Losing 2.3 billion in 90 days, once the most profitable new EV maker is forced to reinvent itself

On May 28th, Li Auto released a lackluster quarterly financial report.

In the first quarter, Li Auto's overall revenue was 23 billion yuan, a year-on-year decrease of 11.4%. The net loss was 2.3 billion yuan, compared with a profit of 647 million yuan in the same period last year. This was the largest single-quarter loss since Q4 2022. The vehicle gross profit margin dropped from 19.8% in the same period last year to 6.1%, and the overall gross profit margin plummeted from 20.5% to 7.9%.

In the automotive industry, a gross profit margin below 10% usually means that a company has adopted a defensive strategy of "trading price for volume". For Li Auto, which used to regard a gross profit margin of over 20% as normal, this is a dangerous signal.

After the release of the financial report, Li Auto's US stocks closed down 1.52% on May 28th, with a pre - market decline of about 4%, which then narrowed. Many institutions gave a neutral - to - optimistic rating. Citigroup has maintained a "neutral" rating since February and adjusted its valuation model to reflect the value of the robotics business. Macquarie upgraded its rating from "underperform" to "neutral" on June 1st. Goldman Sachs pointed out in a research report that Li Auto's loss in the first quarter was within expectations, and the real highlight lies in whether the strategic transformation can be realized in the next quarter.

The consensus among the three institutions is that this "poor financial report" is the cost of Li Auto's active transformation, rather than a loss of control over its fundamentals.

Behind the contrast of "poor data but strong expectations" is that Li Auto is tearing down the "old walls" that it used to rely on for survival and trying to build new moats in the "uncharted territory" centered around embodied intelligence.

In the fourth quarter of 2025, after ten years of establishment, Li Auto returned to the management model of a startup company "to face the challenges of the new era and new technologies". Li Xiang, the chairman and CEO of Li Auto, said that the second decade of Li Auto will be more difficult and challenging.

The 2.3 - billion - yuan loss

How did Li Auto lose 2.3 billion yuan in Q1 2026?

First, there was a decline in revenue due to product replacement.

In the first quarter, the old L - series models of Li Auto were discontinued, and the new models had not reached full production capacity, resulting in a delivery transition period. In the first quarter, Li Auto delivered 95,142 vehicles, a year - on - year increase of 2.5%. However, the vehicle sales revenue was 21.5 billion yuan, a year - on - year decrease of 12.7%. It is worth noting that the pure - electric model i6, with a starting price of 249,800 yuan, had a cumulative delivery of about 49,000 units in the first quarter, accounting for more than 50%. While becoming the sales pillar, it also lowered the average price per vehicle, leading to a decline in revenue. A rough estimate shows that the average revenue per vehicle in the first quarter was about 226,000 yuan, compared with about 266,000 yuan in the same period last year.

Second, there were one - time cost expenditures. Li Auto bore the difference in purchase tax for i6 owners whose deliveries were postponed to the next year due to unexpectedly high orders, with an amount exceeding 500 million yuan. At the same time, it launched replacement compensation benefits for old owners. The combination of these two expenses led to an operating loss of 3 billion yuan in the first quarter, compared with an operating profit of 272 million yuan in the same period last year.

The market that Li Auto was most familiar with was also changing. Under the triple positioning of the 300,000 - 500,000 - yuan price range, family - use vehicles, and extended - range technology, Li Auto once had no direct competitors. In 2023, Li Auto delivered 376,000 vehicles and had a net profit of 11.81 billion yuan. In 2024, the net profit declined to 8 billion yuan, but the delivery reached a peak of 500,500 vehicles.

In 2025, the changes continued to intensify.

According to industry statistics from the China Automobile Dealers Association and others, the number of domestic extended - range models increased from 17 in 2023 to 44 in 2025. "Refrigerator, color TV, and big sofa" are no longer exclusive labels for Li Auto but have become standard features for models priced over 300,000 yuan. In 2025, Li Auto delivered 406,000 vehicles, a year - on - year decrease of 18.8%. The revenue was 112.3 billion yuan, a year - on - year decrease of 22.3%. The net profit plummeted to 1.1 billion yuan, a decline of 85.8%. It is worth noting that this 1.1 - billion - yuan net profit did not come from the main business but mainly relied on 1.92 billion yuan in interest income and investment returns.

When the track becomes crowded, finding a new moat has become a real challenge. Li Xiang has emphasized on multiple occasions that Li Auto is not an automotive company but an artificial intelligence company. At the Q1 2026 earnings conference, he further stated: "In the next 3 - 5 years, the competition in the mid - to - high - end intelligent vehicle market is essentially a competition in embodied intelligence."

For Li Auto, the new walls must be built before the old walls lose their support. Li Auto is also working hard for this.

Li Auto is consciously reducing marketing expenditures and tilting resources towards technology R & D. In the first quarter, the sales and management expenses decreased by 19% year - on - year to 2 billion yuan, while the R & D expenses were 2.7 billion yuan, a year - on - year increase of 8.3%. The annual R & D investment is expected to be 12 billion yuan, with 50% focused on core AI technologies.

As of the end of the first quarter, Li Auto had a cash reserve of 94.3 billion yuan, only 11 billion yuan in interest - bearing liabilities, and a net cash of 83.3 billion yuan. The asset - liability ratio was 51.2%.

The pain of transformation

Li Auto's "new wall" consists of three levels: self - developed chips, large AI models, and embodied intelligence.

The most core hardware is the self - developed chip "Mach M100". Xie Yan, the CTO of Li Auto, revealed at the Q1 2026 earnings conference that the single - unit material cost of the Mach M100 is lower than that of the outsourced chip solution, and its effective computing power is about three times that of NVIDIA's Thor U. The new - generation L9 Livis version equipped with dual chips has a total computing power of 2560 TOPS.

The logic of self - developed chips is to build a systematic barrier. Xie Yan said that when everyone uses NVIDIA chips, information flows too freely, and a company can catch up in terms of performance by poaching a few people. After full - stack self - development and vertical integration from chips to large models, the logic of "catching up by poaching people" no longer holds. Li Xiang added at the meeting, "It took us four years from project initiation to chip mass production, and this time difference itself is a barrier."

At the software level, the Mach VLA large model uses a 3D ViT perception architecture. The forward perception distance has been increased to 300 meters, and the computing volume has increased by 10 times compared with the previous generation. Li Auto said that the ADAS 9.0 version equipped with the Mach M100 chip makes more stable decisions in complex scenarios than the 8.0 version.

At the organizational level, according to public reports, Li Auto's base model department has been adjusted. Three secondary departments related to embodied intelligence, namely embodied engineering, embodied interaction, and embodied behavior, have been added, and the autonomous driving department has been adjusted to an independent secondary department. The humanoid robot project has been launched for R & D. Li Xiang said: "Autonomous driving is the first half of the robotics story, while humanoid robots are the second half." Regarding the commercialization timeline, he said it would take at least three more years.

It cannot be ignored that the process of "building a new wall" faces multiple challenges.

First, there is the pace of chip installation in vehicles. Currently, the Mach M100 is only installed in the new L9 model, and there is no clear timeline for when it will be extended to the L8, L7, and i - series models. If the installation pace is slower than expected, "leading in effective computing power" will only be an empty talk and cannot be translated into a scale advantage.

Second, there is the technical maturity. Li Auto said that it once took two months to update the first - version model of the Mach M100 and adapt it to the vehicle end, although it has been shortened to less than a week, it is still far from the goal of "completing it within a day".

Third, there is the R & D pay - back period. From 2024 to 2026, Li Auto's cumulative R & D investment was about 34.4 billion yuan. Although the Mach M100 has been mass - produced, it is still uncertain whether the hundreds of billions in R & D investment can be recovered within 3 - 5 years.

Fourth, there is the follow - up speed of competitors. XPeng's self - developed Turing AI chip was mass - produced and installed in vehicles in Q2 2025. NIO's Shenji NX9031 chip was also mass - produced in 2025. Tesla's Optimus entered the Gen3 mass - production stage in 2026. When everyone realizes that "embodied intelligence is the next battlefield", is Li Auto's report card convincing enough?

More urgently, there is the transformation to pure - electric vehicles. In March 2024, Li Auto's first pure - electric MPV, the MEGA, was launched, and its monthly sales were far below the expected target of 8,000 units. The new pure - electric i6 delivered 49,000 units in the first quarter, but it still needs a significant increase in production throughout the year. In the second half of the year, the i9 will enter the higher - end pure - electric market and face competition from NIO, XPeng, and Wenjie.

After the new L9 was launched in mid - May, the market feedback was also divided. Li Xiang called the L9 "the first work of an embodied intelligent robot". Official data shows that the model received over 10,000 orders two weeks after its launch, with the Livis version priced at 509,800 yuan accounting for over 90%. However, Citigroup expects the new L9 to have an average monthly sales volume of about 5,000 units and maintains a "neutral" rating.

The capital market also did not buy into Li Auto's AI narrative. After the press conference, Li Auto's stock price fluctuated, with a cumulative decline of nearly 20% in three Hong Kong stock trading days.

Li Tie gave guidance at a conference call: It is expected that the gross profit margin will recover to over 10% in the second quarter and gradually increase throughout the year as the product portfolio improves.

The ramp - up speed of new products and the progress of gross profit margin repair will determine whether Li Auto can smoothly transition through this transformation period.

This article is from the WeChat official account "Chinese Entrepreneur Magazine" (ID: iceo - com - cn), author: Shi Siyu, published by 36Kr with authorization.