Half of Xpeng's Q2 gross profit came from non-car revenue. The next two quarters will test whether it lasts.

Xpeng just posted its strongest quarter on record: 20.7% comprehensive gross margin on 19.74 billion yuan (~$2.74 billion, at ~7.2 CNY/USD) of Q2 2026 revenue, up 8% year over year. Almost none of that improvement came from cars. The car business remained a 12.1% gross margin business, flat quarter over quarter, on 17.05 billion yuan (~$2.37 billion) of vehicle sales. What moved was a 2.7 billion yuan (~$375 million) services-and-other line, up 93.9% year over year, that contributed nearly half of Xpeng's gross profit on roughly 14% of revenue. On a per-dollar basis, that non-car line is now more profitable than the cars.

The constructive question for the rest of 2026 is whether that line is a durable second engine or a partnership-concentrated windfall. Management's own H2 guidance is 17%-19% comprehensive gross margin, already a step down from Q2's 20.7%, and is the first real test of whether the new mix is repeatable or starts to mean-revert as lower-margin MONA volume ramps and partner concentration shows through.

The services-and-other line is not a single business. It is at least three different contracts stacked on top of each other, each with its own risk profile.

The first leg is the Volkswagen R&D and platform partnership, publicly known since 2023 and now contributing technology licensing fees. Per a Leiphone exclusive drawing on insider sourcing, this leg is projected to contribute 500 million to 1 billion yuan (~$70-140 million) per quarter in Q3-Q4 2026, with some of the licensing shifting to a model-volume-linked structure over the vehicle lifecycle rather than a one-time transfer. That is the most credible structural piece of the three: it ties Xpeng revenue to vehicles VW actually sells, not to a fixed annual payment.

The second leg is the Porsche carbon-credit partnership, the headline item in the source. Recognition begins in Q3, with revenue expected in the hundreds of millions of yuan. Carbon credits are a regulated market: regulators tighten pool rules, EV share rises industry-wide, and the per-credit price falls. Xpeng is the supplier here, not the price-setter, and this leg is the most exposed to regulatory mechanics outside the company's control.

The third leg is overseas. Overseas revenue made up more than 25% of Xpeng's H1 2026 revenue, Q2 overseas deliveries crossed 20,000 units for the first time (+81% year over year), and overseas ASP is running above €40,000. Management is targeting 20%+ overseas gross margin and 40,000+ deliveries per quarter by 2027. Overseas is the leg most dependent on volume the company still has to win; the other two are largely contracted.

The counter-material is visible in the same earnings release. Deliveries were 103,295 in Q2 (+0.1% year over year, +64.8% quarter over quarter), and the sequential jump is largely a Q1 comparison artifact. Q3 and Q4 guidance points to storage chip and battery cost pressure remaining roughly stable rather than improving, which means the margin path from 20.7% to 17%-19% is not a function of input costs getting worse. It is a function of mix: more mass-market MONA volume, less of the partner-and-overseas mix that produced the 2.7 billion yuan line in the first place.

The structural read holds if the second engine is repeatable, not exceptional. The 17%-19% H2 guidance is consistent with that read: the partner legs continue, MONA L03 volume pulls gross margin down toward the car business, and overseas volume has to do the work of growing into its target margin profile. The non-car line is not a one-quarter story, but it is also not a flat mix either; it is the variable that explains most of the gap between Q2's 20.7% and H2's 17%-19%.

The next two earnings releases will show whether the 2.7 billion yuan line is the new floor or the old peak. Watch the Volkswagen licensing line first: it is the largest and the most contracted of the three legs. If it holds at 500 million to 1 billion yuan per quarter through Q4, the second-engine read is intact. If it slips, the carbon-credit and overseas legs have to grow into a hole the cars cannot fill.