With the complexity of semiconductor manufacturing sites rapidly increasing amidst a surge in demand for AI, SK Hynix has presented the 'Autonomous FAB' as a key solution for next-generation manufacturing innovation.

Do Seung-yong, Vice President of SK Hynix (Head of DT Division), who attended the NVIDIA GTC 2026 panel session 'Building the Future of Manufacturing' held on the 17th (local time in the U.S.), stated, “In the AI era, we must simultaneously address the dual challenges of expanding production capacity and increasing manufacturing complexity.”

The explanation is that while semiconductor demand is increasing rapidly, there are structural limitations that make it difficult to expand manufacturing infrastructure at the same pace.

Vice President Do Seung-yong explained that the company is pursuing the expansion of production capacity in Korea and globally to meet the increasing demand for AI memory, and that the investment in Indiana, USA, is also part of this strategy.

However, he emphasized that since the new fab requires a significant amount of time from construction to mass production stabilization, efficiency improvements for existing production lines must be carried out in parallel.

The complexity of the manufacturing environment is also rapidly increasing.>
As the proportion of high value-added and customized products such as HBM has increased, the difficulty of fab operations has risen, and decision-making to balance quality, cost, and speed has become more difficult.

Accordingly, pointing out the limitations of existing experience and rule-based automation methods, it was stated that more fundamental manufacturing innovation is needed.

To address this, SK Hynix is pursuing the construction of an autonomous fab with a target of 2030.

The goal is for the factory to learn and make decisions on its own to drastically shorten the transition speed from design to mass production.

Operational AI, physical AI, and digital twin were presented as the core pillars of the autonomous fab.

Operational AI has implemented engineers' judgment and know-how based on data, reducing equipment maintenance and defect analysis time by more than 50%.

Physical AI is expected to improve logistics efficiency and safety and reduce parts inventory by about 30% by intelligentizing transport systems, robots, and autonomous logistics.

Digital twins enable simulation, AI training, and operational optimization without production interruption through NVIDIA Omniverse-based virtual fabs.

Vice President Do Seung-yong emphasized, “We will build a faster and more flexible next-generation manufacturing system by organically combining the three axes,” adding that “the core of manufacturing competitiveness in the AI era is autonomy and agility.”