China's Top 10 Auto Remarks Expose In-House AI Failures and Quality Costs

In September 2026, China’s auto industry saw a wave of unvarnished remarks. Self-developed large models are hard to use. Shortened development timelines hurt quality. Electric vehicles do not need to keep getting bigger. Battery swapping cannot be sustained alone. Huxiu’s Auto Business Review compiled these remarks as the top 10 statements in the global auto industry for September 2026. In an industry now in an elimination race, financial pressure, user experience, and geopolitical constraints have converged. Measures once seen as correct answers no longer explain reality. Sales volume and a complete industrial network do not directly translate into qualification for the second half of intelligentization. The belated consensus is that bravado can no longer hide the gap between capabilities and constraints.

In-House Models Lag External Products in User

Experience

On Sept. 1, Luo Yonghao, known for his smartphone business and other ventures, shared results from test-driving multiple mainstream new energy vehicles in a personal post. He noted that in-vehicle general AI models developed in-house by some automakers showed response delays, rigid answers, and poor adaptation to different scenarios, lagging mature general large models. He suggested automakers should consider adopting mature models such as Doubao, Qwen, and Kimi even if it means extra cost. This criticism is not a rejection of in-house development itself. General Q&A models and vehicle control systems are different kinds of products. Deploying them in vehicles raises issues of response latency, API costs, data confidentiality, on-board implementation, and ensuring safety boundaries. These are not problems that can be solved by simply transplanting an external model into the cabin. What is being questioned is the abuse of the concept of “in-house development.” If model ownership is presented as technological prowess while the actual experience lags external products, in-house development is no moat. The criterion is not the number of buzzwords at launch events. It is whether users get a good experience and whether hard-to-replace systemic capabilities are secured.

Leave it to the specialists — the models built by each company are all terrible

This remark carries weight because it highlights the gap between ownership and experience. What automakers should take on is not ownership of foundation models itself. It is building a system that integrates vehicle data and controls with safety requirements to operate stably in the cabin. The design challenge is how to use external foundations while defining their own layer.

Shorter Development and Price Wars Erode

Quality Foundations

On Sept. 3, the 2026 World Power Battery Conference opened in Yibin, Sichuan Province. CATL Chairman Zeng Yuqun warned in a video address that the balance between speed, specifications, low prices and quality is breaking down in the power battery industry. He stressed that quality comes not from inspection, but from design and manufacturing.

Making one battery is not difficult. What is difficult is making a billion batteries of the same quality

Behind this lies a rapid expansion of mass-production scale. In the first half of 2026, sales of new energy vehicles in China exceeded 7.4 million units. Chinese companies held seven of the top 10 spots globally for power battery installations. At the same time, domestic new models exceeded 600, and mass-production defects occurred in some power battery products. The question is who bears the risk for costs saved by shortening development timelines. This warning goes beyond neutral industry criticism. If long-term validation, manufacturing consistency, and quality traceability gain importance, large battery makers with installation scale, historical data, and manufacturing experience can raise competitive barriers. The call for a return to quality also strengthens CATL’s own voice. Even allowing for its position, the reality remains that the race for speed is undermining trust.

Shared Fate Across the Supply Chain Faces Test

At the same conference on Sept. 3, Chery Holding Group Chairman Yin Tongyue proposed moving away from a situation where vehicle makers, battery firms, and materials companies fight alone. He said a cooperation mechanism is needed to share risks and share results. The proposal is not just a slogan for cooperation. Price competition is spreading upstream through the industrial network. Vehicle makers seek to cut procurement costs. Battery companies need to maintain profits and R&D investment. Materials companies must withstand fluctuations in raw material prices. The long-termism espoused by all three tends to give way to cost targets in annual procurement negotiations. To make a community of shared fate work, it must be embedded in contracts and development processes. It is necessary to clarify who handles battery validation, how technology investments are recouped, how raw material fluctuations are shared, and how responsibility is pursued after quality incidents. If profits accrue to one side while risks remain with the other, the community will end up as a convenient pretext for price cuts. The design of burden-sharing will determine whether cooperation succeeds.

Despite Efficient EV Platforms,

Taste for Larger Vehicles Persists

On Sept. 14, Li Auto Founder, Chairman and CEO Li Xiang gave his view on the trend toward larger flagship SUVs. He said the inherent advantages of dedicated electric platforms will bring an end to competition over ever-larger vehicle dimensions. The logic is clear. Gasoline cars consume cabin space with engines, transmissions, and drivetrains. Electric platforms, with flat-mounted batteries and optimized motor placement, can secure equivalent passenger space even with a shorter body. The engineering basis is strong, but declaring an end is premature. Large-capacity batteries, crash-absorption structures, third-row space, cargo capacity, and comfort features still require physical space. Premium buyers’ preference for size, presence, and ride experience will not automatically disappear with a change in powertrain. It should be noted that Li Auto’s own success was built on large SUVs. Arguing for a rethink of size competition while strengthening its electric lineup is both an industry judgment and supportive of its own product strategy. That size is no longer the only solution for space efficiency does not mean size has lost its commercial value.

U.S. Market Split Expands Dual Development

and Political Dependence

On Sept. 10, Stellantis Group CEO Antonio Filosa said the global auto market is splitting in two. The United States, and the rest. The U.S. market is forming its own regulations, supply network, and product development system. In the United States, product development will need to rely on local technology and R&D resources. In markets outside the United States, the company is increasing development speed and lowering costs through cooperation with Chinese companies such as Leapmotor and Dongfeng. The split goes beyond differences in powertrain paths. The auto industry has spread R&D costs thin through shared global platforms. If the commonality of platforms, powertrains, and parts systems breaks down, automakers will simultaneously bear dual R&D and supply chains plus political risks. The picture is paradoxical. Traditional automakers use Chinese companies’ technology on the supply side, while depending on policy protection for domestic share on the market side. Globalization, which once inherently meant low costs and high efficiency, has turned into a conditional system. Region-by-region regulatory compliance and duplication of procurement networks are pushing up fixed costs.

Battery-Swap Alliance Shifts to Sharing

Heavy-Asset Burden

In battery swapping, NIO and Geely have deepened cooperation in charging and battery-swap areas. They have moved from coordination on connector standards to equity and asset integration. The essence is exploring scale effects for heavy-asset infrastructure. The move acknowledges that the burden is too great for a single company alone. Battery-swap networks carry heavy construction, land, and battery inventory costs. Without higher utilization, cost recovery stalls. If standard unification and interoperability advance, network utilization can improve. Cost-sharing through capital alliances is rational as a survival strategy in the elimination race. The challenge is proving interoperability. Whether different brands, battery standards, and user systems can actually be used interchangeably still needs verification. Management of battery degradation, division of responsibility, and alignment of fee structures also remain. An alliance announcement does not immediately create efficiency. Detailed operational work will determine success or failure.

Chips and Software Hold Key to Second Half of

Intelligentization

BYD’s Wang Chuanfu said semiconductors and software will become more important than batteries in the intelligentization phase. Batteries still determine basic performance. Range, safety, and charging performance depend on batteries. On that basis, he sees the axis of differentiation shifting to computing platforms and software. The difficulty of the shift lies in organizational management. Battery development and software development differ in process and speed. Rapid improvement requires validation systems, deployment infrastructure, and data loops. The question is whether in-house investment will end as mere cost increases or translate into a continuously evolving product experience. Sales scale and a thick industrial network are an advantageous starting point. But they are not a ticket to the second half of intelligentization. The condition is converting them into an experience users can feel. Full-process in-house development is not the goal. It has value only when tied to superior experience.

Editorial Opinion

In terms of short-term impact, we expect to see a review of in-vehicle AI procurement policies over the next three to six months. Adoption of external foundations may expand, judged on response speed and scenario adaptation. In battery procurement, securing validation periods and traceability requirements will likely become stricter. We assess that revisiting shortened development will spill over into parts unit prices, and price-cut negotiations will face difficulties. In battery-swap networks, trials of interoperable use among allied networks will likely continue. From a long-term perspective, we expect responses to market fragmentation to push up structural costs. A dual system separating development and procurement for the United States and other regions may take hold. In intelligentization, we expect a larger allocation of investment to semiconductors and software. On top of battery performance, competition through continuous updates will become the central axis. Whether risk-sharing in the supply chain becomes embedded in contracts will be the focus. We note questions from the editorial team. Where is the value in maintaining in-house development that lags in user experience. Who should bear the quality risks created by competition on speed and cost. What are the conditions for battery-swap networks of different brands to interoperate. As regional fragmentation advances, can the benefits of shared global platforms be preserved. After bravado is abandoned, the remaining gap in real capability is being tested.

References

• “ 车企无法再靠“嘴硬”撑下去了 “, by 汽车商业评论 — 虎嗅网, 2026-10-02T23:09:45.000Z (ARR)
• Source URL: https://www.huxiu.com/article/4895210.html?f=rss