[비즈한국] Although the government is pushing for an Artificial Intelligence Transformation (AX) policy for the manufacturing sector to strengthen industrial competitiveness, the adoption of AI technology at the small and medium-sized manufacturing sites that form the backbone of the industry remains in its infancy.
This is because the majority of small and medium-sized enterprises (SMEs) are still at the foundational stage of smart factory implementation, and the uncertainty of return on investment combined with a lack of specialized personnel hinders the industry's capacity to adopt new technology. Consequently, there is a call for a shift in the policy framework—moving away from simply increasing the number of supported companies toward a phased upgrading of on-site infrastructure and inducing tangible outcomes through inter-ministerial coordination.
3 Out of 4 Smart Factories Still at the ‘Foundational Stage’
According to the ‘1st Survey on the Status of Smart Manufacturing Innovation’ released by the Ministry of SMEs and Startups and the Smart Manufacturing Innovation Promotion Agency in April last year, the AI adoption rate among domestic small and mid-sized manufacturing enterprises was only 0.1%. Only 1.6% of companies had plans for adoption. Among the 163,273 small and mid-sized manufacturing companies possessing factories, the adoption rate of intelligent (smart) factories—a prerequisite phase before AI integration—stood at just 19.5%.
The situation does not improve much when narrowing the scope to companies that have already implemented smart factories. Among those that have adopted smart factories, only 5.2% have either implemented or planned for manufacturing AI. This signifies that most companies are stalled before the AI transformation, which is considered the next step after the smart factory. Furthermore, 75.5% of companies that have introduced smart factories remain at the foundational level, meaning there is another threshold to cross before reaching AI adoption.
This stagnation in adoption is also tied to disparities in company size and region. While the overall AI utilization rate for large corporations in Korea stands at 48.8%, it remains at 28.7% for SMEs. By sector, the utilization rate for the service industry is 53.0%, while manufacturing is relatively lower at 23.8%. By region, the gap between the metropolitan area (40.4%) and non-metropolitan areas (17.9%) is more than twofold. This shows that regional SMEs, which have a high proportion of traditional manufacturing, are struggling to actually integrate digital technology.
Unclear ROI Even After Investing 750 Million Won, Plus a Shortage of Specialized Personnel
The primary factor preventing the acceleration of AI transformation at manufacturing sites is the ‘lack of a Digital Transformation (DX) foundation’ capable of data collection and control. For an AI system to learn process data and perform tasks like quality inspection or facility maintenance, a stable equipment interface and computer network are essential, yet the smart factory adoption rate among small manufacturing firms is only 19.5%.
Even within companies that have built smart factories, qualitative advancement is slow. Among those, 75.5% are at the ‘foundational level,’ and the technologies adopted are mostly Enterprise Resource Planning (ERP) systems (76.3%) that digitize inventory and resource management. The application of real-time process control systems (MES, 14.4%), controllers (16.9%), and process automation robots or AI models remains relatively low.
Data management methods are also not systematized. A significant number of companies claiming to collect manufacturing data rely on manual entry, where workers record production volumes or defects in journals. Such fragmented data lacks continuity over time or standardized formats, making it difficult to apply directly to advanced machine learning models. Without prior investment in basic data infrastructure—such as replacing outdated interfaces on equipment and building communication networks—there are structural limitations to applying AI technology.
The burden of costs and the uncertainty of return on investment (ROI) are also cited as major barriers. According to the survey, the average cost for a small manufacturing company to build a smart factory is 750 million won (the overall average is 1.13 billion won). Considering that more than half of the funding (56.9%) comes from internal capital, the initial investment itself acts as a barrier to entry. An official from the Ministry of SMEs and Startups’ Manufacturing Innovation Division also noted that the biggest burden felt by companies is “the lack of clarity regarding what practical utility and results can be achieved compared to the massive costs of facility construction.”
A shortage of manpower is just as much of an obstacle as cost. In a survey by the Federation of Middle Market Enterprises, the biggest difficulty companies faced during AI adoption was the ‘lack of specialized personnel (41.2%).’ This response ranked even higher than lack of technology/infrastructure (20.6%) and initial investment costs (11.8%). Among the policy supports identified by mid-sized companies as necessary for AI adoption and expansion, ‘training of AI specialized personnel (21.3%)’ also ranked high.
The staffing structure at these sites, revealed by the Ministry’s survey, confirms this. Out of an average of 14.7 employees per company, only 5.4 are involved in smart factory-related tasks. Only 19.5% of companies have a dedicated department or staff, and only 6.6% have a separate budget for related training. While 14.5% of companies plan to increase personnel, nearly half (47.1%) cite cost burdens as the primary reason they cannot implement these plans.
Moving From ‘How Many Did We Supply’ to ‘How Much Has the Site Changed’
Experts point out that given the nature of the domestic industry, where the proportion of traditional manufacturing is high, the manufacturing AX policy must be supported by phased assistance that strengthens fundamental capabilities rather than focusing on short-term results. They argue for a complete overhaul of the policy structure—shifting away from quantitative management focused on the number of supplies, toward Key Performance Indicators (KPIs) that yield tangible, site-perceivable results such as defect reduction, shorter delivery times, and cost savings.
Park Jae-young, a legislative researcher at the National Assembly Research Service, stated, “While the government’s direction to expand supply projects is correct, it should no longer cling to simple figures like the number of smart factories or the rate of automated equipment adoption. The core of smart factory transformation is the improvement of the site’s fundamental health, so we must measure performance with qualitative advancement indicators that can be felt on the factory floor to ensure policy effectiveness.”