What's this about?
"Silicon Valley dismissed Japan's System Integration (SI) culture as an unscalable consultant trap. Writing the system is no longer the scarce work. Integrating it is. In the post-AI world, everyone becomes an AI-powered Japanese SIer."
What is an SIer?
In Japan, a systems integrator — a firm that doesn't primarily build new software but assembles, customises and connects existing systems for a client, then maintains the result for years. Silicon Valley's objection was that it does not scale: revenue grows with headcount, there is no product, and the margins are consulting margins. hardmaru's inversion is that the objection was contingent on writing code being expensive. If generation is cheap, then knowing which systems to connect, in what order, and being accountable for the result, is what remains scarce.
Why is it worth noticing?
- It reframes a dismissed business model as the destination, which is the kind of inversion worth sitting with even if it turns out wrong.
- It rhymes with Anthropic's Model Hardware Standard and the week's science announcements — all integration problems, not generation problems.
- The uncomfortable implication for anyone who chose product over consulting on scalability grounds.