Sat, Jul 09, 2022 page8

• Japan cannot go it alone on chips

  • By Lin Hsiu-min 林修民
  • • 
    • 
    • 

  To advance in the production of next-generation commercial semiconductor chips, Japan has agreed to work with the US on a Japanese manufacturing base for 2 nanometer semiconductors by 2025, a report in the Nikkei Asia said last month.

  Washington and Tokyo have committed to technology collaboration, while the commercial sector is to pursue research on design and mass production.

  The most feasible approaches would be to create a joint venture company between Japan and the US, or for Japan to establish a new domestic manufacturing base funded by the Japanese Ministry of Economy, Trade and Industry.

  The move appears to be a way to counter Taiwan Semiconductor Manufacturing Co (TSMC 台積電), but Japan would be hard pressed to establish a 2 nanometer semiconductor manufacturing base. Its attempt would would be a waste of resources and leave TSMC’s status unchallenged.

  The Japanese government wishes to produce its own computing chips for data centers and flagship smartphones, but the most it can hope to achieve is to facilitate research and patents, which falls far short of its stated goals.

  The Nikkei Asia called the pursuit of research on design and mass production an art. If practiced in the right industry, the range of applications can be quite broad. For semiconductor products, there is a long way to go before a laboratory result fit for research publication becomes fit for mass production. Labs only need to take engineering science into account when creating samples, but whether a product is fit for mass production relies solely on the market.

  The lower a product’s yield rate, the higher the unit cost. When the product price exceeds the price the market is willing to bear, it cannot enter mass production. There are many examples of this in the semiconductor industry, the most well-known being Intel’s 14 nanometer and 10 nanometer chips. As the yield rate for the chips failed to meet the standard for mass production over several years, Intel handed the advanced process technology to TSMC after suffering delays in planned products.

  I have absolute confidence in Japan’s research and development expertise. Like Intel, Japan’s memory and Intel’s CPU technology were the bellwethers in the 1990s.

  However, Japan’s efforts to reduce mass-production costs were widely doubted outside Japan, and even criticized by the Japanese themselves, who regarded their country’s inability to achieve low-cost mass production as the main reason for the downfall of its semiconductor industry. This was recounted in Takashi Yunogami’s 2013 book The Defeat of Japanese-style Manufacturing.

  In the Nikkei Asia report regarding the industry ministry’s plans, there are false expectations that the mere establishment of a 2 nanometer manufacturing plant could actually lead to the production of chips for data center supercomputers or flagship smartphones. Merely making 2 nanometer circuits is far from enough to produce computer chips used at data centers for ballistic simulation.

  The current technology is to use chiplets with advanced packaging technology. Even pure computing chips require at least memory connection technology, just as TSMC applies chip on wafer on substrate technology, or Intel’s use of embedded multi-die interconnect bridge technology to connect chips or memories of different process technologies. Even flagship smartphone chips need advanced packaging, such as the integrated fan-out technology used on the iPhone 7 and later models.

  An advanced chip also requires the support of computer aided design and less-advanced packaging technologies to support different module functions. Therefore, the business is far more complex than a transistor or static random access memory meeting the 2 nanometer gate or fin width.

  Japan is also lacking in IC related industries — Japan’s current forte in the semiconductor industry is equipment and materials. The industry ministry’s connection to national security also poses concern. As the Japanese government has said, advanced 2 nanometer chips are to be used for data center computing and flagship smartphones’ connection to national security.

  The problem is that Japan does not have that chip designing enterprises. Even with the establishment of a 2 nanometer manufacturing factory, it would rely on the US or Taiwan for design orders. The US is responsible for the design of data center computing chips, mainly using the x86 architecture, while Taiwan, the US and South Korea are responsible for flagship smartphone chips.

  However, none of the factories in the current industries are Japanese factories. If the Japanese government wishes to achieve complete autonomy to boost national security, then it will have to establish large-scale chip designing factories.

  Compared with chip foundry below 7 nanometers, Taiwan dominates 90 percent of the global market for production of the most advanced chips. If anything happens to Taiwan, the international advanced chip supply chain would be severely disrupted. As a result, the Japanese government decided to establish its own advanced chip factory.

  An invasion of Taiwan would be a grave danger to Japan, former Japanese prime minister Shinzo Abe said in December last year, given that the Senkaku Islands — known as the Diaoyutai Islands (釣魚台) in Taiwan — the Sakishima Islands and Yonaguni Island are only about 100km from Taiwan.

  If Taiwan falls into Chinese hands, not only would Japan’s 2 nanometer chip factory become a primary target, Japan’s reliance on a crude oil transportation route would also be threatened.

  However, analysts told the Nikkei Asia that China’s placement of a Boeing E-767 Airborne Warning and Control System aircraft in Xinjiang is preparation for attacking Japan. It is obvious that if China invades Taiwan, Japan and its Kadena Air Base could be targets as well. Japan’s proposed 2 nanometer wafer factory would only add one more target for China to consider.

  Japan’s advanced research and development skills provide the country with the time and money to establish an advanced chip factory.

  However, Japan’s intended budget for the semiconductor industry, including the advanced factory — ￥600 billion (US$4.42 million) — is insufficient to produce advanced chips autonomously, given that this budget also includes data center computing and flagship smartphone chip design.

  Taiwan is home to the world’s most complete semiconductor industry clusters and specializations. Not only does this country have advanced chip manufacturing factories and the supply of related equipment and materials, it also has an upstream wafer supply and downstream semiconductor exports for the information and communication industry.

  The most efficient way for Japan to stabilize national security and achieve economic efficiency is to strengthen its partnership with Taiwan instead of building an advanced semiconductor factory.

  Lin Hsiu-min is an adjunct lecturer in Soochow University’s Department of Business Administration.

  Translated by Rita Wang
• Most Popular

  • 1

    Large bamboo viper catch in Tainan puzzles bureau

  • 2

    Danish envoys’ privileges revoked over naming controversy

  • 3

    Horse euthanized after winning the Grand National race

  • 4

    Eight million barrels of oil to set sail for Taiwan this month

  • 5

    Foreign exchange reserves fall below US$600 billion