SDV SectorNews and signals from the software-defined vehicle sector. Global coverage, daily.

Volkswagen Group China named its in-house tech stack VEGA and dated its L3 architecture

At its first Tech Day in Hefei, Volkswagen Group China unveiled VEGA, an in-house stack joining vehicle platforms, electronic architecture and driver assistance, and said its own AI toolchain had cut software feature testing from up to three days to two or three hours.

Close-up of a server rack, with rows of cabling and small indicator lights running along the units.
Illustrative photo · panumas nikhomkhai / Pexels · not the vehicle described

Volkswagen Group China unveiled VEGA at its first Tech Day in Hefei on October 28, presenting it as a single in-house technology stack unifying vehicle platforms, the China Electronic Architecture (CEA) and advanced driver assistance. The group said the stack makes it the only international carmaker in China with a locally developed in-house tech stack, and that every car equipped with CEA is a software-defined vehicle.

The release put dates against the architecture. CEA has been deployed on five locally produced battery-electric models in 2026, and is engineered around a three-month major software capability update drumbeat with a full architecture upgrade every two years. In 2027 it extends to locally produced plug-in hybrid and combustion models, and the first major CEA update is to deliver full central computing for cockpit and ADAS, targeting L3 technology readiness. By 2029, the group said, CEA 3.0 will become a fully Agentic AI-based architecture.

Volkswagen also put figures on its own toolchain. C-Brain, an in-house AI toolchain in which agents take part in engineering solution design and code generation and assist with testing and validation, has cut software feature testing from up to three days to two or three hours, the company said, and reduced the cycle from problem localization through fix to version release from roughly two weeks of cross-team collaboration to twenty minutes.

GAIA 2.0, the group’s in-house AI world model, runs ADAS safety simulations equivalent to five million kilometers of driving a day, drawing on as many as 30 million driving scenario clips. The group said the system is separately targeting the equivalent of ten million kilometers of training data a day for L3. Volkswagen said it is developing its own L4-ready system-on-chip, the C7H.

Why it mattersA carmaker that publishes dated architecture milestones and before-and-after numbers for its own development toolchain makes its software roadmap checkable against what later ships, which is a different kind of disclosure from a platform announcement.

Source: Volkswagen Group China newsroom

The SDV Sector Brief

The month in vehicle software — top stories, what our readers read most, and an editor's take. First Tuesday of the month, 08:30 CET. Double opt-in, unsubscribe anytime.