Fujifilm, Canon, others said to eye stakes in Japan's Rapidus

Kyocera, Fujifilm Holdings, and other Japanese companies are considering investments in Rapidus, a chipmaker aiming to mass-produce advanced semiconductors, according to sources.

Firms said to be eyeing stakes include Honda, Fujitsu, Canon, Nagase & Co., and Argo Graphics.

Japan’s three major banks, along with the Development Bank of Japan, plan to invest up to ¥25 billion (US$160.4 million), while some regional banks are also weighing investments.

Rapidus targets mass production by fiscal 2027, with government support now totaling about ¥1.7 trillion (US$10.9 billion).

🔗 Source: The Japan Times

🧠 Food for thought

Implications, context, and why it matters.

Rapidus’s equity case looks soft next to subsidies

Kyocera joined with Fujifilm Holdings 1. Honda and Fujitsu participated. Canon and others took part as well. Up to ¥25 billion from three main banks with the Development Bank of Japan looks small next to about ¥1.7 trillion in government support.
Rapidus sought a ¥100 billion bank loan in 2024, then turned to investors due to the “distance from commercialization” 2. That shift raises doubts about bank appetite.
Eight founders, including Toyota and Sony, put in ¥73 billion in 2022 2. Repeated capital calls despite 2nm Gate-All-Around (GAA) prototyping by July 2025 3 hint at a long road to profit.

Auto and industrial electronics suppliers face long qualification lead times for Rapidus chips

Rapidus targets 2027 2nm logic mass production 4. Auto adoption will trail consumer gear due to tough qualification, so automakers and their electronics suppliers should plan for multi-year cycles before chips reach production vehicles.
AI-powered autonomous driving and software-defined vehicles need 2nm nodes 4. Suppliers should start co-development now to align with Rapidus’s Q1 2026 Process Design Kit (PDK) release 3 or risk losing ground to TSMC (Taiwan Semiconductor Manufacturing Company) or Samsung.
Industrial automation firms and robotics makers can use Rapidus’s single-wafer front-end process (a production flow that processes wafers one at a time for tighter control) 5. Pair it with AI-optimized yield management (machine learning spots defects, then boosts output) for custom edge AI chips (processors that run AI locally on devices), and engage before the PDK release to secure early allocation in tight first runs.