Original content by the editorial team of Keji Sishao © Youliao Business

Author | Si Shao

In the last quarter of 2024, Li Xiang stood in a position that made his peers envious.

At that time, Li Auto's annual deliveries had just surpassed 500,000 units, securing its position as the top-selling new energy vehicle (NEV) brand. With a net profit of RMB 8.05 billion and a gross margin of 20.5%—these two figures, in the NEV sector where burning cash was as common as drinking water, seemed like results only achievable by a different kind of player. NIO was posting consecutive losses, XPeng was just climbing out of the abyss, and Xiaomi was still anxious about delivery volumes. Li Auto, however, had been labeled by the entire industry as the only player among the new forces to have truly made money.

This was no stroke of luck.

When extended-range vehicles entered the market, they were almost unanimously mocked as "Frankenstein's monsters of oil-to-electric conversions." But Li Xiang identified a real pain point for users: range anxiety. At the time, charging infrastructure was sparse, and highway range remained a mystery. The logic of "refuel to keep going, use electricity to save money" was nearly irrefutable for family users—this was the foundation of Li Auto's success and the reason why all competitors later flocked into this segment.

At the same time, he positioned the brand almost obsessively as a "family-oriented vehicle," locking in the price band of RMB 300,000 to RMB 500,000 for family cars, carving out a relatively untouched high ground amid a price war. This strategy proved effective during the L series era but also sowed a hidden risk: when more and more models relied on the same product logic to cover different price segments, could this high ground still be defended?

Everything seemed to be on the right track.

No one anticipated that a year later, Li Auto's wheels would come to such a sudden halt.

In a single year, the sky fell

In 2025, Li Auto's total revenue was RMB 112.3 billion, down 22.3% year-on-year.

Net profit was RMB 1.14 billion, a staggering 86% drop from RMB 8 billion in 2024.

Quarterly breakdowns revealed an especially tough second half: a RMB 625 million loss in Q3 and a meager RMB 7 million profit in Q4, a 99% year-on-year decline.

When these two figures are placed side by side, no explanation is needed. From a RMB 2.8 billion profit to a RMB 600 million loss, in the same company, the same quarter, twelve months apart. This was not a cyclical fluctuation—it was a cliff dive.

To understand this collapse, we must rewind to early 2025.

Li Auto's foundation was extended-range technology.

But in 2025, this foundation began to crumble—not because extended-range technology became outdated, but because everyone understood its logic and rushed in. AITO, Leapmotor, Seres, and Voyah entered the fray, with Huawei-backed HarmonyOS Intelligent Driving being particularly aggressive, using "far ahead" intelligentization (intelligent) narratives to breach the family user base that Li Auto had once defended.

In the first half of 2025, Li Auto's extended-range models saw consecutive month-on-month declines. The first-mover advantage dissipated much faster than expected once enough followers entered the market.

If the loss of the extended-range segment came from external forces, the failure in pure electric layout (layout) was more like a self-inflicted wound.

In March 2024, the Li Auto MEGA was launched, priced at RMB 559,800, targeting the high-end pure electric MPV segment. However, its debut turned into a PR disaster—the body design was widely ridiculed by the market, with Topic heat (trending topics) overshadowing product strength, and monthly sales fell far short of expectations.

More critically, the MEGA's failure disrupted the entire pure electric product lineup, causing the subsequent i series to fall behind in timing. The July 2025 launch of the i8 faced near-simultaneous competition from the Leapmotor L90, AITO M8 Pure Electric, and Tesla Model YL, forcing an emergency configuration adjustment just one week after launch—an action that spoke volumes.

The first two blows came from external forces. The third, and most fatal, was an internal conflict.

Li Auto's "copy-paste strategy" was logically sound: use similar design languages to quickly cover different price segments, highly efficient when brand momentum was strong. During the L series era, this approach was nearly flawless. But the arrival of the i6 changed that.

After the i6 entered the market at a lower price with more aggressive configurations, multiple frontline sales representatives revealed that 80% of in-store consumers only looked at the i6, ignoring the i8. Users who had originally locked in the i8 quietly switched their orders overnight.

This was not about two models competing for external market share—it was about one new model prematurely harvesting all potential users of another. Li Auto's sales teams stood in showrooms, facing a bizarre dilemma: the company's newer models were selling better, but overall revenue was collapsing.

The high-end i8 struggled to sell, while the L series' core market was also disrupted. A product strategy originally designed for expansion turned into a war of attrition (war of attrition) where left and right hands fought each other.

This internal conflict was harder to quantify in financial reports than external competition—it seeped into pricing systems, sales morale, and product planning. By the time decision-makers realized the problem, the wounds were already deep.

With these three blows combined, full-year deliveries in 2025 fell to 406,300 units, down 18.8% year-on-year. Even more glaring was the ranking: from the undisputed champion among new forces in 2024 to fifth place in 2025, surpassed by Leapmotor, HarmonyOS Intelligent Driving, XPeng, and Xiaomi in sequence.

Full-year net profit was RMB 1.14 billion, an 85.8% year-on-year plunge. If we strip away this figure—the actual operating loss was RMB 521 million, with the only support for "profit" coming from interest and investment gains.

The label of "the only player among new forces to have truly made money" was quietly torn off.

Li Xiang's reckoning

To understand Li Auto's 2025, we cannot ignore Li Xiang himself.

More precisely, we cannot ignore a decision he made in 2022.

In 2022, during Li Auto's high-growth phase, Li Xiang did something that seemed visionary at the time: systematically introducing Huawei's IPD system (Integrated Product Development). This management methodology, one of Huawei's core assets, allowed Huawei to maintain efficient collaboration across its highly complex product matrix.

Li Xiang's logic was clear—to expand from one model to multiple, from one segment to multiple, Li Auto needed a management framework capable of supporting scale. Huawei was the textbook example.

But the problem was that Huawei's IPD was designed for a tech giant with over 100,000 employees. Its core logic was to "componentize" people—R&D, sales, and marketing personnel were disassembled and reorganize (reorganized) into project teams for different product lines, using standardized processes and reporting chains to ensure collaboration. This system worked well at Huawei's scale because it had a thick enough management layer to absorb process friction.

Li Auto, at that stage, could not digest this friction.

What actually happened was that "walls" between project teams grew higher, different product lines began operating independently, decision chains lengthened, and internal communication costs skyrocketed. A NEV company that needed "weekly" market responsiveness was dressed in a management suit designed for stability. Speed was quietly lost in the process.

In July 2025, Li Auto announced it would abandon Huawei's PBC (Personal Business Commitment) performance model and reinstate OKR. This decision itself was a public act of self-denial.

Li Xiang has a well-known trait in the industry: he is willing to admit mistakes publicly.

From early product decisions at Autohome to iterations of the Li ONE, he has left behind extensive, candid retrospectives on Weibo and internal memos. This transparency is rare among founders and has earned him significant trust capital. Many therefore believe Li Xiang is someone who "listens."

But what happened in 2025 revealed another side: his reflections always came half a beat late.

In contrast, He Xiaopeng and Li Bin acted faster. After XPeng hit rock bottom in 2022, it adjusted nearly every quarter—changing management, cutting product lines, bringing in Wang Fengying—with such speed and magnitude that outsiders worried the company might lose direction in panic. Under NIO's years of consecutive losses, Li Bin maintained high-frequency strategic iterations. The point here is not whether their decisions were correct, but that their response to "changing conditions" was significantly faster than Li Xiang's. Both shared a trait: they started making cuts before things fully deteriorated.

Li Xiang was not like this.

In early 2025, Li Auto set an annual delivery target of 700,000 units. By mid-year, market signals were clear enough that the target was lowered to 640,000. Even then, only 406,300 units were delivered—not even 60% of the revised goal. From 700,000 to 400,000, this was not a judgment error—it was a systemic cognitive delay.

Even more telling was the layoffs. In mid-2025, Li Auto underwent a significant round of staff optimizations, intended to cut costs and improve efficiency. But both the execution method and timing faced internal criticism: the damage to morale outweighed the cost savings. This decision later appeared on Li Xiang's own list of reflections.

"The worst version of myself"

At the Q3 2025 earnings call, Li Xiang said something that would be quoted repeatedly:

Li Xiang said his performance over the past few years represented "the worst version of myself."

He reflected during the earnings call that his biggest mistake was managing a direct-sales system with dealer-like approaches, where the professional manager model lengthened decision chains and slowed iterations, failing to adapt to current competitive rhythms.

This was a rare, extremely direct public admission of error. He announced a full return to a startup mode from Q4 onward, replacing reporting with deep dialogue and task delivery with user value.

But one detail is worth noting: these words were only spoken after the company shifted from profit to loss.

Admitting mistakes is not the issue. The issue is that a person who had correctly bet on extended-range technology spent three years eroding his own advantages with a flawed management model and a series of misguided product decisions—then stood up in the worst quarter to say, "I know where I went wrong."

This was not failure—it was something more complex: the faster a car goes, the harder it is to notice the steering wheel has veered.

After the admission

By the end of 2025, Li Auto still had RMB 101.2 billion in cash on hand.

This was the thickest war chest among all Chinese new forces.

In a sense, this was also Li Auto's and Li Xiang's last bit of confidence—no matter how ugly the year had been, they had not yet run out of ammunition. Q4 deliveries rebounded to 109,000 units quarter-on-quarter, with net profit barely turning positive at RMB 20.2 million, like taking a breath after surfacing from deep water.

Li Xiang set a 2026 target of over 20% year-on-year sales growth, to around 480,000 units. The core pillar is the all-new L9 launching in Q2—equipped with Li Auto's self-developed Mach M100 chip and full wire-controlled chassis, this will be Li Auto's first truly proactive strike in the intelligentization (intelligent) arena, rather than following others. In 2025, Li Auto invested RMB 11.3 billion in R&D, half of which went to AI. The 2026 R&D budget is RMB 12 billion, maintaining the same proportion.

The direction is clear, but the numbers are equally clear: Q1 2026 delivery guidance is only 85,000 to 90,000 units. This means that to hit the full-year target of 480,000 units, the remaining three quarters must rebound at nearly double the pace. This curve is steep enough to make one nervous.

Notably, one thing Li Auto did not lose in this crisis was user reputation. L series owner satisfaction remains among the highest in its price segment, and extended-range technology itself has not collapsed. This may be the most critical asset, harder to replicate than the RMB 101.2 billion in cash. A company can lose market share in a year or two, but losing user trust often takes much longer to repair. Li Auto has not reached that point yet.

But the battle is far from over.

The 2025 NEV battlefield is a "weekly competition"—product iteration speeds and intelligent capability evolution have become so fast that no company can afford to breathe. Huawei's ecosystem barrier (barriers) in smart cockpits and autonomous driving, Xiaomi's brand penetration (penetration) through internet playbooks, Leapmotor's breakthroughs in extreme cost-effectiveness for lower-tier markets—every competitor is advancing in trenches using methods Li Auto once knew well.

Li Xiang said, "Return to startup mode." This is easy to say but hard to do. Startup mode means the founder must dive back into products and users, make decisions fast enough to leave professional managers behind, and scrape away the inertia built up over three years of processes and reporting.

This is much harder than admitting mistakes.

The launch of the new L9 in Q2 2026 will be a key moment to watch. It is not just a model release—it is Li Xiang's first answer sheet to prove that "the worst version of myself" is in the past.

The answer, however, is still on the road.

Disclaimer: This article is based on publicly available information or details provided by interviewees. However, Keji Sishao and the author do not guarantee the completeness or accuracy of the information or associations mentioned or displayed in this article, nor do they represent the positions of any institutions or individuals. In case of infringement, please contact for removal. Under no circumstances shall the information or opinions expressed in this article constitute investment advice to any person.