EV & mobility · Deep dive
What EV gross margin actually measures, and why the Chinese makers report it differently
XPeng's gross margin rose from 17.3% to 20.7% over a year in which its cars got less profitable, not more. Both numbers come from the same filing. Here is how that works.
By Michael Westergaard · 12 Sept 2026 · 8 min
In the second quarter of 2026 XPeng reported a gross margin of 20.7%, up from 17.3% a year earlier. Over the same twelve months the margin on the cars it sold fell from 14.3% to 12.1%. Both figures appear in the same press release, three paragraphs apart, and both are correct.
That gap is not an accounting trick. It is what happens when a single percentage has to summarise a company that does more than one thing. Gross margin is the most quoted number in EV coverage and the least comparable, and the reason is always the same: the line above it and the line below it are not the same lines at every company.
The number itself
Gross margin is the first margin on the income statement and the closest thing to a measure of whether the product itself works as a business. Operating margin can be pushed around by how hard a company is spending on growth. Net margin picks up tax and financing. Gross margin is meant to answer a narrower question: after paying to build the thing, what is left?
For a carmaker that question matters more than it does almost anywhere else. The bill of materials on an electric vehicle is most of its cost, battery cells alone being a large fraction of that, and there is very little operating leverage hiding in the model. A carmaker that cannot make a positive gross margin on a car does not grow its way out of the problem by selling more of them.
Trap one: what else is in the numerator
XPeng is the cleanest illustration, because its two numbers moved in opposite directions in the same quarter.
| Period | Gross margin % | Source |
|---|---|---|
| Q1 25 | 15.6% | XPENG Q1 2025 results |
| Q2 25 | 17.3% | XPENG Q2 2025 results |
| Q3 25 | 20.1% | XPENG Q3 2025 results |
| Q4 25 | 21.3% | XPENG Q4 and FY2025 results |
| Q1 26 | 20.6% | XPENG Q1 2026 results |
| Q2 26 | 20.7% | XPENG Q2 2026 results |
Source: XPENG Q2 2026 results
Read that chart on its own and XPeng looks like a company steadily winning the cost war. Now add the second number. In Q2 2026 XPeng's vehicle margin, the margin on car sales alone, was 12.1%, against 14.3% in the same quarter a year earlier. The cars got worse while the company got better.
What closed the gap was a line item called services and others, which brought in RMB 2.70bn in the quarter at a gross margin of 75.1%. XPeng attributes the increase primarily to technical research and development services rendered to a car manufacturer, identified in the release only as "the Manufacturer", along with parts and accessories sales. Press coverage of the quarter identifies the counterparty as Volkswagen Group.
XPeng group margin, Q2 2026
▲ 20.7%
XPeng vehicle margin, Q2 2026
▼ 12.1%
Services and others margin
▲ 75.1%
Services and others revenue
▲ RMB 2.70bn
An engineering contract with a 75% margin is a genuinely good business and XPeng deserves the credit for winning it. It is simply not the car business, and a chart of group gross margin cannot tell you which one you are looking at.
A rising gross margin at a company that sells software to its competitors is not evidence that its factories got cheaper.
Trap two: what else is in the company
BYD is not really a carmaker for these purposes. It is a group that makes cars and also makes components and assemblies for mobile handsets, and the second business is much less profitable than the first.
In the first half of 2026 BYD's overall gross margin was 18.85%, up from 18.01% a year earlier. The automotive and related products segment inside it ran at 22.33%. Electronics and other products brought in RMB 69.41bn, 20.13% of group revenue, and pulled the average down by roughly three and a half points.
| BYD, H1 2026 | Share of revenue | Gross margin |
|---|---|---|
| Automobiles and related products | 79.85% | 22.33% |
| Electronics and other products | 20.13% | ≈5.1% |
| Group total | 100% | 18.85% |
The consequence is specific and easy to get wrong. Set BYD's 18.85% against Li Auto's 11.0% and the gap looks like 7.8 points of manufacturing skill. Set BYD's automotive segment at 22.33% against Li Auto's vehicle margin of 9.4% and the real gap is wider still, but it is now a comparison between two car businesses rather than between a car business and a conglomerate average.
Trap three: what has been added or taken out
Tesla is the company that made this problem famous, and also the one that solved it most openly. It sells regulatory credits to other manufacturers who need them to meet emissions rules. Those credits carry, in Tesla's own description, negligible incremental cost, so close to the whole of that revenue falls through to gross profit.
Because a credit sale is not a car sale, Tesla publishes automotive gross margin excluding regulatory credits alongside the headline figure. It is an unusually honest piece of disclosure: the company volunteers the number that makes it look worse, because the alternative is an automotive margin that moves on regulatory policy in other countries.
Li Auto splits its reporting the same way, and the split is where its recent story actually lives.
| Period | Gross margin % | Source |
|---|---|---|
| Q1 25 | 20.5% | Li Auto Q1 2025 results |
| Q2 25 | 20.1% | Li Auto Q3 2025 results, prior-quarter comparative |
| Q3 25 | 16.3% | Li Auto Q3 2025 results |
| Q4 25 | 17.8% | Li Auto Q4 and FY2025 results |
| Q1 26 | 7.9% | Li Auto Q2 2026 results, prior-quarter comparative |
| Q2 26 | 11% | Li Auto Q2 2026 results |
Source: Li Auto Q2 2026 results
Group gross margin of 11.0% in Q2 2026 against 20.1% a year earlier is the number that gets quoted. Vehicle margin of 9.4% against 19.4% is the number that describes the cars, and it is the worse of the two. Li Auto attributes the deterioration primarily to product mix during a model changeover. Whether that holds is the thing to watch: a mix explanation predicts recovery as the new range settles, and Q2's 3.1-point sequential gain from Q1's 7.9% is at least consistent with it.
What a comparable comparison looks like
Strip each company down to the margin on cars and the picture reorders itself.
| Q2 2026 | Headline gross margin | Margin on vehicles alone | Gap |
|---|---|---|---|
| XPeng | 20.7% | 12.1% | 8.6 pts |
| Li Auto | 11.0% | 9.4% | 1.6 pts |
| BYD (H1 2026) | 18.85% | 22.33% | +3.5 pts |
On headline numbers XPeng beats Li Auto by 9.7 points. On the cars themselves the gap is 2.7 points. The first number is the one that circulates; the second is the one that describes the industry the two companies are actually competing in. And BYD is the only one of the three whose cars are more profitable than its headline suggests, because it is the only one carrying a lower-margin business alongside them.
For scale, a company with genuine pricing power
It is worth seeing what a gross margin looks like when a manufacturer has no real competition.
| Period | Gross margin % | Source |
|---|---|---|
| Q2 25 | 53.7% | ASML Q2 2025 results |
| Q3 25 | 51.6% | ASML Q3 2025 results |
| Q4 25 | 52.2% | ASML Q4 and full-year 2025 results |
| Q1 26 | 53% | ASML Q2 2026 results, prior-quarter comparative |
| Q2 26 | 54% | ASML Q2 2026 results |
Source: ASML Q2 2026 results
ASML sells lithography systems that nobody else can build, and its margin sits above 50% and stays there. The Chinese EV makers are fighting over single percentage points in the high teens. That is not a criticism of either: it is the difference between a monopoly on a production input and a price war in a consumer market, and it is a useful reminder that "good gross margin" has no meaning outside its industry.
Reading the next set of results
Three habits cover most of it.
- FirstFind the vehicle or segment margin before the group margin. If a company reports one, it reports it because the group number is answering a different question.
- SecondCheck what non-vehicle revenue did. A jump in a high-margin services line lifts the group figure without a single cheaper car.
- ThirdCompare like periods. BYD reports segments semi-annually; the US-listed Chinese makers report quarterly. A half-year margin against a quarterly one is not a comparison.
None of this means the headline figure is useless. It means it is a summary of a company, not a measurement of a product, and the EV price war is being fought at the level of the product. Our piece on NIO's brand mix is the same problem seen from the volume side: a group number that held steady while everything underneath it moved.
Sources
- XPENG Q2 2026 unaudited financial results
- XPeng Q2 margin hits 20.7% as overseas revenue tops 25% (Automotive World)
- Xpeng posts EUR 2.5 billion in revenue in Q2 2026 (electrive)
- Li Auto Q2 2026 unaudited financial results
- Li Auto narrows Q2 loss as margin remains nearly half year-earlier level (CnEVPost)
- BYD H1 profit falls 20.5% as overseas growth fails to offset China weakness (CnEVPost)
- BYD half-year figures and segment margins (ad-hoc-news)
- Tesla Form 10-K, revenue recognition on automotive regulatory credits
- ASML Q2 2026 financial results
- Quarterly margin series charted above are held in this site's company files, each point carrying its own source. Gap columns and the implied BYD electronics margin are calculated by Northline Post from the published figures.
Related reading
Northline Post publishes information and opinion, not investment advice. Nothing here is a recommendation to buy or sell any security. The author may hold positions in companies covered.