A modern electric vehicle carries roughly a thousand semiconductor devices. Fewer than a dozen handle the autonomous driving hype that fills investor presentations. The rest manage battery thermal regulation, window motors, seat position sensors, and the timing of windshield wipers. These are 22–40 nanometre chips, mature nodes built on processes a decade or more old, and they are becoming China’s domain.
SEMI China forecasts Chinese foundries will produce 37% of global 22–40nm output by 2026, climbing to 42% by 2028. That five-point jump represents millions of wafers annually, flowing from fabs constructed at a pace unmatched elsewhere. Between 2021 and 2024, China added 32 new fabrication facilities, with the majority committed to 28nm and larger geometries. SMIC and Hua Hong Semiconductor have led this buildout, pouring capital into production lines for nodes that Western headlines routinely dismiss as legacy technology.
The Volume Beneath the Hype
The semiconductor industry operates on a prestige gradient. Three-nanometre GPUs and neural processing units command analyst attention and export control debates. Mature nodes attract none of this glamour, yet they constitute the physical substrate of industrial civilisation. A Bosch engine control unit, a Siemens programmable logic controller, a Haier refrigerator’s inverter board, each depends on chips that do not need to shrink further because their jobs are defined, stable, and massive in quantity.
Automotive applications illustrate this most starkly. A typical combustion vehicle contained perhaps 500 chips in 2010. A contemporary plug-in hybrid carries 1,500 to 2,000. The incremental growth sits almost entirely in power management, body electronics, and sensor interfaces, categories where 28nm and 40nm offer optimal cost-performance ratios. NXP’s S32K microcontrollers and Renesas RH850 families, workhorses of engine management and advanced driver assistance, are manufactured at these geometries. They will not migrate to 3nm because there is no engineering or economic case for doing so.
Industrial automation, white goods, communications infrastructure, and medical devices follow the same pattern. A 5G base station requires sophisticated baseband processing at advanced nodes, but its RF transceivers, power amplifiers, and antenna calibration circuits rely on mature processes. Solar inverters and battery management systems for grid storage demand robust, high-voltage-capable chips that prioritise reliability over transistor density. These are not edge cases; they are the bulk of semiconductor consumption by unit volume, and they are precisely where Chinese capacity is concentrating.
How Domestic Dominance Translates
Proximity between fabrication and final assembly generates advantages that compound across the manufacturing stack. Chinese electric vehicle makers, BYD most prominently among them, have pursued vertical integration into semiconductor design. BYD Semiconductor produces IGBTs and SiC modules for its own vehicle platforms, sourcing wafers from domestic foundries where possible. The distance between a Hua Hong fab in Shanghai and a BYD assembly plant in Shenzhen is measured in hours, not weeks of ocean freight and customs clearance.
This geography compresses iteration cycles. A power management IC that requires voltage regulation tweaks for a new battery chemistry can move from design revision to wafer qualification in weeks rather than quarters. Foreign competitors relying on Taiwanese or Korean foundries for comparable components face longer lead times and less flexible queue positioning. During the 2020–2022 automotive chip shortage, this differential became decisive. Chinese OEMs maintained production volumes while European and North American plants idled for want of 40nm engine controllers.
Cost structures follow from scale and competition. China’s mature node expansion has created surplus capacity among domestic foundries before global demand has fully absorbed it. The resulting price pressure benefits Chinese downstream manufacturers directly and indirectly. Import tariffs and currency exposure disappear from component costs. Inventory carrying costs fall when supply chains shrink to domestic logistics networks. These savings are not dramatic in isolation, but they accumulate across a thousand chips per vehicle, a hundred million vehicles per decade.
The Dependence Being Built
China’s mature node trajectory creates a structural asymmetry in global manufacturing. Western industrial policy, particularly the US CHIPS Act and European Chips Act, directs subsidies toward advanced node reshoring, logic below 7nm, and novel memory architectures. This is defensible as a matter of national security and frontier technology access. It also leaves the foundational layer of industrial semiconductors to Chinese competitive dynamics.
The risk is not immediate supply interruption, as Chinese foundries remain commercially motivated sellers to global markets. The risk is gradual path dependence. As Chinese mature node production reaches 42% of global supply, foreign manufacturers face rational incentives to qualify Chinese sources for cost and availability reasons. Each qualification deepens switching costs. Over a five-year horizon, automotive tier-one suppliers and industrial equipment makers may find their second-source options have withered, not through explicit exclusion but through competitive attrition.
Established foundries outside China, TSMC’s older node operations, UMC, GlobalFoundries, face a volume challenge they are structurally disinclined to meet. Their capital allocation prioritises advanced node development where margins justify the investment intensity. Mature node expansion in the US or Europe must compete for subsidy pools already committed to leading-edge projects. The arithmetic favours Chinese state-backed capacity additions that do not require equivalent returns.
Prestige Versus Position
Frontier AI chips generate geopolitical theatre. Export controls on NVIDIA H100s and the tracking of smuggling networks through Southeast Asia fill diplomatic cables and policy conferences. The assumption underlying this focus is that computational leadership at the frontier determines strategic advantage. This is partially true for specific applications, such as large language model training, certain military simulations, and cryptographic analysis.
It is not true for the material economy that most people inhabit. A country that controls 42% of the chips enabling vehicle production, factory automation, and energy infrastructure holds a position that is more diffuse and harder to counter. Every manufacturing supply chain, every logistics network, every power grid contains these components. Their ubiquity makes them unremarkable individually and collectively indispensable. The country that shapes their cost, availability, and technical evolution shapes the terms on which physical goods are produced worldwide.
China’s semiconductor strategy has been read as a pursuit of technological respectability, the drive to match Western and Taiwanese capabilities at the leading edge. This reading misses the larger movement. The mature node buildout is not a consolation prize while advanced lithography remains constrained. It is the construction of a different kind of power, one embedded in the operating hardware of industrial civilisation rather than its research frontiers. The headlines follow the GPUs, but the dependencies form around the microcontrollers.
