Responding to Japanese demands for artificial language (AI) services, AI giant OpenAI announced on April 15th the opening of a new office in Tokyo as well as the rollout of a new custom language model tailored to the Japanese language. The move highlights the meteoric demand for AI services as well as the continuation of robust US-Japan economic ties.
Following meetings between OpenAI CEO Sam Altman and Japanese Prime Minster Fumio Kishida in 2023, OpenAI has recently announced that Tokyo will be their first office in the Indo-Pacific region. Japan’s proactive approach to AI usage cultivates an environment that facilitates the adoption of the new customized language model tailored to the Japanese language.
OpenAI has expressed excitement to be opening its office in Tokyo, citing the city’s “global leadership in technology, culture of service, and a community that embraces innovation.” OpenAI plans to introduce their new ChatGPT-4 model optimized for the Japanese language. The new model has improved abilities to translate and summarize Japanese text and “operates up to 3x faster than its predecessor.” Tadao Nagasaki, former president of Amazon Web Services (AWS) Japan, will serve as the president of the subsidiary. Japanese companies like air conditioner manufacturer Daikin and financial technology company Rakuten have already adopted a version of ChatGPT for businesses: ChatGPT Enterprise. ChatGPT has also helped local governments like Yokosuka City to adopt AI and increase city employee productivity.
The announcement comes after Altman’s meeting with Kishida in 2023, during which Altman expressed interest in opening an office in Tokyo. The two discussed various matters during the meeting, including safety. Chief Cabinet Secretary Hirokazu Matsuno voiced concerns about privacy and cybersecurity while Altman emphasized OpenAI’s dedication “to work with the government to protect user privacy and safety.” The meeting was arranged by Liberal Democratic Party (LDP) lawmakers, who were preparing to discuss the adoption of AI during the then-approaching G-7 summit. The LDP AI project team has issued a white paper calling for research and development of AI, including ChatGPT.
During the G-7 Summit in Hiroshima, Japan introduced the “G7 Hiroshima AI Process” initiative, aiming to be a trailblazer in the adoption of responsible AI. Through this, G-7 countries discussed international regulation and rulemaking on generative AI.
OpenAI and Semiconductors
Essential to the learning process of generative AI are high-performance graphics processing units (GPUs), compelling OpenAI’s involvement in the growing global semiconductor industry. The current undisputed leader in GPUs is NVIDIA: as of 2023’s fourth fiscal quarter, the trillion-dollar American tech behemoth has captured roughly 80% of the global GPU market share. In fact, NVIDIA GPUs account for more than 90% of the GPUs on which neural network training occurs, the process by which generative AI iterates, optimizes, and self-improves.
However, the meteoric surge in interest for these high-performance GPUs is straining NVIDIA’s capacity to adequately meet demand. Major competition comes from GPU heavyweights Advanced Micro Devices (AMD) and Intel, who maintain around 19% and 1% of the market share, respectively, but NVIDIA still shoulders the vast majority of global supply. Brad Lightcap, OpenAI’s Chief Operations Officer, maintained in an April 15th interview with Nikkei Asia that it is paramount that the industry avoids a shortage, as he expects “demand to be robust into the next few years.”
“Our priority,” continued Lightcap, “is making sure that we don't end up in a world where there's so much demand for AI, but we don't have enough capacity.” Short of detailing his company’s efforts towards increasing supply, he nevertheless affirmed that OpenAI is “always thinking about ways that we can be participatory in a global conversation about increasing total chip capacity.” To that end, Lightcap met with business leaders and executives from the Japanese semiconductor industry to discuss partnerships and opportunities to secure an adequate domestic supply for the company’s new Tokyo branch.
Renewed Efforts: Japan’s Semiconductor Push
Historically, Japan dominated the global semiconductor market. In 1988, NEC, Toshiba, and Hitachi – all Japanese tech companies – were the three largest chip manufacturers in the world. In fact, six of the top ten semiconductor companies in the world by chip sales were Japanese – in total, these companies captured 50 percent of all worldwide sales at the time.
Kazuto Suzuki, Professor of Science and Technology Policy at the Graduate School of Public Policy at the University of Tokyo, describes what followed this period of Japanese tech ascendancy. According to Suzuki, Japanese chip companies gradually shifted away from fabrication and towards the more lucrative development of next-generation semiconductors and semiconductor-production machinery, much like NVIDIA’s massively profitable business model. This gap in fabrication was filled with the rise of the Taiwan Semiconductor Manufacturing Company (TSMC), which produces over 90% of all high-end chips in the world. Japanese semiconductor companies have since lost their competitive advantage in the industry; Japan’s share of chip production has dwindled from 50% in the 1990s to 9% as of May 2024.
But the circumstances that led to a lack of enthusiasm for chip manufacturer subsidies have changed. A global chip shortage, which was exacerbated by onset of the pandemic, has spurred Japan’s government to dramatically ramp up support for its domestic semiconductor industry: between the financial years 2021 and 2023, the country invested 3.9 trillion yen ($24.8 billion) into the sector. A sizeable portion of that funding went towards the construction of a TSMC fabrication plant in Kumamoto, a city located on the southern Japanese island of Kyushu; production is expected to begin at the end of the year.
In a similar vein, the Japanese government has also contributed 92.9 billion yen in subsidies to Japanese computer company Kioxia and American computer drive manufacturer Western Digital, who are jointly building a plant in Yokkaichi, Mie Prefecture, to produce high-tech 3D NAND flash memory products. The same companies are also building a plant in Kitakami in the northeast Iwate Prefecture, which aims to be operational by the end of the year. Further, the Trade and Industry Ministry is also subsidizing 590 billion yen for domestic chip firm Rapidus to construct a manufacturing facility in Hokkaido in partnership with IBM, which is due to start production in 2027.
Japan’s policy of promoting domestic chip production dovetails with its support for US restrictions on Chinese access to cutting-edge semiconductors, such as NVIDIA’s A800 and H800 chips: In supporting the export bans, the Japanese government hopes temper Beijing’s development of AI and weapons systems technologies. These aspirations reflect the sentiments of Japanese media: an April 22 editorial published by Yomiuri Shimbun, one of Japan’s leading newspapers, urged the country to strengthen the foundation of its semiconductor industry, declaring that “stable procurement of the chips is essential from the perspective of economic security.”
Future Expansion
OpenAI’s expansion into Japan marks the Indo-Pacific region’s potential as a frontier in the development of the field of AI. Microsoft currently holds a 49% ownership stake in OpenAI and has robust investments in AI infrastructure, setting up its first Azure data center in Thailand in 2024. This is part of the company’s effort to expand into Southeast Asia, a region observing tremendous growth. Microsoft CEO Satya Nadella stated at a company event in Bangkok that, even in Thailand, there is potential for double-digit growth exceeding $100 billion, driven by the productivity gains enabled by AI. James Donovan, Global Lead for Strategic Partnership (Government, Science, and Security) at OpenAI, discussed during the World Bank Group and the International Monetary Fund 2024 Spring Meetings the company’s plans for Southeast Asia to emphasize AI as a point for collaboration among the Association of Southeast Asian Nations (ASEAN). ASEAN published in February 2024 the ASEAN Guide on AI Governance and Ethics which “focuses on encouraging alignment within ASEAN and fostering the interoperability of AI frameworks across jurisdictions.” A key challenge identified by analysts, however, is the region’s varying levels of digital development making it harder to implement a unified AI regulatory regime.
Despite this, OpenAI’s Tokyo office and their model’s expansion into the Japanese language paves the way for the possible agglomeration of advances in AI in the Indo-Pacific region. This provides a robust point of collaboration between American companies and Asian partners who seek to innovate and bring their partnerships closer towards the future.
Lei Nishiuwatoko is a Spring 2024 Young Professional at the East-West Center in Washington and an incoming M.S. candidate at the Edmund A. Walsh School of Foreign Service at Georgetown University. She graduated from Northeastern University where she obtained a B.A. in International Affairs. Lei has previously interned at the NATO Defense College, WorldBoston, and L.E.K. Consulting.
Vincent Zhang is a Spring 2024 Young Professional at the East-West Center in Washington. He is a senior at the Edmund A. Walsh School of Foreign Service at Georgetown University, pursuing a B.S.F.S. in International Politics with a concentration in Foreign Policy.
Kyle Ta-ay is a Young Professional at the East-West Center in Washington and a Young Leader at the Pacific Forum in Honolulu. He is pursuing an M.A. in International Affairs at American University’s School of International Service. He is also an International Student Advisor at the International Student and Scholar Services, a Research Assistant for the ASEAN Studies Initiative, and the Vice President of Finance of the Graduate Leadership Council.