Semiconductor industry in China

Based on Wikipedia: Semiconductor industry in China

In December 2025, Reuters reported that China had secretly completed a prototype of the most advanced chip-making machine on Earth—an extreme ultraviolet lithography system—in Shenzhen. If the report holds true, it would represent one of the most remarkable technological achievements of the decade, accomplished while facing the most comprehensive export restrictions any country has imposed on another in modern history.

This is the story of how the world's largest consumer of semiconductors is racing to become self-sufficient in making them.

The Scale of China's Chip Appetite

Here's a number that explains everything: 53.7 percent.

That's China's share of worldwide chip sales in 2020. More than half of all semiconductors sold on Earth went to China. In dollar terms, that's $239 billion out of $446 billion globally. But here's the catch—over 83 percent of those chips, some $200 billion worth, were imported from foreign companies.

Imagine being the world's largest market for a critical technology while producing almost none of it yourself. That's the position China found itself in, and it's the gap driving one of the most ambitious industrial policies in history.

Understanding the Players

Before diving deeper, it helps to understand how the semiconductor industry actually works. It's not one business—it's four distinct ones.

Integrated device manufacturers, or IDMs, do everything. They design chips and manufacture them. Think of them as vertically integrated operations, controlling the entire process from concept to silicon.

Pure-play foundries are the opposite of designers. They manufacture chips that other companies design. They're factories for hire, essentially—you bring the blueprint, they make the product. Taiwan Semiconductor Manufacturing Company, or TSMC, pioneered this model and remains its most successful practitioner.

Fabless semiconductor companies occupy the other end of the spectrum. They design chips but own no factories. They send their designs to foundries for manufacturing. This model lets companies focus entirely on innovation without the massive capital expenditure of building fabrication plants.

Finally, OSAT companies—that stands for Outsourced Semiconductor Assembly and Test—handle the final stages. After chips are manufactured on silicon wafers, they need to be cut apart, packaged into protective casings, and tested. It's the less glamorous but essential finishing work.

China has companies in all four categories, though not all are equally developed.

The Long Road from 1956

China's semiconductor industry is older than you might think. The country produced its first transistor in a state laboratory in 1956—just eight years after Bell Labs invented the transistor in the United States. By 1965, China had created its first integrated circuit.

But then something went wrong. For the next thirty-four years, the industry followed a Soviet-style system of industrial organization that separated research from manufacturing. Scientists worked in state labs. Factories produced things. The two rarely talked to each other effectively.

This matters enormously in semiconductors. The industry advances through tight feedback loops between design and manufacturing. You design something, try to make it, discover what doesn't work, redesign it, and iterate. Separating research from production breaks that loop.

By the 1970s, China had approximately forty semiconductor factories, but most of them produced basic diodes and transistors rather than integrated circuits. The Cultural Revolution, which disrupted Chinese society from 1965 to 1975, further stalled progress. Still, by 1972, China was producing third-generation computers—not at the cutting edge, but not hopelessly behind either.

The Reform Era and Its Frustrations

Deng Xiaoping's economic reforms starting in 1978 changed China's approach. The country began trying to catch up with the West rather than pursuing self-reliant development in isolation.

The sixth Five-Year Plan, covering 1981 to 1985, created a "Computer and Large Scale IC Lead Group" under the State Council to modernize the industry. China imported twenty-four secondhand semiconductor production lines from abroad by 1985. The results were disappointing. Only one factory—the Wuxi Factory No. 742, which had been operating since 1960—met its production targets.

Why such poor results? Secondhand equipment in semiconductors is often worthless. The industry moves so fast that by the time you've figured out how to use yesterday's machines, your competitors are two generations ahead. Technology transfer in semiconductors requires more than equipment—it requires people who understand the processes, and processes that were developed iteratively over years of trial and error.

The 1990s brought a new strategy: joint ventures with foreign companies. China partnered with Nortel, Philips, NEC, and ITT, hoping to acquire technology through collaboration. The eighth Five-Year Plan focused resources on developing Huajing, the company operating the Wuxi factory, into a leading integrated device manufacturer. A joint venture with Lucent Technologies followed.

But delays plagued implementation. By the time Chinese factories came online with foreign technology, that technology was already outdated. The industry's pace simply outran China's ability to adopt.

Project 909 and the DRAM Disaster

The ninth Five-Year Plan, from 1996 to 2000, tried something more ambitious: Project 909. The goal was for a domestic firm called Huahong to produce internationally competitive memory chips using Chinese intellectual property and engineers.

Huahong partnered with NEC, the Japanese electronics giant, and actually succeeded in entering production on schedule. But there was a problem—the partnership relied heavily on Japanese expertise, limiting how much knowledge actually transferred to Chinese engineers.

Then the market turned. The global DRAM market—DRAM stands for dynamic random-access memory, the working memory in computers—crashed around 2002. Memory chips are notoriously cyclical commodities. Prices swing wildly based on supply and demand. Huahong, a newcomer without the cost advantages of established players, suffered massive financial losses.

This pattern—entering a market segment just as it turns down—would haunt Chinese semiconductor efforts repeatedly.

The Big Fund

In 2014, China tried yet another approach: the China Integrated Circuit Industry Investment Fund, commonly called the "Big Fund."

The Big Fund pools resources from major state investors including the Ministry of Finance, China Tobacco (yes, the tobacco monopoly is a major investor), China Mobile, and China Development Bank. Its purpose is straightforward: provide the massive capital that semiconductor manufacturing requires.

Building a modern chip fabrication plant—a "fab" in industry jargon—costs somewhere between $10 billion and $20 billion. The most advanced facilities cost even more. These are among the most expensive factories ever built by humans. The Big Fund was designed to write the enormous checks this industry demands.

The strategy showed results. By 2021, China led the world in new fab construction, with eight out of nineteen fabs being built worldwide located in China. Between 2021 and 2023, seventeen additional fabs were expected to break ground.

The American Hammer Falls

On October 7, 2022, the United States government announced the most comprehensive export restrictions on semiconductor technology ever imposed on any country.

The restrictions went far beyond previous sanctions. They targeted not just American-made equipment, but any equipment made anywhere in the world using American technology. Given how thoroughly American intellectual property permeates the global semiconductor supply chain, this effectively meant nearly all advanced chipmaking equipment.

The restrictions specifically targeted artificial intelligence and advanced semiconductor technologies. American citizens and permanent residents were even prohibited from working for certain Chinese semiconductor companies.

In January 2023, Japan and the Netherlands—home to key equipment makers Nikon, Canon, and ASML—joined the restrictions, making them multilateral.

China responded on multiple fronts. It filed a suit at the World Trade Organization. In December 2023, it banned Intel and AMD processors from Chinese government computers and servers, instead requiring eighteen domestically-designed processors from companies called Loongson and Phytium. State-owned enterprises were instructed to transition entirely to Chinese hardware by 2027.

The Companies Fighting the Battle

Several Chinese companies stand at the center of this technological contest.

SMIC: The Foundry Giant

Semiconductor Manufacturing International Corporation, or SMIC, is China's largest contract chip manufacturer and the fifth largest globally. It's partially state-owned and headquartered in Shanghai, though incorporated in the Cayman Islands—a common arrangement for Chinese companies seeking foreign investment.

SMIC can manufacture chips from 350 nanometers down to 14 nanometers. To put that in perspective, the most advanced chips today are made at 3 nanometers. A nanometer is one billionth of a meter—about the width of a DNA double helix. The smaller the number, the more transistors you can pack onto a chip, and the more powerful the chip becomes.

SMIC is roughly two to three generations behind the global leaders. But that gap has implications beyond mere competitiveness. It means China cannot domestically manufacture the most advanced artificial intelligence chips, the cutting-edge processors for smartphones, or the latest graphics cards.

YMTC: The Memory Maker

Yangtze Memory Technologies Corporation—YMTC—represents China's attempt to crack the flash memory market. Flash memory is the storage in your phone, your laptop's solid-state drive, and countless other devices. Before YMTC was founded in 2016, China had no company capable of producing it.

YMTC made rapid progress. By 2020, it was producing 64-layer 3D NAND flash memory using a 20-nanometer process. In April 2020, it unveiled a 128-layer chip—at the time, the most advanced layer count in mass production anywhere in the world.

The company uses a proprietary technology called XTacking architecture, which allows it to achieve competitive specifications through clever engineering rather than brute-force shrinking of transistors. It's an example of China finding alternative paths when the direct route is blocked.

HiSilicon: Huawei's Crown Jewel

HiSilicon is perhaps the most dramatic case study in China's semiconductor saga. A fabless chip designer wholly owned by telecommunications giant Huawei, HiSilicon became mainland China's largest integrated circuit designer.

Then the sanctions hit. In 2020, the United States instituted rules requiring any company using American technology to get licenses before supplying HiSilicon. Since modern chip manufacturing is essentially impossible without American technology somewhere in the supply chain, this effectively cut HiSilicon off from production.

Huawei announced it would stop producing its Kirin smartphone chips on September 15, 2020. HiSilicon was overtaken by rival UNISOC in mobile processor market share.

The story didn't end there. At the end of 2023, Huawei released the Mate 60 smartphone with a Kirin 9000S processor—a chip made entirely with domestic Chinese production. HiSilicon was back, and this time without any foreign dependencies. The chip wasn't as advanced as leading competitors, but its very existence sent shockwaves through the industry.

The Memory Players

ChangXin Memory Technologies, or CXMT, is China's answer in the DRAM market. Based in Hefei, it produces the working memory that computers use for active tasks—different from the flash storage that YMTC makes. As of 2020, ChangXin could manufacture DDR4 and LPDDR4 memory on a 19-nanometer process, with plans to launch next-generation DDR5 memory.

These memory companies matter because memory is both strategically important and commercially enormous. Samsung's two plants in Xi'an alone account for 15.3 percent of worldwide NAND production capacity—a reminder that foreign companies still dominate even within China's borders.

The Lithography Problem

At the heart of semiconductor manufacturing sits the lithography machine. This is the device that projects circuit patterns onto silicon wafers, like a photographic enlarger in reverse—it shrinks patterns rather than enlarging them.

The most advanced lithography machines use extreme ultraviolet light, or EUV. These machines are made by exactly one company on Earth: ASML, based in the Netherlands. An EUV machine costs around $150 million and is arguably the most complex device humans have ever manufactured. ASML is the sole supplier because no one else has figured out how to make them work.

Under the export restrictions, China cannot buy EUV machines. This is the single biggest bottleneck in China's semiconductor ambitions.

In September 2024, China announced it had developed two new deep ultraviolet (DUV) lithography machines—one generation older than EUV. One operates at 193 nanometers with resolution below 65 nanometers. These aren't competitive with ASML's latest offerings, but they represent progress toward self-sufficiency.

Then came the December 2025 Reuters report about a prototype EUV machine in Shenzhen. If true—and such reports are difficult to verify—it would represent a remarkable achievement. The report suggested working chips might emerge between 2028 and 2030.

The Supporting Cast

China's semiconductor industry extends beyond the headline names.

JCET Group, formed in 1972 when the city of Jiangyin converted a local factory to produce transistors, is now the largest semiconductor assembly and testing company in mainland China and third largest globally. It handles the crucial but unglamorous work of packaging and testing chips after they're manufactured.

NAURA Technology Group is China's largest semiconductor equipment manufacturer. Advanced Micro-Fabrication Equipment, or AMEC, is another major player. These companies are building the machines that make the machines that make the chips—a supply chain within a supply chain.

Among fabless designers, UNISOC has emerged as a major mobile processor manufacturer, ranking fourth globally behind MediaTek, Qualcomm, and Apple with 9 percent market share. Loongson Technology develops processors for computers and supercomputers, favored by the government for its "Made in China 2025" push.

Zhaoxin represents an unusual story: a joint venture between Taiwan's VIA Technologies and the Shanghai municipal government, creating x86-compatible processors. The x86 architecture, developed by Intel, dominates personal computers. Having a domestic source matters for government and military systems that cannot rely on foreign chips.

Where Things Stand

The numbers tell a story of both progress and distance remaining.

China has set a goal of 70 percent domestic chip production. Chinese foundries are running at high utilization rates—around 87 percent capacity expected in 2025—driven by a "Design by China, Manufacturing in China" policy. Some business is actually shifting from China to Taiwan under "China+1" sourcing strategies by multinational companies hedging their bets.

The gap in leading-edge technology remains significant. SMIC at 14 nanometers versus TSMC and Samsung at 3 nanometers is not a small difference—it represents roughly a decade of industry progress.

But technology is not static. The EUV prototype report, if accurate, suggests China is not accepting the export restrictions as permanent limitations. The resurrection of HiSilicon's Kirin chips demonstrates that domestic supply chains can be built even under pressure.

The semiconductor industry exists at an unusual intersection: critical for everything from smartphones to weapons systems, concentrated in just a few companies and countries, and advancing so rapidly that yesterday's cutting edge becomes tomorrow's commodity. China's campaign to master it represents one of the largest directed technological efforts in history, comparable in ambition if not in method to the Manhattan Project or the Apollo program.

The outcome remains uncertain. What's clear is that both sides—China pushing for self-sufficiency and the United States pushing to maintain technological advantage—have decided this industry is worth fighting over. The chips that result will power the artificial intelligence, military systems, and consumer electronics of the coming decades. The question of who makes them may shape the century.