LI (Trans): GPM 9.4%; cost inflation not passed on to users

Below is Dolphin Research’s compiled $Li Auto(LI.US) FY26 Q2 earnings call Trans

I. Key takeaways

1) Guidance: Q3 deliveries are guided at 95k–100k units, with total revenue of RMB 26.6bn–28.0bn. The company expects FY CapEx at approx. RMB 6.0bn, including the ultra-fast charging network. Management said full-year cash flow should beat last year, while turning operating and free cash flow positive will hinge mainly on Q4 deliveries.

2) Shareholder returns: To date, the company has repurchased 91.7 mn Class A ordinary shares, including 23.7 mn ADS, for a total consideration of about $631.5 mn. This reflects ongoing commitment to capital return and balance sheet discipline.

3) Key financial metrics this quarter

a) Revenue: total revenue was RMB 25.7bn (+11.7% QoQ, -15.1% YoY). Vehicle revenue was RMB 24.1bn (+11.8% QoQ, -15.7% YoY), with the YoY decline driven by lower deliveries and mix-led ASP pressure, and the QoQ rebound driven by higher ASP and deliveries.

b) Gross profit: GP was RMB 2.8bn (+56.9% QoQ, -53.3% YoY). Vehicle GPM was 9.4% vs. 19.4% a year ago and 6.1% last quarter; overall GPM was 11.0% vs. 20.1% a year ago and 7.9% last quarter. Mix shift was the primary driver of both YoY and QoQ changes.

c) Opex: operating expenses were RMB 5.1bn (-2.0% YoY, +6.9% QoQ). R&D was RMB 2.8bn (-1.2% YoY), while SG&A was RMB 2.3bn (-16.2% YoY on lower staff comp, +11.2% QoQ on higher marketing spend).

d) Profit: operating loss was RMB 2.3bn, vs. operating profit of RMB 827 mn a year ago and a loss of RMB 3.0bn last quarter. OPM was -9.0% vs. 2.7% a year ago and -13% last quarter.

e) Net loss was RMB 1.7bn, vs. net profit of RMB 1.1bn a year ago and a loss of RMB 2.3bn last quarter. Diluted net loss per ADS was RMB 1.69, vs. a profit of RMB 1.03 a year ago and a loss of RMB 2.26 last quarter.

4) Cash flow and cash: operating cash flow turned to a net inflow of RMB 15 mn, vs. an outflow of RMB 3.0bn a year ago and RMB 6.1bn last quarter. Free cash flow was -RMB 1.3bn, vs. -RMB 3.8bn a year ago and -RMB 7.4bn last quarter, with period-end cash reserves of RMB 87.5bn.

II. Call details

2.1 Management remarks

1) Sales mix and market position

a) Despite intensified competition and a full model refresh in H1, Li Auto remained the top-selling Chinese brand in the >RMB 200k NEV market. This underscores its resilience in premium segments.

b) Under its dual-energy strategy, the sales mix is healthy, with EREV and BEV each at roughly 50% of total. Management expects the BEV mix to rise further as new models launch in H2.

c) The i6 ranked top-three by sales in the >RMB 200k segment for six straight months. The i6 and L6 each lead their sub-segments, reinforcing leadership in the RMB 200k–300k SUV market.

d) With new models launching and ramping in H2, management is confident of staying top-three across all brands in China’s >RMB 200k passenger car market. This confidence is tied to product cadence and execution.

2) Product and tech upgrades

a) Starting in Q2, the L series completed a full refresh. Core upgrades include the in-house MACH M100 chip running the MACH VLA model, an 800V active suspension with steer-by-wire chassis, and a third-gen range extender paired with 5C ultra-fast charging batteries.

b) At end-Jul, Li Auto launched the i8 RWD long-range variant. Based on user feedback, it added a powered frunk and zero-gravity seats for driver and front passenger, which lifted sales meaningfully.

c) The next-gen MEGA will debut on Sept 2. The all-new flagship BEV SUV i9 will launch in mid-Sept, expanding the high-end BEV lineup.

3) Battery and charging network

a) Cells, BMS, and packs are now fully in-house, completing the last block of the e-powertrain after motors and power electronics. With integrated vehicle design and 5C fast-charging know-how, management is confident its in-house batteries will lead on quality, safety, and longevity.

b) The in-house battery is already in the new L8, the new L6, and the i8. It will be rolled out across all models over the coming months to raise in-house content.

c) 5C ultra-fast charging has become a prerequisite in users’ purchase decisions. As of end-Jul, Li Auto operated 4,141 stations and over 22,800 piles, forming a 9-vertical-by-9-horizontal highway grid covering 18 national expressways and 300+ cities.

4) Chips and intelligent driving

a) Since May, Li Auto has shipped a full-stack ADAS based on the MACH M100 chip. Over 50k M100 chips have shipped to date with strong quality records, supporting scaled deployment.

b) The end-Jul OTA 9.1 lifted MACH VLA performance by 20%. User adoption of intelligent driving nearly doubled vs. the prior gen, bringing reaction speed on par with or better than human drivers, and adding Efficient and Comfort speed modes.

c) OTA cadence: in Oct, 9.2 will fully pivot to a 3D vision transformer. By year-end, 9.3 will expand VLA parameters by orders of magnitude, enhancing task understanding and reasoning in complex scenes. Faster reaction, longer perception range, and stronger reasoning are the three key upgrades this year.

d) Management sees batteries and chips as the core moats for embodied intelligence. The long-term vision is for the car to become a true intelligent agent that not only assists people but completes tasks more efficiently on its own.

2.2 Q&A

Q: The L series refresh is complete. How has the market responded since launch?

A: High-trim variants account for over 85% of L9 sales, and the new L6 is targeted to stabilize at 10k units per month. The L9, L8, and L6 have all migrated to the latest platform featuring the MACH M100 chip and 5C EREV tech, with the newest models adding a steer-by-wire chassis to fully cover EREV SUVs priced RMB 200k–500k.

High-end models performed above both user and company expectations. The L9’s high trim contributes over 85% of L9 sales, as many users pay for the steer-by-wire chassis and advanced intelligent driving, cementing leadership in the RMB 400k–500k family SUV segment.

The new-gen L6 builds on a base of nearly 400k cumulative deliveries of the prior gen. It improves BEV range, charging speed, and the smart platform, while adding dual zero-gravity front seats and a 29-inch panoramic display, with a goal of sustaining ~10k units per month.

Management acknowledged transitional disruptions from the refresh cycle. Clearing legacy inventory, new-model ramps, and policy shifts created near-term pressure, and processes are being optimized. Next, Li Auto will unlock hardware and AI features via OTA and continue to densify the 5C charging network.

Q: The new MEGA was teased today. What are the key upgrades and sales outlook?

A: Three focused upgrades address prior-gen user feedback, with no quantitative sales target disclosed. The next-gen MEGA directly addresses specific pain points from real user feedback of the prior model, with three main areas of improvement.

First, chassis and handling. Many perceive MPVs as cumbersome in cities, so the new MEGA adds rear-wheel steering, steer-by-wire, and active anti-roll bars to tighten turning radius and reduce body roll, improving urban agility.

Second, the smart platform. The in-house MACH M100 chip upgrades the full intelligent driving stack, adding lateral and rear sensors to improve city NOA, complex intersections, and auto parking, while the cabin moves to the latest Qualcomm chip for better interaction and entertainment. Third, cabin details are tailored to family MPV use, with major upgrades to rows two and three and enhanced ambient-interaction features.

MEGA is positioned as a flagship MPV priced above RMB 500k. Management said sales will depend on conversion, ramps, and market dynamics, and they offered no numeric target. They committed to strong delivery, retail experience, and user operations, and will update sales performance post launch.

Q: Can you quantify the impact of raw material price increases in Q2–Q3, and how will you offset cost inflation and margin pressure?

A: No quantified impact disclosed, with offsets from in-house e-powertrain and chips. Upstream raw materials and key components saw cyclical swings this year, adding temporary cost pressure for the industry and Li Auto and impacting margins.

AI demand lifted pricing for chips and PCBs, and memory prices climbed. Li Auto mitigated some pressure via early volume commitments and long-term procurement agreements, while lithium carbonate also fluctuated this year on the battery side.

The strategy has two prongs: drive ongoing cost-down through better operations, and build structural cost advantages through full-stack in-house tech and an owned supply chain. This dual approach aims to smooth cycles and stabilize margins.

On e-drive, Li Auto self-develops and manufactures motors, controllers, and SiC modules. An integrated architecture optimizes energy use and lowers per-vehicle hardware costs, while batteries are co-developed with the vehicle architecture for deep integration across efficiency, thermal, safety, and packaging, and are being rolled out to more models.

On chips, the in-house MACH M100 uses an innovative dataflow architecture with co-designed hardware and software, delivering structural benefits in compute and cost. Near term, volume commitments and fine-tuned operations help smooth volatility; longer term, scaled deployment of in-house tech should anchor margins.

Q: Given raw material costs and pricing competition, what is the latest gross margin target?

A: Healthy LT GPM at 15%–20%, with a clear stance not to pass cost increases to users. Battery and memory costs rose sharply this year, a common industry headwind. Li Auto’s more intelligent products consume more memory and semis and thus feel a larger impact, while toolings and line equipment depreciation also matter, with stricter treatment for discontinued projects.

As new products roll through this year, GPM is improving. However, higher chip and PCB costs must be addressed, and management explicitly decided not to pass this round of cost increases to customers, keeping pricing stable.

The offset relies on integrated design and supply chain leverage. Li Auto will deepen in-house battery R&D and rollout to raise self-sufficiency and cost control, and on the sales side, leverage the sales partner program to lower selling costs and boost efficiency, passing savings to users.

Management’s long-term view is that a healthy company-level GPM should be 15%–20%. The key variable will be raw material costs over the cycle.

Q: Can you share more on the soon-to-launch i9?

A: A flagship six-seat BEV SUV launching mid-Sept, with pricing and specs not disclosed. Positioned for large families, the i9 is a flagship six-seat SUV that continues Li Auto’s core DNA of family-centric design in the premium segment.

It complements the MEGA in the product matrix. Together, a flagship SUV and a flagship MPV will serve large families seeking BEVs, while the L series and these flagships complete coverage of the RMB 200k–500k high-end NEV market.

Technically, the i9 features an 800V 5C high-voltage charging platform and the latest in-house motors. It leverages the nationwide 5C charging network, and the MACH M100 chip supports both intelligent driving and future embodied intelligence, while the latest high-performance Qualcomm chip powers multitasking and AI agents in the cabin.

User experience focuses on family travel, optimizing interior space, comfort, and interaction for each occupant to deliver a flagship experience. Launch is mid-Sept, but per disclosure rules, pricing and detailed specs will be revealed at the product event, not beforehand.

Q: How are the M100 and the intelligent driving model co-optimized, and what are the quantified upgrade targets for H2?

A: OTA 9.2 triples parameters and lifts compute by 4.6x, with Q4 perception beyond 250m. The M100 entered mass production with the new L9 in Q2 and now covers the L9, L8, and L6, with capacity sufficient for market demand.

OTA 9.2 is a major architectural shift to a full 3D vision transformer. Parameters rise 3x and compute rises 4.6x, improving safety, comfort, efficiency, and navigation holistically for the end user.

There are three quantified Q4 goals. First, effective perception range will exceed 250m to enable earlier speed planning and pathing, reducing hard braking, hesitation, and unnecessary lane changes by over 30%.

Second, key target 3D perception accuracy will improve to within 5cm. Success rates in scenarios such as narrow-road passing and access gates are targeted to rise by 50%.

The third is scene understanding. The system will infer intent within a broader traffic context to make more decisive choices in narrow-road yielding, construction detours, and unprotected turns, cutting unnecessary stops and hesitation by over 20%.

MACH VLA 2.0 for Nvidia Orin and Thor will go live in early Sept. City-driving intelligent mileage penetration has nearly doubled, with all-scenario penetration up about 25% in recent months.

Q: Operating cash flow was near breakeven in Q2 while FCF stayed negative and cash reserves declined. Can FCF turn positive in H2, and how do you view cash reserves?

A: From Q3, operating cash flow should stabilize by quarter; full-year positivity depends on Q4 deliveries. With new models ramping from Q3, management expects to sustain quarterly stability in operating cash flow. The cash position is ample, supporting product innovation, tech breakthroughs, and globalization.

This year, Li Auto is continuing to invest in R&D and CapEx, including ultra-fast charging, with FY CapEx guided at approx. RMB 6.0bn. Whether both operating and free cash flow turn positive in FY will largely depend on Q4 deliveries, though overall cash flow should outperform last year.

Q: What is the key contribution of in-house chips and tight HW/SW integration to advancing intelligent driving?

A: Chips, models, and Halo OS form a full-stack loop, and dual M100s can run multimodal base models on-device. Progress in performance and deployment speed comes from tight coupling of in-house chips with full-stack systems. Organizationally, streamlined structures bring chip and model teams closer, co-designing model architectures that fully exploit M100 compute.

On data and training, in-house chips enable deeper training pipeline optimization. Reinforcement learning built around the MACH platform significantly boosts model capability within a world-model framework, while platform data management and shadow data systems speed iteration.

System-level optimization via the in-house Halo OS tightly integrates applications with the underlying chips. This raises resource efficiency and system performance, improves engineering quality, and shortens dev cycles, forming a tightly integrated full stack across chips, models, and OS.

The value of in-house chips now extends from intelligent driving to embodied intelligence. Vehicles with dual MACH M100s can run multimodal foundation models on-device with voice, language, and video inputs for general problem understanding, environment comprehension, and task planning, moving beyond pre-defined functions.

Q: What is the overseas strategy and latest progress?

A: EREVs in Middle East/Central Asia, BEVs in Europe, with i6 sales starting in Q4. Overseas expansion is a long-term strategy, progressing steadily in market entry and product rollout. In the Middle East and Central Asia, L-series EREVs are the spearhead, with the new L9 launched in Kazakhstan and Uzbekistan in Jul and plans to launch in Dubai in Sept to formally start Middle East sales.

Li Auto has a strategic partnership with Allur, a leading Kazakh auto group, to advance local assembly. Through localized adaptation and assembly of current models, Li Auto will progressively build a global footprint across R&D, product, manufacturing, sales, and service.

In Europe, BEVs take priority. The i6 will debut at the Paris Motor Show in Oct and go on sale in Europe in Q4, while in right-hand-drive markets, Li Auto will launch the MEGA in Hong Kong and Singapore by year-end and introduce a right-hand-drive i6 to complete the lineup.

Management also highlighted uncertainties inherent in going overseas, especially geopolitics and regulatory regimes. Li Auto aims to position the brand as premium abroad, pacing expansion prudently, adapting to local conditions, ensuring compliance, building after-sales networks, and investing in brand.

Q: In today’s competitive landscape, will Li Auto sustain high R&D to develop embodied humanoid robots?

A: No comment on humanoid robots; focus stayed on chips, batteries, and e-drive investments. Management did not discuss a humanoid robot product or budget but outlined the broader R&D strategy. After a first decade of entrepreneurship, Li Auto will build moats via sustained R&D, with in-house chips a core long-term strategy from early on.

Chips are a core competitive advantage, and AI models are the competitiveness itself. Beyond chips and AI, Li Auto is also investing in core e-powertrain components and will roll out Li Auto-branded batteries across all models from H2.

Management stressed that in-house development does not imply suppliers’ products are inferior. The in-house MACH M100 does not diminish respect for Nvidia, and in-house batteries do not detract from leaders like CATL.

In the era of embodied intelligence, chips and batteries will be the most important competitive advantages, while e-powertrains and great products define product competitiveness. The choice to develop in-house reflects a desire, like Apple or Huawei, to control core components that determine competitiveness.

<End of text>

Risk disclosure and disclaimer:Dolphin Research Disclaimer and General Disclosure