ECARX Holdings, listed on Nasdaq as ECX, reported third-quarter revenue of USD 241 million for the three months ended 30 September, up 10 per cent year on year, with product sales of USD 223 million up 23 per cent.
The gap between those two numbers is the story. Product revenue is growing more than twice as fast as the total, which means software and service lines are not carrying the quarter. Hardware is.
Volumes and launch cadence
The company shipped about 721,000 units in the quarter, with 21 new models across five brands entering production. It added 23 new design wins, and cumulative installations across the first nine months reached roughly 1.4 million vehicles.
Launch cadence is the metric that matters for a tier-one cockpit supplier, because a design win is only revenue once the vehicle reaches volume. Twenty-one launches in a single quarter is a heavy integration load, and it explains why the product line, rather than the software line, is doing the growing.
The memory problem
ECARX said gross margin came under temporary pressure from rising global memory chip costs, and that it is responding through pricing adjustments, product mix optimisation, supply chain management and integration of its software capabilities. It maintained its full-year revenue guidance.
This is the same squeeze hitting every Chinese smart-cockpit supplier. A modern cockpit domain controller carries a fixed bill of DRAM and NAND, and memory prices have risen sharply through 2026. Suppliers with annual price-down obligations written into their contracts cannot pass that through immediately, so the difference lands on gross margin until the next sourcing round or the next model year.
The mitigation ECARX describes is a mix shift towards higher-value platforms, naming its Antora and Pikes product lines. Higher silicon content per vehicle raises revenue per unit, but it also raises memory content, which is why mix alone does not solve the problem. It only outruns it while memory prices are stable.
The European thread
The most consequential item in the release is buried in the market development section. ECARX completed its acquisition of the Flyme software business and, under a cooperation arrangement, will work with a European luxury carmaker to build an AI cockpit innovation joint laboratory on Flyme Auto, developing next-generation intelligent cockpit systems for that manufacturer’s China-market models, with the resulting cars expected on the road by the end of 2026.
That structure is becoming standard. A European brand keeps its global cockpit architecture and hands the China market to a local software stack, because the Chinese market’s expectations for maps, voice, mini-programs and app ecosystems cannot be met by a European head unit team on a European release cycle.
The consequence is subtle and worth stating. Every such deal gives a Chinese software layer a reference customer with a European badge, and gives the European brand a dependency it will eventually have to unwind in its own market.
August and September moves
In October the company began internal testing of Super Aicy under its Flyme AIOS, and signed a memorandum of understanding with Arm to explore next-generation AI compute platforms for automotive and other applications.
Neither is revenue this year. Both point at the same bet: that the cockpit becomes the place where AI compute is bought in a car, and that whoever owns the runtime owns the silicon decision.
Editor’s note: This is an adapted translation of the original Gasgoo report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://auto.gasgoo.com/news/202611/10I70477463C601.shtml.