Our Work

Semiconductors info series header

The Global Semiconductor Supply Chain: Four Phases

The semiconductor supply chain can be divided into four general phases: R&D, Design, Manufacturing/Fabrication, and Assembly, Test, Packaging (ATP)/Advanced Packaging. Throughout these phases, semiconductors require key material inputs and unique equipment that further complicates the supply chain. Much of the supply chain is also concentrated in a handful of countries, posing challenges both to supply chain resiliency and U.S. national security.

4 phases of Semiconductor supply chains

The following table provide a summarized assessment of the four main phases in the semiconductor supply chain. Phases highlighted in blue are those in which the U.S. has secure domestic capabilities while phases highlighted in red are those in which U.S. capabilities are minimal and vulnerable.

Supply Chain Phase Main Countries Involved
Phase 1: Research and Development U.S. firms are world leading, but China is rapidly attempting to advance capabilities.
Phase 2: Design  U.S. firms are world leading, but China is rapidly attempting to advance capabilities.
Phase 3: Manufacturing/Fabrication Fabrication occurs in Taiwan (20% of global production), South Korea (19%), Japan (17%), China (16%), U.S. (12%). The most advanced chips (5nm and 7nm) are made only in Taiwan, South Korea, and Japan. The U.S. can only produce 10nm and China can only produce 14nm, but both the U.S. and China are aiming to enhance their production capabilities.
Phase 4: Assembly, Testing, Packaging (ATP) and Advanced Packaging

ATP: China, Taiwan, and Southeast Asia possess nearly all global capacity. The U.S. has just 3% of global ATP capacity.
Advanced Packaging: Highly concentrated in Taiwan and South Korea.

For both ATP and Advanced Packaging, China is attempting to advance its capacity.

     
Phase 1: Research and Development (R&D)

The semiconductor industry requires significant research and development (R&D) investments in order to continuously innovate chip design for more advanced technologies like AI, IoT, and autonomous vehicles. In 2019, the industry globally invested approximately $90 billion in R&D (20% of global semiconductor sales), of which $72 billion was contributed by the U.S., and experts estimate that the industry will need to invest another $3 trillion in R&D over the next 10 years to meet rising technological demands. The U.S. Innovation and Competition Act, if passed, would invest $52 billion for domestic semiconductor research and design.

Presently, the U.S. is the global leader in semiconductor R&D due to its highly educated workforce and innovative ecosystem. In 2019, six of the seven leading companies in semiconductor R&D intensity were U.S.-based. However, China has been aggressively investing in semiconductor R&D. China has surpassed the U.S. in filing the largest total number of semiconductor academic research papers and patents annually and China’s 2021-2025 Five Year Plan increased central government research spending by 11%, with semiconductors designated as one of the seven areas prioritized for funding and resources.

Phase 2: Design

In the chip design phase, engineers must devise and assemble a collection of interconnected circuit elements to perform a specific function. Engineers receive a request containing requirements for what function or task the chip must complete. They then create the logic/circuit design which contains memories, processing units, sensors, and other technical building blocks necessary for the circuit to function. Engineers will then complete the physical design and layout on the chip and test the prototype to ensure it can carry out the required functions. Once verified, the chip design can be sent out to manufacturers for mass production.

The design phase requires highly advanced technical know-how and a highly educated and trained workforce. The U.S. is therefore the global leader in this phase. 10 out of the top 20 global semiconductor design companies and four out of the top five global electronic design automation (EDA) and semiconductor intellectual property (IP) companies are headquartered in the U.S. Leading U.S. semiconductor design firms include Qualcomm, Intel, Broadcom, and AMD. The U.S. Commerce Department assessed in 2021 that the “U.S. design ecosystem is robust and world leading.”

However, China has taken steps to increase its control of semiconductor IP and advance its design capabilities. In 2017, there were nearly 1,400 chip design companies in China, triple the number that had existed just four years earlier. Collectively, these companies had $31.9 billion in revenues in 2017. While these companies are still reliant on design tools and software from foreign companies and lack the capabilities to autonomously design higher-end chips, China is making a significant push for self-sufficiency in this area.

Phase 3: Manufacturing/Fabrication

In the semiconductor manufacturing (also referred to as fabrication) phase, fabrication facilities (typically called ‘fabs’ or ‘foundries’) make disc-shaped silicon wafers into individual microchips according to the specified design, with each chip generally the size of a fingernail. As the U.S. Commerce Department notes, the fabrication process is “complex and highly specialized…a semiconductor manufacturing plant involves thousands of process machines, lasers, ultra-precision optics, and advanced robotics. The fabrication process is one of the most advanced in the world, involving cutting-edge techniques and equipment, operating at subatomic-level precision.” Due to this complexity, the fabrication phase is highly capital intensive. Constructing an advanced foundry can range from $5 billion to $20 billion in initial costs. Fabrication facilities also require inputs of ultrapure gases, ultrapure water, dry air and nitrogen, and high-quality reliable electrical power. A single fabrication facility can use 100 MW of power (more than a standard auto plant) and as much water as small city. Therefore, attempting to build new fabrication facilities is incredibly costly and is only reasonable in areas with the requisite resources, power, and skilled labor.

While U.S.-headquartered companies such as IDM, Intel, Analog Devices, Maxim Integrated Productions, Microchip Technology, Micron, ON Semiconductor, and Texas Instruments are global leaders in fabrication, most of their actual manufacturing capacity has been off-shored, primarily to East Asia. Therefore, while U.S. semiconductor companies account for 47% of global chip sales, only 12% of global chip manufacturing actually takes place in the U.S., a sharp drop from the 37% of global manufacturing capacity based in the U.S. in 1990.

In terms of absolute production capabilities, approximately 75% of chip fabrication is in East Asia, with global production distributed as follows:

  • Taiwan: 20%
  • South Korea: 19%
  • Japan: 17%
  • China: 16%.
  • U.S.: 12%
  • Europe: 9%
  • Elsewhere (such as Singapore and Israel): 6%

Semiconductors Global Fabrication CapacityTaiwan in particular is recognized as the global leader in fabrication, especially for the contract foundry market (the non-vertically integrated market, where firms that develop designs contract out production to a foundry to produce the microchips) where it controls 63% of the global market share and a single company, TSMC controls 53% of the global market share. This means that new entrants to the semiconductor market who either only design advanced chips or who require chips for their advanced products but lack the financial resources to build a completely in-house vertical production line for microchips will have to go through Taiwan (or South Korea, which controls 18% of the global contract market) for fabrication.

Equally important to absolute production capacity is the ability to produce the most advanced types of chips. The most advanced chips at 5nm can only be produced by TSMC in Taiwan and Samsung in South Korea. The most advanced chips able to be made in the U.S. are just 10nm (Intel) and China’s most advanced capabilities are only in 14nm chips.

China is rapidly attempting to enhance its fabrication capabilities, even amidst U.S. efforts to impede them. The Trump Administration blacklisted China’s most advanced semiconductor company, Semiconductor Manufacturing International Corporation (SMIC) and pressured other countries such as the Netherlands to block the sale of advanced technology and machinery to SMIC to prevent China from advancing its semiconductor capabilities.

Despite these efforts, China’s share of global fabrication capacity is expected to grow to 28% by 2030. The Chinese government is devoting $100 billion in subsidies to the industry and is developing 60 new manufacturing facilities. In 2019 alone, of the six new semiconductor production facilities in the world, four were in China and none were in the U.S. As a result, the Semiconductor Industry Association (SIA) predicts that by 2030 the U.S. share of global semiconductor manufacturing capacity will fall to 10% while East Asia’s will grow to 83%.

Phase 4: Assembly, Test, Packaging (ATP) and Advanced Packaging

After fabrication, the chips are assembled into finished semiconductor components, tested, and packaged for incorporation into finished products. This process is less capital intensive than the fabrication phase and requires large floor space but only low-tech workers, resulting in ATP facilities mainly being located in areas with low labor costs.

Approximately 60% of global ATP capacity is in China and Taiwan, with the rest largely in Singapore, Malaysia, the Philippines, and Vietnam. Of 360 global ATP facilities, 100 are in China, 100 are in Taiwan, and 43 are in Southeast Asia. While U.S.-headquartered companies have 28% of the market share of ATP revenues, nearly all of the actual ATP production from U.S. companies is outsourced to East and Southeast Asia. The U.S. only has 3% of global semiconductor packaging capacity (excluding testing capacity).

As an alternative to the standard ATP process, Advanced Packaging is when multiple chips and/or more than one integrated circuit are combined into one package. This allows for higher chip density, the integration of different chip functions in a single package, and an increased ability to use commercial ‘off-the-shelf' chips for custom solutions. Advanced Packaging requires more capital and higher-skilled labor than traditional ATP. The U.S. does have a small amount of Advanced Packaging capabilities through companies like Intel, Texas Instruments, Micron, and GlobalFoundries. However, the bulk of Advanced Packaging occurs in Taiwan and South Korea. While China currently lacks strong Advanced Packaging capabilities, it has made developing and expanding this technology area a key priority since 2014.

 

Previous: Part 1, Overview

BENS National Office | 1030 15th Street NW, Suite 200 East | Washington, D.C. 20005
Phone: (202) 296-2125 | Fax: (202) 296-2490 | Email: bens@bens.org

© 2024 Business Executives for National Security | Privacy Policy | Donor Privacy Policy