The Audit Numbers: Products Were Listed, but Nobody Bought Them
Data that the Korea Technology and Information Promotion Agency for SMEs (TIPA) submitted to Rep. Cha Kyu-geun, as reported by SBS Biz, shows that KRW 2.74 billion in government funding went into manufacturing data "product processing" projects in 2024–2025. Those projects listed 46 products each year, 92 in total. Not one of them sold in either year, and funding still rose by more than KRW 400 million over the period. The program did not start at zero, though. In 2023, 4 of 40 listed products sold, for about KRW 50 million.
Transactions across the whole library also fell sharply. They went from 94 deals worth about KRW 348.2 million in 2024 to 30 deals worth KRW 1.1 million in 2025, a 99.7% decline. TIPA said the drop followed its decision in 2024 to stop directly subsidizing data purchases. At the Ministry of SMEs and Startups' comprehensive audit on Oct. 23, Rep. Kim Won-i noted that KRW 12.7 billion had gone into the manufacturing data library over four years, while SME-to-SME transactions totaled only KRW 700 million. The two figures should not be compared directly. KRW 2.74 billion covers only product processing over two years, while KRW 12.7 billion covers the entire library over four years.
Why Nothing Sold: Once Purchase Subsidies Ended, Demand Was Missing
The audit pointed to three causes. Companies feared leaking trade secrets, data quality varied widely from one company to the next, and the support structure rewarded listings, so the focus was on adding more products.
The buyer's side makes the problem clearer. Data generated under another plant's equipment and process conditions rarely transfers cleanly to your own plant. Without metadata, labels, and collection conditions, buyers also have no way to price it. For most SME manufacturers, the realistic goal is not selling data for revenue. It is making the data useful for their own AI adoption and for scoring well in government program evaluations.
Where Policy Is Heading: In-Kind Credit, Platform Consolidation, and a National Library
Planned in-kind credit: TIPA plans to count data that smart factory participants contribute as an in-kind share of project costs. The start date and valuation criteria have not been released, so this is still a plan.Platform consolidation: Minister Lee So-young said the platforms will be merged and upgraded. Along with KAMP's transition to "Manufacturing AI 24," channels for trading and contributing data are likely to be reorganized.MOTIR roadmap of Oct. 29: The Ministry of Trade, Industry and Resources launched the data subcommittee of the M.AX Alliance and released a manufacturing data roadmap. Its four tasks are standardization, strategic collection, secure management through data vaults, and building manufacturing foundation models and specialized manufacturing AI firms. The government's 2027 budget proposal includes KRW 115.8 billion for the national manufacturing data library. The National Assembly is still reviewing it.Ownership principle: The Ministry of SMEs and Startups' position is that manufacturing data belongs to the company that generates it. The first step is to check whether your contracts and your vendors' terms reflect this.Rep. Kim's suggestion that the government buy data from closed businesses is only a proposal at this stage.A Five-Step Check to Turn Plant Data into an Asset
Inventory: Build a table for each piece of equipment (MES, PLCs, inspection equipment, energy meters) showing what is collected, how often, where it is stored, and for how long.Rights: In your smart factory vendor and cloud contracts, check who owns the data, what formats you can export it in, and what happens to it when the contract ends.Classification: Sort data into three groups: trade secrets (recipes, process settings), data that can be shared after de-identification, and data that can be made public. Protect trade secrets with access controls and confidentiality markings so they meet the legal test for confidentiality management.Quality documentation: Record labels, anomaly events, collection conditions, and unit and tag conventions. Buyers and AI vendors ask about these first.Path to use: Use the data first to train your own AI projects. Next, share it with supply-chain partners. Consider contributing it to the library once in-kind credit takes effect.What to Sort Out Before Applying for 2027 Programs
If you plan to apply for 2027 smart factory or manufacturing AI programs, check each notice for how much weight the data acquisition and management plan carries in evaluation.Data vouchers, which fund buying and processing data, differ from manufacturing data programs in the ministry in charge, eligibility, and permitted use. Keep the two separate.Decide whether to contribute data based on the in-kind credit amount, the level of de-identification, any limits on use, and whether usage records are disclosed.KITIM supports manufacturers at each stage. We assess your current data, review the data rights clauses in vendor contracts, and write the data plan for smart factory and manufacturing AI applications. If you are working out how to organize your plant data and use it in your next application, please reach out through KITIM's consultation request page.