Li Auto's vehicle gross margin has halved in a year, and the root cause lies in the battery ledger.

In last year's second quarter, Li Auto's vehicle gross margin was 19.4%; this year's second quarter it was only 9.4%. Spread across each vehicle, revenue per vehicle fell by about 15,000 yuan over the year, while cost per vehicle rose by about 12,000 yuan. Li Xiang said on the earnings call that Li Auto would not pass on battery and memory chip price increases to users. Car prices can't be raised, raw materials are getting more expensive, and the gross margin gets squeezed away.

Battery makers earn more than automakers keep from selling cars

CATL's net profit attributable to shareholders in the third quarter was 21.246 billion yuan. That is net profit, while automaker financial reports usually look at gross margin; comparing the two directly would exaggerate the gap. On the same basis: CATL's power battery system gross margin in the first half was 20.63%, while Li Auto's vehicle gross margin in the second quarter was 9.4%. Both figures are sales revenue minus the cost of those goods, so CATL's gross margin on selling batteries is more than twice Li Auto's on selling cars.

Comparison of Li Auto's vehicle gross margin and CATL's power battery system gross margin; "after battery cost reduction" is this article's estimate (Data sources: Li Auto 2026 Q2 financial report, CATL 2026 interim report; Chart: this article)

The thickest slice of meat in the supply chain is in the battery maker's bowl. Lithium carbonate briefly broke through 210,000 yuan/ton in June this year, and from September a consumption tax was imposed on batteries; price increases travel down the chain, and once they reach the automaker there is nowhere left to go. If Li Auto wants to pull its gross margin back to the 15% to 20% Li Xiang mentioned, batteries are a hurdle it cannot avoid.

Li Auto designs its own cells, saving roughly 2,000 to 10,000 yuan per vehicle

Self-developed batteries have already been installed in the L8, L6, and i8, with a full-line switch within a few months; the 2026 i6's cells are designed by Li Auto, manufactured on its behalf by Sunwoda and CALB, and the battery pack is made by Li Auto itself. In September, Li Auto invested 2.65 billion yuan for a stake in Sunwoda Power, becoming its second-largest shareholder.

The all-new Li Auto L8 battery pack: five-layer bottom protection structure and dual waterproof sealing (Image source: Li Auto official website, 2026-07-01)

How much can be saved? The company has not disclosed. Based on my estimate using public data, roughly 2,000 to 10,000 yuan can be saved per vehicle, equivalent to recovering about 1 to 4 percentage points of vehicle gross margin. This is an estimate, not an official figure.

The calculation is simple: market cell price, multiplied by the battery capacity in the official configuration table, then multiplied by a savings ratio of 10% to 20.63%. Cell prices use mid-October quotes from SMM, after tax about 0.37 yuan/Wh for lithium iron phosphate and about 0.52 yuan/Wh for ternary. Battery capacity is 51 kWh for the L6, 96 kWh for the i6, and 97.8 kWh for the i8, so the cells in one vehicle are worth about 19,000 to 51,000 yuan. The lower bound of 10% comes from Leapmotor's statement that self-developed cells cut costs by about 10% versus external sourcing; the upper bound of 20.63% is CATL's power battery system gross margin, equivalent to taking back all of the battery maker's gross profit, and is only a ceiling. Comparing the savings with Li Auto's roughly 245,000 yuan revenue per vehicle in the second quarter gives 0.8 to 4.3 points. Calculating across the whole fleet using DBS's estimate of Li Auto's 30 to 40 GWh annual electricity usage gives about 1.1 to 4.5 points; the two methods roughly agree.

Estimated per-vehicle savings from self-developed cells, not official figures (Data sources: SMM via Dongwu Electric New Energy, Li Auto official website configuration tables, Wallstreetcn, CATL 2026 interim report; Chart: this article)

This calculation also needs a discount. The 5C cells Li Auto uses are more expensive than the ordinary cells corresponding to SMM quotes, so the absolute amount saved could be larger; but the fixed investment in the battery R&D platform has already exceeded 500 million yuan, and management has said that contract manufacturing must leave partners a reasonable gross margin, and some of the savings may be passed on to buyers. The real result will most likely fall in the lower-middle part of the range.

Sunwoda Power lacks money, so it will listen to automakers

Sunwoda Power had a net loss of 3.169 billion yuan in 2025 and was still losing 324 million yuan in the first half of this year; its debt ratio at the end of June was about 84%. A battery maker that loses money year after year and has debt of more than 80% most lacks cash and stable large orders, and Li Auto has provided both. Whatever specifications, production scheduling, and prices the automaker wants, it is hard for Sunwoda to push back. Whether the 10% lower bound above can be held depends on this level of cooperation.

Cell workshop at Shandong Sunwoda New Energy (Zaozhuang); an ultrasonic welding machine performs tab welding (Image source: provided by the interviewed company, via Consumer Daily/Sina News, 2026-09-23)

In October, Xiaomi invested 1.7 billion yuan for a stake in Sunwoda Power at the same price as Li Auto's round. Sunwoda Power was already one of the cell manufacturers for Xiaomi's Longjia battery. With two major customers pushing orders in, the factory's utilization rate is high, fixed cost per cell is lower, and there is room for contract manufacturing prices to move down.

For circumstantial evidence, look at BYD. It makes its own batteries, and its automotive-related business gross margin in the first half was 22.33%. That basis includes components and other businesses, and the gap does not come entirely from batteries, but it shows where gross margin can stand for an automaker that keeps battery profit in-house.

Cost reduction will show up first in vehicle gross margin, starting in the fourth quarter

The L8, L6, and i8 have already switched, the 2026 i6 launched on October 28, and the first batches of the MEGA and i9 still use CATL. Sunwoda's Zaozhuang Phase II is planned at 24 GWh and entered trial production in May; this phase alone is close to 60% to 80% of Li Auto's annual demand, and although it also has to supply Xiaomi and other customers, capacity should not block the switch. In the third quarter only some models benefited; the fourth quarter is the first