Gasgoo Munich - "Just do your job."
The phrase has defined Li Auto’s state since the start of 2024: a long to-do list and a frantic pace.
At the start of 2026, two seemingly subtle decisions thrust Li Auto into the media spotlight.
On one side: contraction and pragmatism. Multiple outlets reported that “Li Auto plans to close about 100 inefficient stores in the first half of 2026,” while emphasizing dynamic adjustments based on actual operations.
When reached for comment, Li Auto told Gasgoo that the reports are “basically accurate,” adding that the company has no unified official response regarding the reasons yet.
On the other side: radical leaps. Almost simultaneously, an impromptu all-hands meeting called by CEO Li Xiang lasted nearly two hours, swinging Li Auto’s strategic compass toward a grander, more unknown frontier—artificial intelligence and embodied intelligence.
On one hand, the realism of a traditional auto business “slimming down” by closing stores; on the other, the idealism of betting the future on an “All-in AI” strategy.
The sharp contrast between these two scenes perfectly sketches the complex predicament and strategic choices Li Auto faces at the critical crossroads of 2026. It is attempting to shed its successful “dad car” label and undergo a complete self-reinvention.
As store closures and an AI leap happen in tandem, Li Xiang attempted to bridge an internal cognitive gap with a company-wide speech—only to expose the deep anxiety of this EV giant at the crossroads.
A Disconnect from the Top: Is AI Suffering ‘Acclimatization Issues’ at Li Auto?
“I don’t understand what the boss is saying.”
Following the internal all-hands meeting on January 26—called without prior notice by Li Xiang—many Li Auto employees took to the company’s internal social media platform to express, almost in unison, that they “didn’t understand” or questioned the significance of the gathering.
The meeting stretched on for nearly two hours. For the most part, Li Xiang shared his views on AI trends and stressed several critical timelines related to the technology.
He noted that 2026 will be the final opportunity for any company hoping to become a top-tier AI player; by 2028 at the latest, Level 4 autonomous driving technology will certainly become a reality. Globally, he argued, the number of companies simultaneously developing foundation models, chips, embodied intelligence, and operating systems will not exceed three—and Li Auto aims to be one of them.
Looking ahead, Li Auto will not only reinforce its brand positioning as a “mobile home” but also expand further into embodied intelligence.
Beyond these points, Li Xiang announced that Li Auto will enter the humanoid robotics sector and plans to launch related products soon. He stated the company will actively recruit top talent, even considering rehiring former employees who had left for other robotics startups.
To achieve this, Li Auto will restructure its R&D teams into divisions such as the foundation model team, software entity team, and hardware entity team. Both automotive and robotics projects will be managed under the hardware entity umbrella.
Even so, the bulk of the all-hands meeting centered on Li Xiang’s personal interpretation of AI development trends.
Image Source: Li Auto
Li Xiang certainly showed his “true colors,” but many of the execution team down below remain completely baffled.
Why were employees “stunned”?
This all-hands meeting—deemed “incomprehensible” by staff—acts as a mirror, reflecting the cognitive gap within Li Auto during its strategic pivot.
“Essentially, that meeting wasn’t about short-term goals. It was about long-term vision—which is why it’s called a vision, not a plan, and had no specific targets,” analyzed Sun Shaojun, founder of Carfans. “The confusion stems from the fact that it had no direct bearing on the short-term interests of most employees.”
The facts bear this out.
One industry insider offered a further analysis: An engineer tuning the chassis for the L9, now classified under the “hardware entity team” in the new structure, might realize the company’s strategic center of gravity is shifting toward distant foundation models and robotics. The automotive business he works on becomes just one option among many “hardware entities.”
A software engineer developing city NOA features, assigned to the “software entity team,” would now have to collaborate with a brand-new “foundation model team.” The output of that team will directly determine what functions he can implement. Cross-team communication costs, friction over technical interfaces, and the uncontrollable nature of the underlying model’s “black box” could all become new obstacles in his daily work.
For a long time, Li Auto fostered an execution culture oriented toward short-term sales and delivery targets. This culture was incredibly efficient during the expansion phase, helping Li Auto quickly rise to the top tier of new automakers.
“Li Auto’s team has always been biased toward short-term goals,” Sun Shaojun pointed out. So when Li Xiang suddenly started talking about long-term goals, employees’ first reaction was that it didn’t fit his usual “pragmatic” persona.
That “pragmatic” persona stems from a path dependence on past successes.
Image Source: Li Auto
There is truth to that. Unlike other new forces that traded massive losses for growth in their early years, Li Auto was the fastest to achieve quarterly profitability, thanks to its meticulous cost control and precise grasp of market demand.
Even years ago, when XPENG and NIO poured vast sums into smart driving R&D and charging infrastructure, many in the industry gave Li Auto a thumbs-up: making money first and surviving was the most practical approach.
Today, looking at the trio—NIO, XPENG, and Li Auto—none have emerged from the auto market completely unscathed. Yet, relatively speaking, Li Auto, which was so single-mindedly “selling cars to make money” in the early days, is now the one among the “three brothers” worrying the least about anything outside its core business of building cars.
Who would have thought that being “too focused on the job” would become one of the dilemmas now constraining Li Auto.
What Threw Li Auto Off Rhythm?
“If this vision were being discussed by XPENG or NIO, their employees might understand it better,” Sun Shaojun stated bluntly.
Ultimately, it comes down to the fact that Li Auto’s foray into AI has been too shallow—and perhaps too late. But that doesn’t imply Li Xiang himself is short-sighted or arrived “late to the party” regarding AI layout.
At the end of 2024, Li Xiang used multiple interviews to signal Li Auto’s pivot to AI, at which time news about the company’s VLA (Vision-Language-Action) large model was everywhere.
Notably, during the 2024 Li Auto AI Talk, Li Xiang stated that the company would definitely pursue embodied intelligence, but “the timing isn’t right now”: If Level 4 autonomous vehicles haven’t been solved yet, how can we tackle something more complex?
Yet, a mere year later, embodied intelligence has been moved right onto the meeting agenda.
Sun Shaojun believes that, like most new automakers, Li Xiang likely harbored the “human-car-home” ecosystem concept early on. It was just that too many variables along the way disrupted the original rhythm.
In the original roadmap, Li Xiang likely expected to devote more energy to frontier fields like AI and robotics after the i-series gained a foothold. One could even say that had the i-series succeeded, its robotics projects might have made a high-profile debut long ago.
But reality intervened. Two models failed to win sufficient market recognition, forcing the company to pull its strategic center of gravity back to the auto business itself. This forced “callback” disrupted the established rhythm, compelling Li Xiang to invest significant time deep in the trenches of supply chain, R&D, and sales to ensure the success of upcoming models.
This diversion of attention undoubtedly delayed the advancement of its long-term AI layout.
Image Source: Li Auto
Sun Shaojun argues that the root of Li Auto’s current challenge isn’t simply cost or technology, but the stumble of a critical link in its product matrix—the setback of the i-series, particularly the market performance of the Mega and i8 falling short of expectations. This failure directly hammered the company’s overall profitability, and a decline in profit often amplifies and aggravates underlying internal conflicts.
In fact, if the Mega and i8 had successfully continued Li Auto’s usual “hit-maker logic,” the public and operational pressure the company faces today would not be so immense. Li Auto’s current predicament is less an active “chase” or “gamble” than a passive adjustment forced by product failures.
Time is running short for Li Auto to prove itself again on the automotive product line. For now, it has no choice but to stockpile enough momentum and space for its next expedition into AI and robotics.
Build Its Own ‘Ecosystem’? Or Become Just a ‘Member’?
Since the start of 2025, AI has undoubtedly been a buzzword for most enterprises. When companies across every sector talk about AI, what exactly are they talking about? For automakers, what grander narrative lies behind their AI layout?
Sun Shaojun put it succinctly: The AI ecosystem.
“Either I build an ecosystem independently, or I simply join one.”
Right now, Li Auto is undoubtedly aiming for the former.
Image Source: Li Auto
For the automotive industry, AI is not an option but a profound transformation of the very form of future survival.
The endpoint of this transformation aligns with the fundamental logic behind tech giants like Huawei and Xiaomi, which are going all out to build a “human-car-home” full-scenario ecosystem. They control the initiation and fulfillment loop of user demand through hardware carriers like phones, cars, and smart home devices.
Sun Shaojun illustrated the scenario: “In the morning, you leave home and your car senses your presence and unlocks. You state your destination, and the car drives itself there while your home automatically initiates cleaning mode. On the way, you order breakfast, and the car routes through a pickup point. Upon arrival at the office, the car parks itself. This seamless experience relies on an AI ecosystem that connects every scenario.”
Without participating in or leading such an ecosystem, a standalone automaker risks being marginalized in the future.
Therefore, for new forces like “NIO, XPENG, and Li Auto,” the choice is stark: either build an open platform to attract others to your ecosystem like Huawei, or build your own closed-loop ecosystem like Xiaomi. Doing neither makes it difficult to establish a lasting foothold.
This also explains why XPENG is developing robots and NIO is building phones—they are all struggling to expand the boundaries of their own ecosystems.
For Li Auto, the pressure stems precisely from this.
Image Source: Li Auto
It isn’t a sudden U-turn or a blind gamble on AI; in its strategic blueprint, building an AI ecosystem was always a planned step. However, the acceleration of two external variables disrupted its original rhythm: first, the unexpected failure of its i-series models forced the company to divert massive energy back to stabilizing its auto business foundation; second, competitors like Huawei and Xiaomi are advancing in AI and ecosystem integration far faster than industry expectations, moving up the timeline for full-scenario intelligence competition.
What Li Auto is doing now is “catching up” and “accelerating” within a tight time window, hoping to secure an autonomous seat for itself in the multiple-choice question of “build an ecosystem or join one.”
This competition is about far more than just building cars.
At the end of the interview, Sun Shaojun offered an open-ended prediction: “A third player will definitely enter the fray.” The market needs a disruptor; users need a new choice.
However, Gasgoo believes that as Li Xiang attempts to unify thinking with a speech, employees on the floor are thinking about the vehicles to be delivered tomorrow and sales targets for next quarter. Whether this top-down strategic leap succeeds depends not only on whether Li Auto can precisely convert its cash reserves into ecosystem competitiveness, but even more on whether it can bridge the palpable cognitive gap within its ranks.
At this moment, Li Auto is like a long-distance runner who suddenly realizes he must leave his comfort zone, only to find his opponent has switched to a jet aircraft.
Its AI story is not a relaxed tech narrative, but a race of necessity.
