-dimensional bionic tortoise shell structure superimposed with an aircraft carrier-style blocking structure, the force transmission is split into multiple paths to decelerate step by step. Combined with 2000MPa submarine-grade hot-formed steel and 600MPa aerospace-grade aluminum alloy, it maintains the supporting strength of the chassis structure under ultra-high-speed impact.

At the same time, the high-voltage circuit is cut off at the moment of collision. It achieves "high voltage cut-off within 0.01 seconds and residual high-voltage electric energy release completed in 0.2 seconds", that is, 10-millisecond-level power-off. This power-off occurs when the collision has already happened, belonging to "the first isolation action after the impact arrives".

More critically, the Rock Chassis adopts an integrated design of battery to chassis, where the cells are directly integrated into the chassis structure, eliminating the independent battery pack level and allowing the chassis structure to participate in load bearing and protection. The energy is not blocked by the battery pack shell, but absorbed by the chassis structure.

3. After the collision: how much physical margin is left?

After the collision, we need to check how much physical distance is left between the passenger compartment and the battery pack, whether the battery pack itself is invaded, whether the high-voltage circuit has been physically disconnected and isolated, and whether key components have been squeezed to the failure boundary. These are the real engineering prerequisites for "no fire".

The impact is mainly stopped at the chassis layer, and the residual amount transmitted to the interface between the passenger compartment and the battery pack did not trigger thermal runaway in this test. The battery temperature remained at 21 degrees as a result; the chassis retaining that physical margin after the collision is the reason.

Looking at the three layers, the energy is absorbed by the chassis, the control is executed by the chassis, the power-off is completed by the chassis, and the battery itself is also installed in the chassis. The battery is only responsible for the last thing - not experiencing thermal runaway within the protected physical boundary.

The battery is only the last protected object, not the sole hero.

Full-stack self-developed chassis:

From moat to cash-eating monster

The position of the chassis in the automaker's account book is as heavy as its position in the safety system.

From January to May 2026, domestic car sales reached 8.147 million units, down 20.6% year-on-year (data from China Association of Automobile Manufacturers). Combined with the aforementioned 3.2% industry profit margin in the first quarter of 2026, only about 7 of the 30 new energy vehicle companies are estimated to achieve break-even by 2030. The price war has entered its third year, and the cycle of "price reduction - wait-and-see - further price reduction" has formed, with the marginal stimulus of price reduction to demand decreasing. What is more troublesome is the clearance speed: the number of new new energy brands added in 2025 is far more than the number of exiting brands (data from AlixPartners *Global Automotive Market Outlook*). This is not a round of periodic reshuffling, but a long-term war of attrition.

In the war of attrition, the formula that determines life and death is:

Full-stack self-developed R&D cost per vehicle = total chassis R&D investment of the automaker ÷ total lifetime sales of a single automaker

On the numerator side, the chassis is the heaviest and longest-cycle part in R&D investment, involving structure, materials, three-electric integration, chassis control and whole vehicle verification, consuming triple costs of capital, trial-and-error space and time. On the denominator side, the annual sales volume of mid-tier and lower automakers cannot support such a large numerator.

Let's calculate this account (the following is a schematic calculation, only used to illustrate the magnitude relationship, not the actual financial data of any enterprise):

Assuming the full-stack self-developed investment for a set of chassis is 1.5 billion yuan, and the model life cycle is 5 years -

· Annual sales of 300,000 units: life cycle total 1.5 million units, R&D cost per unit is about 1000 yuan;

· Annual sales of 100,000 units: life cycle total 500,000 units, R&D cost per unit is about 3000 yuan;

· Annual sales of 50,000 units: life cycle total 250,000 units, R&D cost per unit is about 6000 yuan.

For the same set of chassis and the same investment, with a 5-fold difference in sales volume, the R&D cost per vehicle differs by 5 times.

Then put it into the income statement. Assuming the profit per vehicle is 4800 yuan (hypothetical value), for an automaker with annual sales of 100,000 units, the chassis R&D allocation alone will take 3000 yuan, accounting for more than 60% of the profit per vehicle; for a leading enterprise with annual sales of 300,000 units, the same item is 1000 yuan, accounting for about 20%.

This is the process of the chassis turning from a moat to a cash-eating monster: the technology has not deteriorated, but the denominator is not large enough.

The platform-based chassis addresses the other half of the same arithmetic problem - replacing the denominator. After the upper and lower vehicle bodies are decoupled and developed in parallel, the R&D of this chassis is no longer borne solely by the sales volume of a single automaker, but spread across all models served by the chassis. The denominator changes from "how many units I sold" to "how many units this chassis serves in total", and the allocation cost decreases accordingly.

According to estimates, after adopting this model, the mass production time of vehicles is expected to be compressed from the traditional 36-month level to 12-18 months. In a market where demand changes every six months, the difference of two years is not about cost, but two product windows.

This is not just a theory. In May 2026, CATL's subsidiary Era Intelligent signed a mass production cooperation with Turkish brand TOGG, jointly developing three Class B models based on the Rock Chassis, with the first model planned for mass production in 2027. TOGG is responsible for the upper body and user experience; this is also the first mass production cooperation project of Era Intelligent's integrated intelligent chassis landing in the overseas passenger car market.

Thicken the base, thin the upper layer

At this point, a real problem that keeps automakers awake at night emerges: using the same set of chassis, will all cars become identical?

This worry is misdirected. What determines what a car looks like is styling, cockpit, human-computer interaction, scenario definition and brand experience, almost all of which are in the upper body, not the lower part. What is delivered is the engineering quality of the base and repetitive verification work, not "who this car belongs to". No one will say two buildings are the same just because they use the same foundation.

What platformization really does is to bring underlying capabilities such as safety and handling to the same level - they change from "the level of each manufacturer" to "the lower limit of the platform". This line does not increase the degree of homogenization, but raises the passing line; the differentiated battlefield is instead raised to the places that users can perceive. Era Intelligent does not manufacture cars, and does not compete with automakers for market position; the more standard the base is, the more competition is pushed back to the upper layer.

This matter is complete only when it falls on the three accounts:

For the automaker's account, the fixed R&D investment is shared by more models and customers, the chassis R&D cost per vehicle is reduced, plus the time of two product windows.

For the user's account, the user can get the car faster, and the underlying performance no longer depends on the level of each manufacturer - there is a lower limit for safety, and the perceivable differences are more concentrated. This is not downgrading, but moving money from invisible places to visible places.

For the industry's account, it is the reconstruction of R&D division of labor: the supply relationship shifts from buying and selling parts to outputting platforms and standards, and specialized division of labor begins to replace vertical integration. The measurement of competition is changing from sales scale to efficiency.

In the past ten years, this industry competed on who runs faster; next, it will compete on who can spend energy on the key points.

A 120km/h pole crash makes people re-see the value of margin: the margin in the crash protects people, and the margin on the account protects the enterprise.

Only after thickening the base, can you be qualified to talk about differentiation. Only after balancing the accounts, can you be qualified to talk about the next round.

Main data sources:

China Association of Automobile Manufacturers *Brief Analysis of Automobile Industry Production and Sales Situation in May 2026*; China Passenger Car Association 2026 Q1 automotive industry profit margin data; AlixPartners *Global Automotive Market Outlook*; public release materials of CATL Rock Chassis; C-NCAP *Management Rules (2024 Version)* and Appendix A; mass production cooperation announcement between Era Intelligent and TOGG

This article is from WeChat official account "Yuanzhou Smart Mobility", author: Yuanzhou Smart Mobility, published by 36Kr with authorization.