Japan Just Built What TSMC Feared Most

Summary

The video presents Japan’s Rapidus as an ambitious attempt to restore the country’s role in advanced semiconductor manufacturing and reduce reliance on TSMC. Its central claim is that Rapidus produced working 2-nanometer transistors at its Chitose, Hokkaido facility in July 2025, about three months after installing its first EUV lithography machine.

The narrator stresses that this is a significant prototype milestone—not proof that Rapidus can manufacture chips reliably or profitably at scale.

Technology and manufacturing approach

• What “2-nanometer” means: The node name describes a generation of manufacturing technology and transistor density, not a literal 2-nanometer feature size.
• Gate-all-around transistors: At these scales, conventional FinFET designs face problems such as current leakage. The newer gate-all-around (GAA) structure surrounds the conducting channel with the gate, improving control of the current. The video cites IBM’s 2021 prototype and its claimed performance or power-efficiency gains over earlier-generation chips.
• IBM technology and training: Rapidus was founded in 2022 with backing from eight major Japanese companies and banks. It partnered with IBM and the Belgian research center imec. Around 150 engineers trained at IBM’s Albany research center, learning the process alongside IBM researchers before applying it in Hokkaido.
• EUV and pilot production: Rapidus’s IIM-1 facility began construction in 2023, with an ASML EUV machine installed in 2024 and an experimental line operating in April 2025. The company says its first 2nm transistors passed initial electrical tests and met customer expectations.
• Individual-wafer processing and data: Rapidus is emphasizing processing wafers individually rather than relying mainly on batch processing. The company says this produces more data for AI-assisted process analysis and could accelerate learning. Its proposed service model, RUMPS, is intended to offer faster turnaround from chip design to finished product.

Japan’s industrial motivation

The video contrasts Rapidus’s effort with Japan’s past strength in semiconductors. Japanese firms once held a large share of the market but lost ground as the industry shifted toward the foundry model pioneered by TSMC. In that model, dedicated manufacturers make chips for fabless companies, while many Japanese firms remained vertically integrated.

The narrator frames Rapidus as both an industrial comeback effort and a supply-chain security project. Taiwan’s dominance of advanced chip production—described in the video as roughly 90%—creates risks in the event of a natural disaster or conflict. A successful Japanese alternative could diversify supply and challenge the assumption that advanced manufacturing must remain concentrated in Taiwan.

Funding, customers, and ecosystem

The project depends heavily on continued public and private investment. The video says government support had reached about ¥2.45 trillion by April 2026, with additional public-private financing, while Rapidus estimates it will need about ¥4 trillion to reach mass production. Japan has also changed financing rules to allow direct state equity investment.

Named prospective customers and collaborators include Tenstorrent, Preferred Networks, Sakura Internet, Esperanto, Broadcom, IBM, Fujitsu, and Canon. The video says Broadcom requested samples for evaluation, while the Japanese government is supporting development of AI processors intended for Rapidus production. It also describes a process design kit delivered to early partners in the first quarter of 2026.

In October 2026, Rapidus announced the Rapidus Core ecosystem, including 17 design-company partners intended to help customers develop chips for its manufacturing process. The narrator argues that this ecosystem matters because TSMC’s advantage is not only its factories but also its established design tools, intellectual property, and engineering support.

Competition and risks

The video emphasizes the large gap between Rapidus and TSMC:

• Rapidus’s line is described as producing about 7,000 wafers per month, compared with TSMC’s reported capacity approaching 100,000 wafers per month for its 2nm process.
• Rapidus’s initial yield target is 50%; TSMC’s reported initial yield is around 70% and increasing.
• Rapidus still needs to prove customer-certified production, commercially viable yields, firm orders, and access to much more financing.
• TSMC is expanding in the United States and Japan, including plans for advanced production in Arizona and a move toward 3nm manufacturing in Kumamoto.

Rapidus’s announced goal is to begin 2nm mass production in the second half of fiscal year 2027, with full commercial foundry services later. It is also researching 1.4nm technology, with longer-term ambitions for a second Hokkaido plant and more advanced processes. The video presents those plans as ambitious but uncertain.

Overall assessment

The narrator’s conclusion is that Rapidus is not positioned to replace TSMC for major high-volume customers in the near term. Its potential importance is instead as a proof of concept: advanced chipmaking may be reproducible outside Taiwan, and specialized chips, faster turnaround, and supply-chain diversification could give Rapidus a viable role. Its success will depend on turning a promising pilot line into dependable mass production and securing enough customers and capital.

Reviews, guides, or tutorials: None are featured; this is an industry analysis and project overview.

Main speaker and sources: A single video narrator provides the analysis. The account refers to statements from Rapidus leaders Atsuyoshi Koike and Tetsuro Higashi, company announcements and partner activity, IBM and imec’s technical work, and outside assessments attributed to EY Cosmo Securities analyst Kazuyoshi Saito and a Reuters analysis.