Semiconductor Cold War - U.S.–China Capacity Race and Global Supply Chain Restructuring

Semiconductor Cold War: U.S.–China Capacity Race and Global Supply Chain Restructuring

Competition in global semiconductor and related technologies has become a core element of great power rivalry. While the United States still maintains a leading position in the global semiconductor industry, with its companies accounting for nearly half of the global semiconductor market revenue, this dominance is now being challenged. In 2024, China's semiconductor sales accounted for 20% of global sales, and the country has proposed further expansion plans to achieve deeper domestic substitution, aiming to achieve 50% self-sufficiency in semiconductor production within a few years. The United States has responded with the slogan of "reindustrialization," revitalizing its domestic semiconductor industry. The trade war, initially characterized by tariffs and accusations, has branched out into a new battle of production capacity and technology.

A 30-Year Supercycle: Semiconductors Continue Expanding

Over the past three decades, semiconductor industry revenue has fluctuated at times but has overall maintained strong, rapid growth, with each trillion-dollar milestone being reached increasingly quickly. According to WSTS, global semiconductor sales (shipments of integrated circuits, discrete devices, and optoelectronics) reached USD 630.5 billion in 2024—exceeding prior expectations and surpassing the USD 600 billion mark for the first time. Driven by rising chip demand in AI, automotive, and industrial applications, average monthly semiconductor sales in 2024 grew more than 15% compared to 2023. Regionally, annual sales grew in the Americas (+45.2%), China (+20.0%), and Asia Pacific/Other (+12.2%), while Japan (−0.3%) and Europe (−8.1%) saw declines.

In 2024, multiple semiconductor product categories related to high-performance computing performed exceptionally well. Logic device revenue reached USD 215.8 billion, ranking first among all categories. Memory revenue ranked second and surged 78.9% in 2024. DRAM— a subset of memory—grew 82.6%, the fastest of all product categories.

Downstream Innovation Helps the Market Surpass Cycles

Across the semiconductor value chain, upstream includes materials, equipment, and EDA tools; midstream includes design, manufacturing, and packaging/testing; downstream applications span consumer electronics, communications, industrial, and automotive. The semiconductor industry has clear cyclic-growth characteristics, with high-growth phases driven by innovation from downstream end markets. According to InsightAce Analytics, semiconductors— as the backbone of data centers, AI, autonomous vehicles, smartphones, and other emerging technologies—are expected to surpass USD 1 trillion by 2030.

Future demand will be driven by long-term trends across multiple industries, forming a diversified and parallel growth pattern. The table below illustrates core downstream trends and corresponding semiconductor needs:

PwC data shows that the semiconductor industry is shifting from “consumer-electronics-driven” to diversified, high-value applications in computing, networking, industrial, and automotive. Among these, servers/networking and automotive have become the core engines of long-term growth:

(1) Servers and networking lead with an 11.6% CAGR, driven by cloud computing, data-center expansion, AI accelerators, and 5G infrastructure.

(2) Automotive semiconductors (CAGR ~10.7%) are rising with EV adoption, ADAS, and software-defined vehicles, driving demand for SiC/GaN devices, sensors, and AI/MCU processors.

U.S. Pushes Domestic Capacity Expansion—Against the Trend

Since the late 1990s, the U.S. has remained the global leader in semiconductor sales. In 2024, U.S.-headquartered semiconductor companies generated USD 318 billion in revenue, representing 50.4% of global sales. The U.S. also leads in R&D: total semiconductor R&D investments reached USD 62.7 billion in 2024, up ~5.7% YoY, representing 17.7% of industry revenue—second only to pharmaceuticals/biotech.

In exports, U.S. semiconductor shipments reached USD 57 billion in 2024, ranking sixth among all export categories. Notably, about 70% of U.S. semiconductor industry revenue comes from overseas markets. 

However, the U.S. faces structural imbalances: U.S. semiconductor companies account for 57% of global wafer demand, but only ~10% of global wafer manufacturing capacity is located in the U.S. Most production relies on fabs in Mainland China, Taiwan, and Japan.

After decades of declining domestic manufacturing share, the U.S. government is driving a new wave of domestic industrialization. As of July 2025, companies have announced more than USD 500 billion in private-sector investments across the U.S. semiconductor ecosystem. The Semiconductor Industry Association (SIA) projects U.S. fab capacity will triple by 2032. These projects are expected to create or support more than 500,000 U.S. jobs (68,000 facility jobs, 122,000 construction jobs, and 320,000 jobs across the broader economy).

In July 2025, the U.S. enacted legislation strengthening the Advanced Manufacturing Investment Credit (AMIC), increasing the credit rate from 25% to 35%. SIA also supports extending this credit beyond 2026 and expanding it to include R&D, accelerating re-industrialization and attracting further investment.

Despite ambitions to expand domestic semiconductor production capacity, multiple factors are hindering this expansion: on the one hand, the world's leading chip design industry, dominated by companies like Apple, Broadcom, and Nvidia, is facing challenges from overseas; on the other hand, US wafer fabs are facing a severe shortage of manufacturing talent. According to a joint study by the SIA and Oxford Economics, the US is facing a significant shortage of technicians, computer scientists, and engineers. It is projected that by 2030, the semiconductor industry alone will be short 67,000 professionals, while the overall technical talent gap across the economy will reach as high as 1.4 million.

China Accelerates to Catch Up—Poised to Lead Foundry Capacity

Like the U.S., China is investing hundreds of billions of dollars in subsidies to build a self-sufficient chip ecosystem, aiming to secure AI leadership and strengthen national security.

According to forecasts, China's investment in wafer manufacturing equipment already accounts for 30% of global investment. Combined with an expansion rate of four to five new wafer fabs annually, this investment will propel China to become a global leader in wafer foundry services by 2030. Currently, China already has SMIC, one of the world's top five foundries, with a market share of approximately 5%. Other Chinese foundries, such as Hua Hong Semiconductor and Yangtze Memory Technologies (YMTC), are rapidly expanding with the help of new government subsidies.

In fact, China’s current wafer manufacturing capacity has reached 21% of the global total, a level that is roughly on par with South Korea and is approaching the level of the Taiwan region. In comparison, Europe and Japan maintain a stable supply–demand balance in semiconductor foundry services, with a large portion of their capacity used to meet internal market demand. In Southeast Asia, especially Singapore and Malaysia, the region accounts for 6% of global foundry capacity, but these capacities are entirely dominated by foreign foundries, as the region lacks local semiconductor foundry enterprises.

There is no shortage of institutions optimistic about the outlook for mainland China’s wafer foundry industry. According to data released by the Semiconductor Equipment and Materials International (SEMI) in June 2024, global semiconductor fab capacity is expected to achieve year-on-year growth of 6% and 7% in 2024 and 2025 respectively, reaching a historic high of 33.7 million 8-inch wafer equivalents per month in 2025. Meanwhile, a 2025 report by Yole Group also indicates that mainland China is likely to become the world’s largest semiconductor foundry center by 2030.

In addition to capacity expansion, China’s semiconductor industry is gradually moving toward more advanced process technologies. Currently, Semiconductor Manufacturing International Corporation (SMIC), Hua Hong Semiconductor, and Nexchip have all entered the global top ten foundries. Among them, SMIC has achieved breakthroughs in sub-10-nanometer advanced process nodes, and since 2022 has invested more than USD 7 billion annually in equipment procurement and R&D, demonstrating strong momentum in technological catch-up.

Downstream Trends: Five Innovations Shaping the Semiconductor Industry

The future development of the semiconductor industry will not only rely on capacity expansion, but more importantly on finding new footholds in emerging innovative applications to achieve differentiated competition and value creation. For the segments with the greatest market potential in the future, market consensus is concentrated in five major directions: artificial general intelligence (AGI), advanced autonomous driving, humanoid robots, quantum computing, and brain–computer interfaces. These frontier applications not only place unprecedented technical requirements on computing-power density, energy-efficiency ratios, real-time communication, and multimodal sensing, but are also driving full-chain innovation and upgrades in semiconductors—from architecture to materials. At the same time, these fields have become among the most concentrated directions of R&D investment in countries such as China and the United States, and their influence is profoundly reshaping the technological roadmaps, ecosystem structure, and capital flows of the semiconductor industry.

1. Artificial General Intelligence (AGI)

The development of artificial intelligence is limited less by algorithms than by two practical constraints: first, the need for massive amounts of high-quality data, and second, the need for powerful computing capabilities to process that data. Driven by clear evidence of AI’s value, investment into these two areas continues to accelerate. Next-generation semiconductor technologies are the core support for achieving AGI. Large-scale AI models require logic chips that are faster and more energy-efficient, combined with high-density 2.5D or 3D packaging technologies to enhance computing performance.

High-bandwidth, low-latency memory is also crucial when training massive datasets. R&D directions are gradually shifting towards neuromorphic computing and processing-in-memory (PIM). Domain-specific neural processing units (NPUs) have already been applied in edge devices, while PIM places computational units next to or inside DRAM to reduce energy consumption and latency for data movement. In the future, fully neuromorphic hardware may simulate brain architectures, significantly improving computation per watt and per cubic centimeter, providing a new semiconductor foundation for AI development.

2. Autonomous Driving

As advanced autonomous driving technology progresses, future vehicles will see rapidly increasing demand for sensors, AI processors, and power semiconductors, making semiconductors the core support for mobility. By the 2030s, as more new vehicles with high-level autonomous driving capabilities reach the market, the demand for sensors, AI processors, and power devices may continue rising, making semiconductors central to future mobility.

After full autonomous-driving commercialization, the number of semiconductors required per vehicle will increase significantly. Currently, traditional vehicles contain about 200–300 semiconductors, while Level-3 and above vehicles may exceed 1,000 semiconductors. This trend not only expands the automotive semiconductor market, but also raises the requirements for high-performance processors, sensors, and power-management chips, driving structural upgrades across the semiconductor industry.

3. Humanoid Robots

Global population aging and labor shortages are accelerating the development of robotic technologies. Robots can operate around the clock with minimal downtime, working roughly three times longer than humans, and their application scenarios have expanded from food service to healthcare, security, and home care. With the advancement of AI and autonomous technologies, humanoid robots can perform complex movements and learn autonomously in real time, and will increasingly integrate into both enterprise and home environments in the future.

Progress in robotics heavily depends on semiconductors, including processors, sensors, and micro-electromechanical systems (MEMS). AI processors handle decision-making and real-time data analysis, supporting edge computing so robots can operate independently even with unstable networks. CMOS image sensors, ToF and LiDAR technologies enable environmental perception, MEMS sensors enhance motion precision, and PMICs and power semiconductors improve stability. Semiconductor innovation directly determines the intelligence level and application depth of robots.

4. Quantum Computing

Quantum computers, leveraging superposition and entanglement, can explore multiple computational paths simultaneously, surpassing classical computers in problems such as large-number factorization and molecular simulation. Despite the promising prospects of quantum computing, in the short term it still relies on traditional semiconductor chips for control, qubit stabilization, and quantum error correction (QEC). “Hybrid quantum computing” combines quantum computers with supercomputers to solve practical problems. The development of quantum computing will drive demand for high-performance, low-latency semiconductors and boost markets for memory, control chips, and accelerators, opening new frontiers for semiconductor applications.

5. Brain–Computer Interfaces (BCI)

The concept of brain–computer interfaces relies on electrodes to detect brain signals and uses electronic circuits to enable communication between the brain and external devices. Because brainwaves represent massive and extremely complex data inputs, processing them requires advanced ultra-low-power AI accelerators, analog-to-digital converters, and amplifiers. This is especially important for non-invasive BCIs, where brain signals are weak and require precise amplification.

For invasive BCIs, which often require long-term implantation, minimizing size, heat generation, and power consumption is crucial. Biocompatible materials and packaging methods are also essential to ensure safety and functionality inside the body. Therefore, the demand for application-specific integrated circuits (ASICs) tailored for BCIs is likely to increase. In addition, low-latency processing of digitized brain signals may drive demand for system-on-chips (SoCs) incorporating AI accelerators and digital signal processors (DSPs). These chips are responsible for interpreting data and converting it into intentions and actions, requiring advanced low-latency performance. Furthermore, transmitting brain signals to external receivers requires low-power, short-range communication chips such as RFICs or Bluetooth Low Energy (BLE).

About US:

Bosson Research (BSR) is a leading market research and consulting company, provides market intelligence, advisory service and market research reports for the automobile, electronics and semiconductor, and consumer good industry. The company assists its clients to strategize business policies and achieve sustainable growth in their respective market domain.

Bosson Research provides one-stop solution right from data collection to investment advice. The analysts at Bosson Research (BSR) dig out factors that help clients understand the significance and impact of market dynamics. Bosson Research (BSR) bring together the deepest intelligence across the widest set of capital-intensive industries and markets. By connecting data across variables, our analysts and industry specialists present our customers with a richer, highly integrated view of their world.

Contact US:

Tel: +86 400-166-9288

E-mail: sales@bossonresearch.com

URL: www.bossonresearch.com