For decades, the global semiconductor industry operated on an unspoken assumption: the world would design chips in the U.S. and Europe, manufacture most of them in Taiwan and South Korea, and assemble them in China. That era is now accelerating toward an uncertain end.

China’s determined, multi-billion-dollar push to build a self-reliant semiconductor ecosystem is no longer a distant policy goal; it is actively reshaping supply chains, rattling stock prices, and forcing executives from San Jose to Stuttgart to rethink their long-term strategies.

Three years after export controls from the U.S. and allies cut off access to advanced chipmaking tools, Beijing has doubled down. The result is a fragile but fast-moving domestic industry that is already altering the balance of power in mature-node chips and quietly making headway on the cutting edge.

The Mature-Node Surprise

While headlines focus on 3-nanometer and 5-nanometer processors for smartphones and AI servers, the quieter revolution is happening in “legacy” chips: 28-nanometer and above. These power everything from cars and washing machines to military drones and power grids.

According to a new analysis from TechInsights and the Semiconductor Industry Association, China is on track to produce nearly 40% of the world’s mature-node chips by 2027, up from roughly 25% in 2021. That surge has already triggered price drops across microcontrollers, power management ICs, and display drivers.

“For non-cutting-edge applications, Chinese fabs are now genuinely competitive on cost and, increasingly, on reliability,” says David Wu, a Shanghai-based supply chain consultant. “Global automakers and industrial giants are quietly qualifying Chinese-made chips as Plan B. But for some components, Plan B is becoming Plan A.”

The AI Wildcard

The most dramatic shake-up, however, involves advanced AI chips. When U.S. sanctions cut off Nvidia’s H100 and A100 from Chinese customers, officials in Beijing responded not by retreating but by turbocharging domestic alternatives.

Huawei’s Ascend 910B, despite yield and software stack challenges, has found willing buyers across China’s cloud and AI sector. Meanwhile, startups like Biren Technology and Enflame have received fresh rounds of state-backed funding, together raising over $2 billion in the past six months alone.

Chip equipment makers in the Netherlands and Japan, bound by their own export curbs, report a sudden uptick in demand for older-generation machines that remain legal to ship. Industry insiders say Chinese firms are stockpiling these tools to build out “transitional” lines, hoping to leapfrog to next-gen processes by 2026.

Global Ripples

The effects are already visible across the industry:

South Korea’s Samsung and SK Hynix, once reliant on China for assembly and some memory production, have accelerated plans to diversify to Vietnam and the U.S. But they also face a new reality: China’s YMTC is now a genuine rival in NAND flash, having recently shipped its fifth-generation 3D NAND chip with industry-competitive density.

European auto suppliers, including Bosch and Infineon, have launched “dual-source” procurement programs, auditing Chinese foundries as backups. Some are taking equity stakes in Chinese chip startups to secure supply.

The U.S. CHIPS Act, intended to revive domestic fabrication, is now competing with an even larger Chinese incentive package estimated by some analysts at over $40 billion in direct subsidies plus low-interest loans. “It’s not an arms race yet,” says a Washington-based trade official who declined to be named. “But on subsidies alone, China is outspending everyone.”

The Cautious Reality

To be clear, China remains years behind in extreme ultraviolet (EUV) lithography, the linchpin of sub-7nm production. Its domestic lithography leader, SMEE, has only recently delivered its first 28nm-capable scanner, while ASML’s machines have been doing that for a decade.

Sanctions have also fractured access to advanced EDA software and certain chemicals. Several ambitious Chinese AI chip projects have missed deadlines or pivoted to less ambitious designs.

But there is a growing consensus that further isolation will not stop China, but only slow it. “Every time a new restriction comes out, a new domestic solution appears six to twelve months later,” says former Intel engineer Mark Chen, now advising cross-border tech funds. “The quality isn’t always there initially. But the sheer volume of engineers and capital means they iterate fast.”

What Comes Next

For the rest of the global tech industry, the message is clear: the era of a single, unified semiconductor supply chain is ending.

Multinationals are already building parallel supply lines, one for China, one for the rest of the world. That adds cost and complexity but hedges against geopolitical shocks. Meanwhile, chip equipment suppliers from Tokyo Electron to Lam Research are seeing divergent trends: rising sales of maintenance and older tools to China, flat or declining orders for EUV-related gear.

One thing is certain. China has made its ambition explicit: by 2030, it aims to produce over 70% of the chips it consumes at home. Whether or not it hits that number, the attempt alone is forcing a restructuring of the global industry, one that will be felt for a generation.