Sony Retains Chip Division, Partners with TSMC in Joint Venture to Produce Image Sensors
nashnova research
Sony CEO Hiroki Totoki ruled out spinning off the semiconductor unit, announcing instead a joint-venture framework with TSMC worth a combined ¥747 billion to mass-produce smartphone image sensors by 2029 — a clear bet on sharing the capital burden rather than going it alone.
Why won't Sony spin off its chip business?
CEO Hiroki Totoki made the call at the Paley International Council Summit in New York: the semiconductor unit stays inside Sony.
His core argument is synergy — Sony makes films and cameras, so on-set feedback flows directly to the sensor R&D team.
He also stressed that image sensors will grow critical for physical AI devices — think the "eyes" of AI-powered robots. This means → Sony sees the unit's value extending well beyond today's smartphone cameras, betting on AI hardware as a future entry point.
If it stays, who pays the bills?
The semiconductor unit is capital-intensive — its return on invested capital has long trailed Sony's entertainment divisions.
In plain terms = chip profits, measured against the money poured in, have consistently underperformed movies and gaming.
Sony's answer: bring in TSMC as a joint-venture partner to share both technology and cost. This reflects a strategy of "keep the business, split the check" — not a white-knuckle stand on principle.
How is the joint venture structured?
Per NHK and Asahi Shimbun, the two companies formally announced expanded cooperation in August 2026: Sony invests ¥465 billion (~$3 billion) and leads R&D; TSMC invests ¥282 billion, contributing advanced process technology and manufacturing.
Combined outlay: roughly ¥747 billion, with additional Japanese government subsidies expected.
The division of labor is straightforward: Sony decides *what sensor to build*; TSMC figures out *how to build it efficiently*.
Can the production timeline hold?
The partners plan to install development equipment and production lines at a factory in Kōshi, Kumamoto Prefecture, set for completion by March 2026.
Smartphone image-sensor mass production is targeted for 2029.
This means → there is roughly a three-year ramp between factory completion and volume output. Whether that window holds on schedule is the key test of whether the "share the burden" strategy actually works.
市场有风险，内容仅供研究参考，不构成投资建议。