The Blind Spot: Japanese Engineers and Claude Code
Claude Code is eating software engineering alive. The people who built it say so openly. Boris Cherny, the creator of Claude Code, told Fortune that the title "software engineer" may start disappearing by the end of 2026. Engineers at Anthropic itself report feeling like they come to work every day to put themselves out of a job. Fortune reported that 100% of code at both Anthropic and OpenAI is now AI-written.
Meanwhile, Japanese engineers are writing cheerful blog posts about their productivity gains.
The Japanese reaction
Search Japanese tech blogs -- Zenn, Qiita, company engineering blogs -- and you will find a consistent pattern. The posts are practical and optimistic:
- "How I shipped a Flutter app in one month with Claude Code (as a non-engineer)"
- "4 ways Claude Code improved our development workflow"
- "From AI-skeptic to Claude Code convert"
- "A manager returns to hands-on engineering thanks to Claude Code"
The tone is enthusiastic. The focus is on tips, tricks, and workflow optimization. There is almost zero discussion of what this means for the profession itself.
The Western reaction
The English-language conversation is completely different. It is dark, anxious, and structural:
- Cherny predicts everyone becomes a product manager, everyone codes, and the engineer role dissolves.
- Anthropic engineers describe "conflicting feelings" -- productive now, irrelevant later.
- The phrase "permanent underclass" is circulating among tech workers.
- Non-engineers are shipping production software with no training.
The people building the weapon are warning you about the weapon. The people being aimed at are celebrating how sharp the bullet is.
The blind spot
Japanese engineers are optimizing sprint velocity with a tool that is designed to make sprint velocity irrelevant.
They see Claude Code as a productivity multiplier. They do not see it as a replacement. This is the blind spot: if a non-engineer can ship a Flutter app in a month, the value of "I can write code" drops to near zero. What remains valuable is knowing what to build, why, and for whom. That is strategy and domain expertise, not engineering.
The Japanese tech blog ecosystem reinforces this blind spot. It rewards practical how-to content. It does not reward structural analysis or uncomfortable predictions. So engineers keep writing about their workflows while the ground shifts beneath them.
What survives
Code is a commodity now. What cannot be commoditized:
- Domain knowledge -- understanding a specific industry deeply enough to know what problems matter.
- Taste -- knowing what to build and what not to build. Judgment that no prompt can replace.
- Systems thinking -- seeing the whole board, not just the next move.
- Ownership -- deciding, shipping, and being accountable for outcomes.
The engineers who thrive will be the ones who stopped calling themselves engineers a long time ago.