The thesis
Of the wave of Chinese EV startups that launched alongside NIO and XPeng in the mid-2010s, Li Auto is the one that solved the unit-economics problem first. It did so by refusing to bet everything on a pure battery-electric car: its early lineup ran on extended-range electric vehicles (EREVs), which pair a large battery and electric drivetrain with a small gasoline engine that only generates electricity and never turns the wheels. That hedge against China's uneven charging infrastructure let Li Auto sidestep the range anxiety that slowed rivals, and in 2023 it became the first Chinese EV startup to post a full-year profit, built on family SUVs marketed under the line 'create a mobile home, create happiness.'
The last eighteen months have complicated that clean profitability story, even as the underlying scale keeps growing. Li Auto has delivered more than 1.76 million vehicles since 2019 and holds roughly RMB100 billion (about $14 billion) in cash, but its 2025-2026 push into pure battery-electric SUVs has hurt deliveries and margins, and the company posted a net loss in the first quarter of 2026. Founder and CEO Li Xiang has responded by declaring 2026 the last window for Li Auto to become a top-tier AI company, pivoting a meaningful share of R&D toward an internal humanoid-robotics program even as the core car business fights to stabilize.
From an auto-media founder to CHJ Automotive
Li Xiang was not a car engineer before he was a carmaker. He built his reputation running Autohome, one of China's largest automotive information and classifieds sites, which gave him an unusually granular view of what Chinese buyers actually wanted from a vehicle before he ever designed one. He founded the company in Beijing in July 2015 under the name CHJ Automotive (later renamed Li Auto), with a thesis shaped directly by that experience: family buyers in China's dense, charging-infrastructure-poor cities wanted the interior space and smart features of an electric car without the range anxiety, and nobody was building it.
The first product, the Li ONE, embodied that bet exactly. It was a six-seat, extended-range SUV that began deliveries in December 2019, validating the EREV approach at a moment when most of the industry treated range-extended hybrids as a stopgap on the way to "real" EVs. Li Auto listed American depositary shares on Nasdaq in July 2020, raising about $1.1 billion, and followed with a dual-primary listing on the Hong Kong exchange in August 2021 that raised roughly $1.5 billion, giving it capital and public-market discipline earlier than most of its EV-startup peers.
The EREV machine, and the electric line built on top of it
The L-series SUVs (L6, L7, L8, L9) turned the EREV formula into a genuine volume business: by mid-2026 Li Auto had delivered more than 1.76 million vehicles cumulatively and built out a proprietary supercharging network of more than 4,100 stations and nearly 23,000 stalls to support the newer battery-electric side of the lineup. The company relaunched its flagship SUVs through 2026 to defend that volume: an all-new L9 in May, an all-new L8 in June, and a revamped L6 in July, each aimed at reasserting the EREV line against intensifying range-extended competition from BYD, Xiaomi, and others.
The harder chapter has been Li Auto's move into pure battery-electric vehicles. Its first BEV, the MEGA MPV launched in March 2024, badly missed sales targets and forced early price cuts. The company regrouped with the i8 six-seat SUV (July 2025) and the i6 five-seat SUV (September 2025), and the i6 in particular has become a real volume seller, but the transition period dragged on margins through 2025 and into 2026 even as it broadened Li Auto's addressable market beyond EREV buyers.
Betting on in-house silicon: MindVLA and the M100 chip
Where most Chinese automakers buy their autonomous-driving compute from Nvidia, Li Auto spent roughly four years building its own: the Mach M100, a 5-nanometer chip the company says delivers 1,280 trillion operations per second per chip, with two chips paired for 2,560 TOPS of combined compute in its top-end vehicles. The M100 was designed alongside Li Auto's own AI software, MindVLA (Vision-Language-Action), an autonomous-driving architecture the company unveiled in March 2025 and describes as a step toward genuinely autonomous, map-independent driving rather than incremental improvements to lane-keeping and adaptive cruise.
The all-new L9 'Livis' that launched in May 2026 was the first vehicle to ship with both the M100 chip and the MindVLA stack running together, and Li Auto says commercial-scale deployment across more of the lineup is underway through 2026. The strategic logic mirrors what Tesla and a handful of Chinese rivals have concluded: owning the compute stack, not just licensing it, is increasingly treated as a prerequisite for competing on autonomy rather than an optional differentiator.
Direct sales, family branding, and the moat
Li Auto sells almost exclusively through its own retail stores rather than independent dealers, which lets it control pricing and the in-store experience the way Tesla and NIO do, and it has pushed that network aggressively into China's smaller Tier 3 and Tier 4 cities to capture demand for premium family vehicles outside the saturated first-tier markets. Combined with a narrow early product line (one platform, one powertrain concept) that let it reach profitability faster than better-funded EV-startup peers, the direct model has supported strong owner satisfaction and referral-driven sales, which in turn has helped defend average selling prices against the brutal price war among Chinese EV makers.
The moat is real but narrower than it looks: EREV technology is no longer proprietary (BYD, Xiaomi, and others now sell range-extended competitors directly against the L-series), so Li Auto's durable advantages are increasingly the brand's family positioning, the supercharging and retail footprint, and the in-house AI stack rather than the powertrain concept that built the company.
From record profits to a 2026 loss
Li Auto's financial arc has been unusually volatile for a company its size. It delivered a record 500,508 vehicles in 2024, but full-year 2025 deliveries fell about 18.8% to 406,343 as EREV competition intensified and the BEV transition dragged, and net income collapsed 85.8% to RMB1.1 billion on revenue of RMB112.3 billion, down from RMB144.5 billion in 2024. The slide continued into 2026: first-quarter revenue fell 11.4% year over year to roughly RMB23 billion (about $3.33 billion), with a net loss of RMB2.3 billion and gross margin collapsing to 7.9% from 20.5% a year earlier, as the i6 ramp and a model-refresh cycle weighed on pricing power.
Management responded by cutting full-year 2026 delivery-growth guidance from about 40% (roughly 550,000 units) to about 20% (roughly 490,000 units). Second-quarter 2026 deliveries came in at 98,330 units, down 11.5% year over year but at the upper half of Li Auto's guided range, and the balance sheet remains a genuine strength: roughly RMB94 billion in cash at the end of the first quarter, among the strongest of any Chinese EV maker. Full second-quarter financial results are due August 26, 2026.
What to watch: the i9, the robotics pivot, and Q2 earnings
Three threads will determine whether 2026 is a trough or a turning point. First, the i9, Li Auto's largest and first all-electric flagship SUV, is set to launch in September 2026 with an 800-volt platform, 5C charging, and a dual-motor, 400-kilowatt system; it needs to succeed where the MEGA stumbled if the BEV line is to carry real weight in the lineup. Second, Li Auto opened its first overseas production line in July 2026, assembling the L6, L8, and L9 at partner Allur's plant in Kostanay, Kazakhstan, alongside new dealer partnerships in the UAE and Saudi Arabia, its first serious push into international markets.
Third, and most unusual for an automaker, is the robotics pivot. In May 2026 Li Auto reorganized its AI unit into an independent autonomous-driving department plus three new embodied-AI teams, formalizing an internal humanoid- and wheeled-robotics program code-named Nexus that had been developing in secret for roughly a year. Li Xiang has framed 2026 as the last realistic window for any company to simultaneously master foundation models, chips, and embodied AI, and has committed more than half of R&D spending to AI-related work. Whether that bet pays off, and whether the core car business can recover its margins in the meantime (management has guided to roughly 10% gross margin in the second quarter), are the two questions that will define Li Auto's next few years.
Sources
- Li Auto - FY2025 and Q4 2025 unaudited financial results (SEC 6-K)
- StockTitan - Li Auto Q1 2026 results (RMB2.3B loss, 7.9% gross margin)
- CnEVPost - Li Auto reshuffles teams to deepen AI and robotics push
- CnEVPost - Li Auto teases i9 flagship SUV ahead of September launch
- CnEVPost - Li Auto launches all-new L9 with MindVLA and M100 chip
- CnEVPost - Li Auto begins first overseas production in Kazakhstan