Volkswagen wants to use the electronic architecture co-developed with XPeng in its mainstream European models.

Germany's Manager Magazin reported on October 26 that Schäfer, head of Volkswagen's core brand, is preparing to submit a proposal to management: about three-quarters of the next-generation models from Volkswagen, Skoda and Cupra would switch to the CEA electrical/electronic architecture developed in China. The E/E architecture can be understood as the car's nervous system—it determines how many controllers the car has, how wiring runs, and how software is updated over the air. CEA was developed jointly by Volkswagen China Technology Company, CARIAD China and XPeng, and has already been mass-produced in China this year. According to the report's estimates, switching European volume models to CEA would save Volkswagen 4 to 5 billion euros. The proposal hasn't been approved; the final decision rests with the Volkswagen Group board.

For software in Western markets, Volkswagen set up a joint venture with US-based Rivian, putting in $5.8 billion. Now the head of its own core brand is proactively proposing to hand European volume cars to this Chinese architecture. CEA has already been installed and running in China, which is why Volkswagen has the confidence to use it to crunch the European numbers.

The market's pricing of XPeng still treats it as a car-selling company. At the October 28 US close, XPeng's market cap was about 60 billion yuan. My judgment is direct: XPeng's technology cooperation should be valued separately; on its own, it can support over 40% of this market cap.

01 Nearly half of Q2 gross profit came from beyond car sales

As discussed in the first article of the series, nearly half of XPeng's Q2 gross profit came from "services and others." This item includes technology R&D services as well as parts and accessories. Its revenue doubled year-on-year, with a gross margin of 75.1%, versus 12.1% for car sales in the same period.

The company explained clearly why revenue doubled: an R&D project for an automaker passed several key milestones, and more parts were sold. That automaker is Volkswagen. Bernstein estimates that Volkswagen contributed about 1.2 billion yuan in R&D service revenue to XPeng in Q2.

This money follows project milestones. Once a milestone is passed, work already completed is recognized as revenue all at once, so individual quarters can jump. To judge what it's worth, look at how much can be stably earned per year over the entire contract cycle.

02 From signing to rolling off the line, the ID.UNYX 08 took two years

In March this year, the first car jointly developed by Volkswagen and XPeng, the ID.UNYX 08, rolled off the production line. From signing the technology cooperation agreement to rolling off the line took two years. XPeng's platform and software were thus installed into a Volkswagen mass-production car.

March 2026: the ID.UNYX 08 rolls off the line. Image from XPeng Motors.

This contract collects more than development fees. The platform and software agreement between the two covers two models for the Chinese market, including technology licensing and royalty charges. According to XPeng's interim report, part of the licensing fee is tied to sales volume—as Volkswagen sells cars, XPeng recognizes revenue accordingly. Once R&D is done, money can keep flowing in along with Volkswagen's sales.

The CEA line goes further. According to Volkswagen's official statement, CEA went from concept to mass production in just 18 months, the fastest the group has ever developed a new E/E architecture, with no reduction in validation standards. The two sides subsequently signed an agreement to expand cooperation, extending CEA from pure-electric platforms to Volkswagen's combustion and plug-in hybrid platforms in China. With the same architecture installed in more models, Volkswagen does less duplicate development, and the volume of what XPeng contributes also grows.

The European proposal emerged on this foundation. Once it was proven in China, the European team would seriously compare it on cost.

03 Guosheng estimated the Volkswagen cooperation at $3.9 billion

Valuing the Volkswagen cooperation separately is something brokerages have already done. Guosheng Securities' XPeng report published August 22, 2024 split the company into two parts: the core car-selling business valued on revenue, and the Volkswagen cooperation valued on net profit, on the grounds that this business has high gross margin and fast growth. At the time, it calculated that the Volkswagen cooperation would bring XPeng 1.1 billion yuan in net profit in 2024, and at 25x P/E—pricing on 25 years of profit—valued it at $3.9 billion.

I agree with this split. For every yuan of car sales, after deducting the cost of building and selling the car, not much is left. For every yuan of technology service revenue, after deducting R&D and adaptation investment, much more is left. Mix the two kinds of money together and apply the same P/S ratio (market cap divided by revenue), and the technology part gets discounted at car-selling prices.

The method itself is simple. Revenue recognized on technology contracts, minus corresponding R&D, allocated expenses and taxes, is the annual net profit of this business. Give it a separate multiple, add it to the valuation of the car-selling business, and you get what XPeng should be worth.

Guosheng updated twice afterward. The multiple was cut by more than half, the profit forecast more than doubled, and the standalone valuation stayed above 10 billion yuan. That 2024 estimate only counted the domestic market.

04 The technology business is worth about 27 billion yuan

I recalculated the profit forecasts and multiples Guosheng gave across several reports against today's market cap. All profit and multiple figures in the table come from publicly available brokerage reports; I added no profit assumptions of my own.

Table 1 What XPeng's technology is worth: recalculated per Guosheng Securities' public methodology