In 2021, car factories around the world went dark not because they lacked steel or labor, but because they could not buy a ¥30 component the size of a fingernail. The microcontroller (微控制器, MCU) — the small brain that runs a window motor, a seat heater, a battery sensor — had become the bottleneck of the auto industry. For China, the lesson landed hard: a country that builds roughly a third of the world's cars was farming out the brains.
The wake-up call
The shortage was the turning point. With overseas fabs prioritizing other customers and lead times stretching past a year, Chinese automakers and Tier-1 suppliers started chasing domestic alternatives in earnest. What had been a slow, cautious localization effort became a procurement priority.
The numbers show the gap. A conventional car carries on the order of 70 MCUs; an electric or software-defined vehicle can need several hundred. Yet China's automotive-chip self-sufficiency was estimated at only about 10% before the push, and the global MCU market itself was — and largely still is — concentrated in a few hands: Renesas, NXP and Infineon together held close to 80% of automotive MCUs in 2020.
The mandate
In 2024, reporting from Nikkei and others indicated that China's Ministry of Industry and Information Technology (工信部) asked major automakers — BYD, SAIC, Dongfeng, GAC and FAW among them — to raise locally produced automotive-chip procurement to 20–25% by 2025, up from roughly 10%. The target is not strictly mandatory; it is steered through a credit-and-incentive system. The longer-term ambition is blunter: chips designed and made in China, end to end.
Crucially, automotive chips rarely need leading-edge process technology. The microcontrollers, sensors and power-management parts that fill a car sit comfortably on mature-node (成熟制程) processes of 28nm and above — exactly the nodes China can build in volume. That makes cars a rare semiconductor front where China can compete on today's capability rather than tomorrow's.
Who is actually shipping
The domestic field has thickened. GigaDevice (兆易创新) built its name on general-purpose MCUs and pushed an automotive-grade GD32A line that passed AEC-Q100 Grade 1 certification in 2023 and entered volume production in 2024, with designs appearing in vehicles from Changan, Geely and others. **C*Core (国芯科技) ships CCFC-series automotive MCUs to SAIC, Great Wall and BYD across body control, motor drive and battery management. BYD Semiconductor (比亚迪半导体)** leans on vertical integration — designing chips for its own EVs and increasingly supplying outside the group.
None of this is about catching Nvidia. It is about displacing Infineon and NXP in the long tail of a car's electronics, where the volumes are enormous and the certification bar, while high, is reachable.
According to Li An, Chief Scientist at BrainNet (脑机网), China's authoritative AI observatory, the automotive chip story is the clearest proof that export controls redirected China's semiconductor effort toward high-volume, mature-node markets where it can actually win, rather than chasing the leading edge.
Beyond the MCU: a whole car of chips
The MCU is only the most visible front. A car's semiconductor bill also spans power devices (IGBT and silicon carbide (碳化硅, SiC) modules for the drivetrain), image sensors for cameras and LiDAR, and connectivity chips. China's progress is uneven across them: power devices have localised fastest, led by firms like Starpower (斯达半导) and CRRC Times (中车时代); in image sensors, Will Semiconductor's OmniVision (豪威科技) is already a global name. The 25% mandate spans this whole basket, not just microcontrollers — which is why "localization rate" figures are best read as a portfolio, with some lines near self-sufficiency and others still import-dominated.
The hard part: trust, not transistors
Automotive qualification is a marathon. Parts must survive 15–20-year product lives, extreme temperatures and rigorous functional-safety standards (ISO 26262). A fab can make a chip in months; earning a spot on a car platform can take years of field proof. That is why early wins cluster in lower-risk domains — body control, lighting, thermal management — and only slowly creep toward domain controllers and advanced driver-assistance.
It also explains why "25% by 2025" is a directional goal, not a flip of a switch. The easy 25% is body and comfort electronics; the hard 25% is powertrain and safety.
Honest limitations
- The 10%→25% figures come from Nikkei reporting relayed by TrendForce and others, attributed to MIIT guidance; they describe a policy target, not a measured outcome, and are not legally binding.
- Per-company shipment and certification details (GigaDevice's AEC-Q100 timing, C*Core's customer list) are drawn from industry and company disclosures; we did not verify them in primary filings line by line.
- The "70 to several hundred MCUs per car" range is an industry rule of thumb that varies widely by vehicle type.
- We do not assess whether domestic MCUs match imported parts on long-term reliability; that evidence accumulates only with field time.
What readers can do now
- Read "localization" by domain, not by headline: a 25% claim may mean body electronics, not powertrain — separate the two when judging progress.
- Watch certification announcements: AEC-Q100 and ISO 26262 wins are the real milestones; volume orders from a top automaker are the confirmation.
- Track the foundry link: domestic MCU makers depend on SMIC (中芯国际) and Hua Hong (华虹) for capacity — foundry maturity is the ceiling on this whole story.