It is an AI program, but what really decides the outcome is data. Without accumulated process data, no AI has anything to learn from, which is why data collection and validation is a separate stage.
Figures below follow the 2026 call. They change by year, so check that year's call before applying.
Eligibility and conditions of the program
Processes and metrics to improve
Scope · cost · schedule
Follow the procedure in the call
The 2026 call has a data collection and validation stage separate from the AI factory build. If you have no collected data yet, it makes sense to start by reviewing that stage.
A track that supports AI tailored to the industry and processes of a manufacturing site. It addresses process optimization and predictive maintenance with technologies such as AI agents and on-device AI.
Under the 2026 call, the AI factory build offers up to 9 months, up to KRW 200 million and a support ratio of up to 50%, and data collection and validation separately offers up to 6 months and up to KRW 50 million. The large–small win-win type also has an AI track, with government support of up to 50%.
The point where this program most often goes wrong is whether the data exists. Predictive maintenance needs failure history together with the equipment condition just before each failure; process optimization needs conditions and results tied to the same axis. Plan AI first without that data, and much of the build period is spent cleaning data.
As of the 2026 call.
| Item | 2026 call |
|---|---|
| Eligibility | Korean SMEs and mid-sized manufacturers |
| AI factory build | Up to 9 months · up to KRW 200 million · support ratio up to 50% |
| Data collection and validation | Up to 6 months · up to KRW 50 million |
| Example technologies | AI agents, on-device AI and similar |
| Example applications | Process optimization, predictive maintenance |
| Win-win AI track | Government support up to 50% of the build cost |
The overall flow is the same whatever the type.
Start with the current year's call: the funding cap, cost-share ratio, eligibility and application window. Requirements differ by type, so read the call for that specific type rather than the combined announcement.
Decide what to build and prepare the business plan with your supplier. Survey equipment communication and the signals you can collect at this stage, or the plan will not be realistic.
Applications go through the smart factory programme portal. The adopting company is the applicant; for consortium types, both the company's and the supplier's details are required.
After evaluation and selection you sign an agreement and start the build. Changes outside the agreed scope need a separate procedure, so it is better to fix the scope at the planning stage.
The build is inspected and project costs are settled. Improvements and upgrades needed in operation continue in later stages.
| Category | Details |
|---|---|
| Apply at | Smart factory programme portal |
| Operator | Smart Manufacturing Innovation Agency 044-300-0953 · 0957 |
| General enquiries | SME integrated call center 1357 |
With the information below, writing the plan goes faster.
First confirm which signals each machine can provide and whether the values match reality. How to judge collection routes is covered in the equipment communication survey.
"Adopting AI" is a means, not a goal. If you do not decide which judgement to automate and what data that judgement needs, the model you build will not be used.
The right program depends on your situation.
The basic type: companies apply directly, with no lead organization. You set the build scope yourself, so the quality of the business plan largely decides selection and results.…
Win-Win TypeA lead organization such as a large company or public institution contributes funds and the government adds part of the build cost. The company share is usually the lowest, but not everyone can apply, so check the route first…
Gyeonggi ProvinceA regional program funded by Gyeonggi Province and its cities and counties. Only manufacturers with a site in Gyeonggi are eligible; the call usually opens early in the year and selection ends in the first half.…
Smart ServiceIf smart factories cover manufacturing processes, Smart Service covers service delivery. Adopting companies and suppliers apply together as a consortium.…
The figures above are as of the 2026 calls and change every year. Always check the original call for the year in question for funding caps, cost-share ratios, schedules and eligibility.
Source: Ministry of SMEs and Startups and the Smart Manufacturing Innovation Agency, "2026 Integrated Call for Smart Manufacturing Innovation Support Programs", Bizinfo (bizinfo.go.kr). Checked 30 August 2026.
What companies most often check at the review stage.
That is exactly why the data collection and validation stage exists. Still, before applying you should survey which values each machine can provide, or the plan will not hold.
It depends on how the system is built. It can be run in-house on premises, and at manufacturing sites this is often a contractual requirement. The on-premises entry sets out the criteria.
Processes where conditions and results are linked through data, such as molding conditions and defects, or coating conditions and inspection verdicts, where cause and effect sit on the same axis.
Failure history together with the equipment condition data just before each failure. If maintenance records exist only on paper, setting up a recording system comes first. See the preventive maintenance entry.
Work back from what you want the AI to judge. With U-AI, IYULAB checks applicable areas along five axes before setting the scope.
Terms and procedures on this page follow the public standards and official sources below. Figures are as of each source's publication date.
| Topic | Manufacturing AI smart factory |
|---|---|
| Section | Smart factory grants |
| Standards applied | 1st Smart Manufacturing Innovation Survey (Ministry of SMEs and Startups and the Smart Manufacturing Innovation Agency, announced 28 April 2025) · Smart Manufacturing Innovation Agency, smart factory programme portal (smart-factory.kr) |
| Public record | GS certification grade 1 of 2 grades · 7 patents registered |
| Reviewed by | IYULAB engineering team |
| Last updated |
Public record: GS certification, grade 1 of 2 grades (U-CMMS 3.0, Telecommunications Technology Association) · 7 patents registered (10-2141840, 10-2141847, 10-2147747, 10-2354546, 10-2408497, 10-2408501, 10-2698341; searchable by number on KIPRIS) · Certifications and patentsReviewed by the IYULAB engineering team · Last updated:
Once we see your equipment and the scope you want to build, it becomes clear which program fits.
Request a consultation