Li Auto and XPENG sell technology: who pays the next development bill?

Both want customers, but they start from different positions

Competition between automakers is moving from showrooms into purchasing departments. Li Auto wants external buyers for chips and power modules. XPENG wants to sell more companies the technology capabilities developed through its Volkswagen partnership. For industry executives, the central development is a change in who pays for engineering: investment previously supported by a manufacturer's own vehicle sales is seeking a second set of customers.

At a September 21 media briefing, Li Auto's Ma Donghui said chips and silicon-carbide modules could be supplied externally, while battery packs customized for its own vehicles were unsuitable for outside sales. His comments established a product boundary for earlier reports. ^[1]On September 18, 36Kr had reported plans involving Mach chips, Sike Semiconductor and range-extender systems, alongside financing plans for some businesses. Specific commercial progress on range-extender systems remains a media-reported claim. ^[2]

XPENG's comments point more directly to new customers. In a September 17 interview with National Business Daily, He Xiaopeng confirmed discussions with automakers and companies outside the automotive industry beyond Volkswagen, covering technology and chips. He mentioned the possibility of announcing cases at the Paris Motor Show in October. Management has confirmed discussions; that statement did not disclose new counterparties, contract values or production dates. ^[3]

Together, these developments show an industry reconsidering how to share its development bill. XPENG has an external partnership to replicate. Li Auto has internal applications to turn into a supplier business. Treating both as already established, highly profitable businesses would obscure the different problems each must solve.

Own-brand sales cannot carry every development investment

Automakers develop chips, software and electronic architectures primarily to differentiate their vehicles. The constraint is economic: engineering, tools and validation require spending upfront, while recovery depends on subsequent deployment. More intense vehicle competition and shorter product cycles increase the pressure to spread those costs beyond a single manufacturer's sales.

Financial disclosures provide context. Li Auto's vehicle margin was 9.4% in the second quarter of 2026, against 19.4% a year earlier. The company mainly attributed the change to product mix; quarterly R&D expenses were approximately RMB2.8 billion. ^[4]Those figures do not establish the cause of any particular supply initiative. They do explain why management needs existing technology assets to work harder. Serving external vehicles could add another route to recovering development spending.

XPENG's revenue mix makes the attraction clearer. In the same quarter, vehicle revenue was RMB17.05 billion at a 12.1% gross margin. Services and others generated approximately RMB2.70 billion at 75.1%. Multiplying the rounded disclosed figures produces estimated gross profit of RMB2.06 billion and RMB2.03 billion respectively. A much smaller revenue stream contributed almost as much gross profit as vehicle sales. ^[5]

The operating implication is straightforward: technology partnerships deserve management as a distinct business. However, the services category includes parts and accessories, so its revenue is not all licensing fees. Growth also reflected contractual milestones in technical R&D services. The 75.1% figure describes that quarter's combined segment, not a sustainable licensing margin or net profitability.

Boards should focus on incremental cash returns after customer adaptation, validation, support and ongoing maintenance. External sales improve returns when reuse grows faster than engineering headcount and delivery costs. Otherwise, the research organization has simply taken on additional contract-engineering work.

XPENG sells development efficiency; Li Auto starts with components

XPENG has a useful reference. Volkswagen says the China Electronic Architecture, jointly developed by Volkswagen Group China Technology Company, CARIAD China and XPENG, has entered production application and spans multiple platforms and powertrain types. Volkswagen also attributes shorter development cycles and other benefits to its new development process. ^[6]This provides an engineering outcome inside a customer's organization, beyond XPENG's own claims.

The next pitch should therefore center on shorter development schedules, less integration work and a maintainable software architecture. A customer buying chips and software together ultimately compares total project cost and time to market. Avoiding a missed product cycle can matter more than the price of a single component.

A successful Volkswagen project does not automatically become a reusable product. If every additional customer requires extensive interface redesign, new tools and larger teams working on site, sales can grow with little scale benefit. XPENG's key task is to extract a stable common technology layer and price brand-specific functions and customization separately. Its second customer beyond the established partnership will test whether that boundary works.

Li Auto's more practical entry point is a component with independently defined specifications, quality obligations and supply responsibilities. Buyers can evaluate silicon-carbide power modules on conversion efficiency, thermal behavior, reliability and delivered cost. Their connection to vehicle branding is relatively weak. Chips also offer an opportunity, but they require the surrounding software support.

A chip running well inside a Li Auto vehicle demonstrates successful internal integration. An external buyer must migrate models, debug development tools, validate performance and arrange support over its product lifecycle. The complete commercial offer needs a development environment, adaptation support and a long-term supply commitment. A headline computing-performance figure does not resolve a purchasing team's project risk.

The likeliest sequence is that more standardized components win initial projects, chips enter applications with manageable adaptation requirements, and broader system partnerships expand later. This is an assessment of commercial sequencing, not a disclosed customer roadmap. Keeping customized battery packs outside the offer illustrates why internal technical value and external marketability require separate business cases.

The hardest task is earning a competitor's trust

Buyers worry about more than a successful demonstration. They need to know whether the supplier will prioritize its own vehicles tomorrow. When buyer and supplier compete in the same price segment, technology roadmaps, engineering resources and capacity allocation become contractual issues.

A separate company can clarify accountability, but incorporation alone does not solve the problem. Customers need enforceable arrangements covering separation of business data and vehicle plans, continuity during supply interruptions, control of software updates, replaceable interfaces and responsibility for defects. An external-sales business needs its own service budget and delivery commitments rather than indefinite reliance on engineers borrowed from the parent.

Early buyers are most likely to include companies with limited development resources, urgent launch schedules and little appetite for rebuilding underlying technology. International automakers' China projects also have reasons to buy capabilities already suited to local development speed. Reuters' September 17 report described XPENG's effort to reach foreign automakers beyond Volkswagen. That is a customer-development direction, not evidence of new international volume-production contracts. ^[7]

Buyers should retain the ability to define products and manage suppliers. Cabin experience, brand interaction and feature choices help determine whether customers buy a car; underlying modules can be purchased against cost, timing and switching requirements. A dangerous procurement bargain would exchange control over upgrades and future cost decisions for speed on a single launch.

For suppliers, accepting clear boundaries can make deals easier. Allowing buyers to own the brand-facing layer, with explicit interfaces and exit provisions, makes technology more acceptable in strategic programs. Excessive dependence on the supplier's ecosystem may increase the value of one contract while shrinking the pool of potential customers.

Order quality matters more than the number of partners

Over the next one to two years, the most likely result is a series of clearly scoped technology projects and component orders, rather than new automakers rapidly becoming industry-wide platforms. Automotive validation takes time and buyers have established suppliers. External revenue should initially concentrate in a few customers before expanding with production programs.

XPENG is currently better placed to expand technology-service revenue because prospective buyers can evaluate an existing automaker partnership. The important milestone is whether the second and third projects reuse the same technical base, require less incremental engineering and convert acceptance milestones into continuing cash receipts.

Li Auto should first demonstrate that customers will pay for standardized products and return for subsequent vehicle programs. Reliability, delivery and repeat orders will say more about business value than a major financing round. Vehicle management should separately track the resources consumed by external projects so that new revenue does not conceal the cost of delays to internal programs.

The first beneficiaries should be buyers able to shorten development schedules and suppliers with mature technology, defined products and reliable delivery. Pressure will fall on undifferentiated integration layers and on automakers too small to justify duplicating every underlying technology. Each internal development program needs a fresh comparison with buying a mature solution and concentrating resources on the finished product.

Three outcomes will distinguish a durable business: named customers with defined production programs, declining adaptation costs for subsequent customers, and reliable cash collection after revenue recognition. If each revenue increment requires another large customization team, the activity will resemble engineering services. Reusable products that generate repeat orders would support an independently growing business.

Li Auto and XPENG are subjecting engineering to two tests: first, improve their own vehicles; then persuade another company's purchasing team to pay over time. Passing the second test would genuinely expand the commercial reach of their technology.