Ask any factory boss whether AI matters and you'll get a yes, since 98% of manufacturers say they're exploring it. Then ask whether the company is actually ready to use it. Only 20% can say yes with a straight face.

Most of the other 80% are stuck in the pilot loop: a proof of concept, an internal presentation, a steering committee, another proof of concept. Nothing ever reaches the floor.

Where the gap is

Cost barely registers in the survey data. The biggest barriers are softer things.

The five real barriers (% of manufacturers)
Insufficient understanding of AI — people don't know what it can actually do
47%
Lack of talent / skills — no one internally knows how to start
42%
Data infrastructure not ready — ERP exports messy, siloed, inconsistent
38%
Unclear ROI — management won't fund what it can't measure
35%
Risk & governance concerns — legal and compliance are cautious
29%
Sorted big-to-small, one accent colour on the vital bar — the TIS way.

Money isn't on that chart. A Claude or ChatGPT subscription costs less than one hour of CNC machine time. Nobody knows where to start, and the people who'd benefit most (BD managers, QA engineers, planners) aren't the ones being sent for training.

The lunch-break economy

Southeast Asia's numbers look contradictory until you spot the pattern. A McKinsey and EDB study puts 46% of SEA companies past the pilot stage, ahead of the global 35%. Yet in Singapore, 75% of individual employees use AI, while only 15% of SMEs have adopted it at company level.

Translation: the adoption is happening at lunch break. The BD manager uploads Excel to ChatGPT between meetings. The supervisor drafts an SOP on a personal phone at the mamak. Officially the company has no AI, but it runs on it every day.

Why I call this good news

Because the gap between companies is really a gap between individuals, and an individual can move this quarter without waiting for the IT department's blessing. Nothing stops you becoming the person in your office who turns a day of quote preparation into an hour.

In 2024, being in the 80% was normal. In 2026 it's a queue, and the people walking past it trained themselves.