Every decade gets its two-letter transformation. DX moved records and workflows into software: the factory got dashboards, and humans kept making the calls. AX — AI transformation — is a different animal. It moves the decisions themselves into learning systems, and in manufacturing that distinction is the whole game.
The term has hardened fastest in Korea and Japan, where AX now anchors national programs and boardroom agendas, per public reporting. The direction is unambiguous: Samsung has said it intends to run AI-driven factories across its manufacturing by 2030, and the major industrial groups are reorganizing around the same target. Analyst projections put enterprise AI spending on a steep multi-year climb, with manufacturing among the largest categories. The budget line exists. The results are less evenly distributed.
The failure mode is pilot purgatory. A model that performs in the lab dies at the handoff to the line, and the autopsy usually reads the same: fragmented equipment data that no model can learn from, governance that nobody designed, and no way for the model's output to touch the process. AI that cannot adjust a parameter is a dashboard with better adjectives. That is DX wearing an AX badge.
Working AX in a factory looks unglamorous and specific. Inspection data flows into models trained on the process, the models catch drift before it becomes scrap, and — the step most programs never reach — the loop closes: parameters adjust in real time, on equipment precise enough to act on what the model knows. The operator's job shifts from running one machine to supervising a fleet. Yield stops being an audit result and becomes a control variable.
The strategic read: AX succeeds where the physical interface is precise, and nowhere else. The sequence is unforgiving — instrument the process, unify the data, then let the models act. Companies that treat AX as a boardroom initiative will get slideware. The ones that start on the line, at the point where a decision becomes a micron of motion, will get yield. That is where transformation stops being a metaphor.
Observations drawn from public company disclosures and industry reporting; spending projections reflect third-party analyst estimates.