Can Japan Compete Again in AI? A Silicon Valley VC's Honest Take
Naoki Shibata, Managing Partner at Silicon Valley VC firm NSV and author of After AI, argues Japan has a real shot at erasing its digital trade deficit through vertical AI applications.
- Japan’s AI lag versus the US is roughly 1–1.5 years — the same gap seen during the smartphone and cloud eras, and no worse.
- Only the US and China are building foundation LLMs; every other country is at least a year behind on the bottom three stack layers.
- GAFAM headcount has been flat since 2021–2022 while revenue still grows double digits — Shibata calls this the clearest proof of AI-driven productivity gains.
- The bottom three AI layers (GPU/chips, cloud infra, frontier models) are effectively closed to non-US/China players due to R&D budgets that are 100–1000x larger.
- Vertical AI (industry-specific applications) is where nearly all startup deal flow is concentrated now; the general-purpose SaaS layer is already dominated by incumbents adding AI features.
- AI’s TAM expands beyond software budgets into human labor budgets — the cost comparison shifts from SaaS subscriptions to headcount, making the addressable market far larger.
- Japan’s specific forcing function is acute labor shortage, not margin pressure — different from the US but equally strong as an adoption driver.
- Historical pattern holds: infrastructure wins first (Cisco in 2000, Nvidia now), but the application layer eventually captures the largest share of market value.
2025-11-19 · Watch on YouTube
