The premise is simple but widely ignored: on a semiconductor line, a climbing defect rate provokes an instinct to buy — a new inspection system, more sensors, another engineer — before anyone has established when the drift began, where it first surfaced, and what changed upstream. The technical purchase becomes a substitute for the business diagnosis.
Globally, this reflex is expensive. Capital equipment in advanced fabs runs into tens of millions per tool, and every unnecessary acquisition compounds depreciation, floor space, and integration debt. The deeper cost is analytical: a team that reaches for hardware stops asking why yield moved. In an environment where AI vendors are selling anomaly-detection and predictive-maintenance platforms into every industrial buyer, the risk is that firms layer sophisticated tooling over an undiagnosed process and mistake instrumentation for understanding. More data does not resolve a problem that was never framed as a question.
The strategic point cuts across sectors. The teams that win are not the ones with the most sensors but the ones disciplined enough to separate symptom from cause — to treat a metric as a prompt for investigation rather than a trigger for procurement.
For Japanese manufacturers, this lands squarely on the monozukuri tradition. The genchi genbutsu instinct — go and see the actual thing — is precisely the diagnostic habit the article defends, yet it erodes when procurement cycles reward visible capex over invisible root-cause work. Japanese fabs and materials makers, sitting mid-stream in the global supply chain, cannot afford to absorb tooling they do not need.
For Japanese SIers and RPA vendors, the warning is sharper. Much of the domestic integration business is structured to sell and implement tools — a new platform, a bot, an inspection suite — because that is what can be scoped and billed. But automating or instrumenting a process that was never properly diagnosed simply hardwires the defect. The higher-value position, and the one that resists commoditization by AI agents, is diagnostic consulting: framing the business question before recommending the technical fix. SIers that reposition from tool-installers to problem-framers will command better margins; those that keep selling boxes will find their work absorbed by cheaper automation.
The takeaway for executives: audit whether your organization's response to a bad metric is a purchase order or a question. The former scales cost; the latter scales capability.
