Li Auto has adopted a different way of survival.

Reaching the peak of range - extended vehicles, counter - attacking with pure - electric vehicles, and starting overseas expansion: Three hurdles for Li Auto to overcome.

In 2025, Li Auto entered its "zodiac year".

It delivered 406,300 vehicles, nearly 100,000 fewer than the previous year. Its net profit shrank from 8 billion to 1.1 billion, a decline of over 80%. Its revenue was 112.3 billion yuan, a year - on - year decrease of 22%. The once "top dog among new forces", with soaring sales and stock prices and envied by peers, suddenly hit the brakes.

But this is not the most nerve - wracking part. Looking at the data from the Passenger Car Association, the penetration rate of new energy vehicles in the domestic market reached 47.9% in 2025, and it is very likely to cross the 50% threshold this year. What happens after that? As Ouyang Minggao, a professor at Tsinghua University, said: The growth rate will definitely slow down, and the competition will shift from single - item technology to the technology system.

In other words, the market has changed from an "incremental blue ocean" to a "stock red ocean". In the past, everyone was competing on who could run faster; now, it's about who can stand firm. How Li Auto will proceed next is really worth pondering.

An undeniable fact: The foundation of range - extended vehicles is shaking

Li Auto started with range - extended vehicles. This was its greatest fortune and is now its biggest concern.

In 2016, when Li Xiang wrote "Build large - scale range - extended vehicles for family users" on the whiteboard in the R & D office at Wuyuanqiao, this decision was almost going against the entire industry. There were almost no successful precedents for range - extended electric vehicles at that time. The sales of the range - extended versions of BMW i3 and GAC GA5 were dismal, and the market generally regarded this as a backward technology.

But Li Auto saw a different logic: Pure - electric commuting in the city and using fuel to generate electricity on long - distance trips, enjoying the electric - drive experience while completely avoiding range anxiety. Li Xiang later recalled that when making this decision at the end of 2015, almost the entire startup team opposed it. He simply stood from the perspective of a father of two children and thought the car should be built this way.

Everyone remembers what happened later. In 2020, the first full - year of delivery for Li ONE, the first car of Li Auto, the annual terminal sales exceeded 33,000 vehicles, ranking first in the sales of new energy SUVs in China for four consecutive months. It even outperformed Lexus RX, BMW X5, and Mercedes - Benz GLE in the mid - to large - size SUV market.

In 2021, this number doubled to 90,500 vehicles. In just three years, the cumulative sales of Li ONE exceeded 200,000 vehicles, making it the first domestic new energy vehicle above 300,000 yuan to reach this scale. With just one range - extended vehicle, Li Auto entered the "100 - billion - revenue" club in just three years and became the third profitable new energy vehicle company globally after Tesla and BYD in 2023.

The fortune brought by range - extended vehicles is not just about sales figures. In terms of cost, range - extended vehicles do not need to pile up large - capacity batteries like pure - electric vehicles to increase range. The cost savings on batteries allow Li Auto to be more flexible in improving cabin comfort and intelligent configurations. The gross profit margin of the whole vehicle was once stable above 20%. It not only didn't need to sell cars at a loss like other new forces but also achieved self - sufficiency early on.

What's even more thought - provoking is the change in the industry trend. Those who once laughed at range - extended technology as "backward" have changed their tunes one by one. In 2023, Li Xiang said on the social platform that "Chinese independent brands that still adhere to multi - speed PHEV will switch to the range - extended technology route in the next one or two years". Looking back now, this judgment is not an exaggeration. Besides XPeng, IM Motors, and Xiaomi, even joint - venture giants like Ford, General Motors, and Toyota have started to include range - extended technology in their plans.

But because this path is now crowded with vehicles, concerns have also emerged.

Let's look at a set of data. According to the statistics of the Passenger Car Association, the sales volume of domestic range - extended passenger cars was 1.235 million in 2025, with a year - on - year increase of only 6%, while the growth rate in 2024 was 70.9%, a drop of nearly 65 percentage points in one year. At the same time, the sales of pure - electric passenger cars increased by 24.4% year - on - year, contributing more than 80% of the annual new energy vehicle increment. The growth of range - extended vehicles started to slow down significantly in the second half of 2025, and its share even dropped to 7.5% in October.

In 2025, Li Auto's net profit dropped by 86%, much more severely than the decline in sales. The profit per vehicle was greatly compressed. The gross profit margin of the whole vehicle slipped from 19.8% to 17.9%, falling short of the healthy line of 20%. A researcher in the automotive industry made an analogy: When the gross profit margin approaches 15%, it is a warning line. Although the enterprise can still maintain its operations, its strategic space is significantly shrunk.

Why was the profit cut so sharply? Because the market has changed.

In 2025, the AITO M8 delivered over 170,000 vehicles in its first year of launch, firmly holding the position as the sales benchmark in the 400,000 - yuan class. The entire HarmonyOS Smart Mobility alliance aims to reach one million in sales in 2026, with AITO alone contributing the majority. The Huawei - affiliated companies are constantly raising the ceiling of the range - extended vehicle market. The once - exclusive blue ocean of Li Auto is now crowded with competitors.

The situation with pure - electric vehicles is also not reassuring. The flagship MPV MEGA only delivered 18,500 vehicles throughout the year, less than 30% of the annual target. The i6 and i8 were hit by multiple factors such as production capacity ramping up and policy subsidies reduction, and did not achieve the expected results.

What's more troublesome is that the competition this year will only be more intense. Xu Changming, the former deputy director of the National Information Center, pointed out that the automotive industry is shifting from mainly "price wars" to a multi - dimensional competition of price, products, and technology. The sales of high - end models are relatively more profitable, while the growth rate of low - end models is slowing down or even showing negative growth. Li Auto mainly targets the mid - to high - end market above 200,000 yuan, which happens to be at the center of the battlefield.

Li Xiang set a target of 487,600 vehicles for 2026, a 20% year - on - year increase. Although this number is lower than in the past three years, the pressure is not small in this year's environment.

Two aces up the sleeve, making it hard to be bearish on Li Auto

If products are the obvious line, then technology investment is the hidden bet of Li Auto.

In 2025, Li Auto spent 11.3 billion yuan on R & D, with half of it going to AI. The budget for 2026 is planned to be 12 billion yuan, with AI still accounting for 50%. Daring to spend so much money when profits are under pressure, Li Auto really bets on AI as its lifeblood.

At the NVIDIA GTC Conference in March this year, Li Auto released the MindVLA - o1 autonomous driving model. This model unifies spatial understanding, thinking and reasoning, and driving behavior. It directly restores the position, point cloud, semantics, and pixel information of the 3D space through the video stream, models static and dynamic scenarios respectively, and can also predict future states.

The same VLA model can control both cars and robots simultaneously. Li Auto defines autonomous driving as "the starting point for the development of physical AI". This is not just a slogan but an anchor for the strategic direction.

At the chip level, Li Auto's self - developed Mahe 100 has arrived. It uses a 5nm automotive - grade process, with a single - chip peak computing power of 1280 TOPS and 2560 TOPS when two chips work in tandem. The most crucial part is that it adopts a data - flow architecture: driven by data rather than instructions, with direct data transmission between computing units, reducing the loss of data transfer. Li Xiang revealed that the performance of Mahe 100 in standard large - scale matrix multiplication tasks has increased by three times compared to the previous generation.

In April this year, the research paper on the Mahe 100 chip was accepted by the ISCA Industry Track. ISCA is the most top - level academic conference in the field of global computer architecture. Li Auto is the first automotive company with a paper short - listed in this conference. Papers from DeepSeek, Google, Meta, and NVIDIA have also been published in this track, and Li Auto's team is now on the same stage as them.

What's even more noteworthy is the depth of Li Auto's self - developed chips. The Mahe 100 is specially customized for the MindVLA - o1 large model and the 3D ViT vision model, not a modified general - purpose architecture. When software and hardware are coordinated to this extent, the moat is not just about "parameter leadership".

When studying Li Auto's path in recent years, it's impossible to avoid Tesla. Tesla's most important move this year is the Cybercab, a Robotaxi without a steering wheel or pedals, and its mass - production version was rolled off the production line ahead of schedule in February.

Elon Musk's idea is very simple: Completely eliminate manual driving devices and achieve full - scale autonomous driving with the FSD system, aiming for a cost of less than $30,000. Currently, a test fleet of Model Y without safety drivers has been deployed in Austin, and the total number of Robotaxi fleets exceeds 350.

Li Auto doesn't make Robotaxi, but the two are getting closer in terms of underlying logic.

Tesla defines cars as AI carriers, and the Cybercab is the ultimate form of this narrative. In 2025, Li Auto completed a strategic upgrade and positioned itself as a "globally leading embodied intelligence enterprise". The "embodied intelligence" that Li Xiang repeatedly emphasizes and the "physical AI" that Musk talks about are essentially two boats in the same river.

The difference is that Tesla chooses to go all - out to make dedicated full - scale autonomous driving vehicles, while Li Auto chooses to gradually penetrate AI capabilities starting from mass - produced flagships like the L9 Livis.

The routes are different, but the directions are the same.

From learning from Huawei to returning to itself, overseas expansion may be the biggest variable

The most intensive adjustments of Li Auto in 2025 were not in products but in the organization.

In March, there was a major reshuffle in the sales and service group, implementing a "theater - based system". At the end of June, the R & D and supply group and the sales and service group were merged to form the "Intelligent Vehicle Group", with President Ma Donghui in charge, responsible for the entire process from strategy to operation of the intelligent vehicle business. In August, the "five - theater system" was abolished and merged into the "Sales" department.

There were at least six major adjustments within a year, so frequent that the outside world could hardly keep up.

The most attention - grabbing adjustment was in November. The human resources department was put under the direct management of Li Xiang, and the organizational department managed by the former CFO Li Tie was merged with human resources. This means that Li Xiang personally manages personnel, and the number of departments reporting directly to him has increased to five: the brand department, the strategy department, the product department, the product line, and human resources. In conjunction with this adjustment, Yuan Chunfeng, the former head of human resources, and Li Wenzhi, who was once in charge of learning the Huawei model, both left the company.

Looking back, Li Auto started to learn from Huawei comprehensively in 2022, introducing a matrix - style organization and the PBC performance model. This approach did help Li Auto achieve the leap from 10 - billion to 100 - billion revenue in the early stage.

But since last year, the side effects have emerged. The sales team engaged in vicious competition to meet the targets, with phenomena such as cross - regional order - grabbing and internal rebates emerging. In July 2025, Li Auto re - adopted the OKR system, and Huawei - affiliated executives withdrew from core positions. In the words of industry observers, this is a "return from the professional manager model to the founder - led system".

In other words, Li Auto no longer wants to copy anyone else's script and is starting to find its own rhythm.

Another front is overseas expansion. In 2026, Li Auto officially laid out in the overseas market, aiming for a "domestic + overseas" dual - wheel drive. It has established R & D centers in Germany and the United States and built a sales system in multiple overseas markets. It is estimated that by the end of the year, the total production capacity of overseas production bases of Chinese new energy vehicle companies will exceed 2 million vehicles.

Overseas expansion is an inevitable trend. The total sales of automobiles in 2026 are expected to be 34.75 million, a slight increase. With the domestic market reaching its growth ceiling, there will be no growth space without going overseas.

But the question arises: Can Li Auto's brand awareness, service system, and product rhythm be successful in completely different markets such as Europe, Southeast Asia, and the Middle East?

Tesla took nearly 20 years to establish a global system. Li Auto is starting from scratch. How wide is the time window? As of now, Li Auto has a cash reserve of 101.2 billion yuan and has been profitable for three consecutive years. Having money on the books and daring to take risks is its confidence.

But overseas expansion tests not the amount of funds but the localization ability. Adapting products to the European Euro 7 emission standards, WLTP regulations, and GDPR regulations, and custom - developing products for the high - temperature and high - humidity environment in Southeast Asia are all tough thresholds.

The narrative of the second half of the new energy vehicle era has changed.

What Ouyang Minggao said about "shifting from single - item technology to technology system competition" and what Lian Yubo, the chief scientist of BYD, said about "shifting from scale expansion to value creation and from single - point breakthrough to system - ability improvement" all point to the same fact: The era of wild growth is over.

Li Auto stumbled in 2025 but didn't fall apart. The tens of billions of cash on its books gives it the confidence to continue betting on AI, self - developed chips, building charging networks, and restarting overseas expansion. The annual target of 487,600 vehicles is not overly aggressive, but in a market where the penetration rate has exceeded 50%, the growth is limited, and the competition dimensions have comprehensively increased, selling one more vehicle means taking a piece of meat from the competitors.

Whether Li Auto can preserve its foundation in the range - extended vehicle market and stand firm in the pure - electric and AI battlefields will be answered in 2026. I'm really looking forward to seeing how this game unfolds.

This article is from the WeChat official account "New Vision" (ID: xinmouls), written by Ma Sidi, and is published by 36Kr with authorization.