Kolmar Korea picked to help build South Korea's next wave of AI factories
South Korea's AI Factory Alliance just added a single cosmetics company: Kolmar Korea. The move signals a clear policy intent-push AI beyond automation and into autonomous manufacturing under the government's Manufacturing Artificial Intelligence Transformation (MAX) Strategy.
For public officials, this is a live testbed for how policy, procurement, and industry standards can compress the path from pilot to national scale.
- Government target: grow AI factories fivefold by 2030-from ~100 to 500.
- Kolmar Korea is the only cosmetics firm in the Alliance.
- Goal: 95% production accuracy via autonomous AI systems by 2029.
Why this matters for government
MOTIE is coordinating with the Ministry of SMEs and Startups and the Ministry of Science and ICT to fast-track AI-native production. Cross-ministry alignment will determine whether AI factories stay as pilots or become a core capability across sectors like auto, shipbuilding, electronics, logistics, and beauty.
This is also export policy in action. Cosmetics are a national bright spot, with US$8.52 billion shipped in the first three quarters of 2025, up 15.4%. Embedding AI into production can shorten lead times, cut defects, and keep Korean brands responsive as product cycles compress.
What Kolmar will build
Kolmar Korea will develop a fully autonomous production line that monitors itself in real time and adjusts without constant human oversight. The plan runs four years and four months, targeting 95% precision by 2029.
Modular AI models will span planning, manufacturing, quality control, filling, and packaging. The company has operated a smart factory since 2019 and reports a 42% drop in defect rates-useful baseline data for setting national metrics.
From automated to autonomous
Smart factories automate preset tasks. AI factories learn from sensor data, predict issues, and tweak variables on the fly. That shift is the heart of the MAX Strategy and Korea's Smart Manufacturing Innovation 3.0 Policy.
The government increased its 2026 budget for AI transformation programs by 84% to ₩455.2 billion to support the transition. The signal is strong: build factory intelligence, not just factory equipment. MOTIE
What policy makers can do next
- Standards first: Define shared data schemas, event logs, and quality taxonomies so AI modules are portable across vendors and sites.
- Model governance: Require version control, bias testing, fallback modes, and human-in-the-loop checkpoints for high-risk steps.
- Funding discipline: Tie grants and tax credits to measurable outcomes-defect rate, OEE, energy per unit, rework, changeover time, and lead time to launch.
- SME enablement: Build regional testbeds and reference lines where smaller firms can plug in, learn, and copy proven configurations.
- Procurement as a lever: Favor "AI-ready" equipment (open APIs, standard protocols, secure data pipelines) in public purchasing.
- Workforce: Co-fund upskilling for operators, maintenance, and quality teams-AI literacy, data interpretation, and exception handling.
- Security and safety: Mandate cyber baselines, model monitoring, and incident reporting that match the risks of autonomous control.
- Global alignment: Coordinate with export partners on certifications so AI-made products clear compliance with less friction.
Factories that think (and learn)
Kolmar's system will surface anomalies, self-correct, and keep quality steady across small-batch, multi-product runs. That directly addresses a common friction in beauty: fast-changing consumer demand and short product cycles.
The company has already expanded capacity abroad. Its second US facility includes AI-based quality checks, showing how domestic advances can translate to export markets and near-market production.
How to measure progress
- Quarterly movement toward 95% precision and corresponding defect reductions.
- Time-to-changeover and first-pass yield in small-batch runs.
- Energy use per unit and unplanned downtime trends as AI control scales.
- Supplier participation: how many SMEs integrate shared AI modules and standards.
- Export performance tied to AI-enabled production lines.
Bottom line
Kolmar Korea's selection gives government a clear proving ground: take a proven smart factory, apply autonomous control, and publish what works. If agencies lock in standards, governance, and skills now, scaling to 500 AI factories by 2030 becomes a realistic target-not a slogan.
If your agency is planning AI factory pilots and needs practical upskilling paths for evaluators and operators, review these options: AI courses by job role.
Your membership also unlocks:
