Kwak No-jung, CEO of SK hynix, likened the memory market in the age of artificial intelligence (AI) to a constantly changing "curved road," emphasizing the need to seek new business and technology opportunities from the perspectives of customers and partners. He outlined plans for the company to evolve into a 'full-stack AI memory' provider, offering memory solutions optimized for AI workloads, including high-bandwidth memory (HBM).
On September 9, SK hynix reported that Kwak made these remarks at the '2026 SK hynix Future Forum' held the previous day at the Supex Center in Icheon. He stated, "While the past memory market was a straight road where the competition was about who could run faster, the current market resembles a curved road that changes constantly."
He added, "It is important to not only understand our internal capabilities but also to grasp the speed and direction of changes in the external environment and adapt flexibly. We need to shift our perspective to 'outside-in' and reassess what we have taken for granted from the viewpoint of our customers and partners to find new possibilities amid the changes in the AI ecosystem."
The forum, themed 'Golden Time in the AI Era: AI Memory Solution Creators Changing the World,' focused on the changes in work methods and memory technology due to the spread of AI. Participants included industry representatives from both domestic and international AI and semiconductor sectors, as well as academic experts.
There are predictions that AI will transform the work methods in the memory industry. Tariq Shihab, head of products at Anthropic, forecasted the establishment of a 'software factory' in the semiconductor industry that utilizes AI for circuit simulation, power measurement, and system optimization. He anticipated that as AI integrates with hardware like robots, its application scope in manufacturing will expand.
SK hynix is already applying AI in its production processes. The company is systematizing the use of AI to analyze process and equipment data during mass production, and when anomalies occur on the production line, AI analyzes relevant data to suggest causes and solutions. They are preparing the 'GaiA' platform for fab production operations and are also utilizing safety monitoring robots and vision-language model (VLM) drones for high-altitude tasks.
As AI advances, the focus of the memory business is expected to shift from supplying general products to system optimization. This is due to the evolution of AI from 'reactive AI,' which simply responds to queries, to 'agentic AI,' which performs long-term tasks, resulting in an increase in the data that needs to be stored and processed.
In response, SK hynix plans to strengthen its 'full-stack AI memory' strategy by combining various memory types, including HBM, DRAM, CXL, and SSD, tailored to AI workloads. The company aims to co-design with ecosystem partners, including AI services, software, and accelerator companies, from the system level to optimize performance, power efficiency, and total cost of ownership (TCO). This strategy aims to expand their role beyond that of a simple memory supplier to a business that co-designs AI systems.
The technology expected to determine next-generation competitiveness is '3D memory.' As miniaturization approaches physical limits and power consumption increases due to data movement, the importance of vertically connecting memory and logic is growing. SK hynix has identified key technological directions, including expanding data bandwidth, ultra-short-distance transmission, and utilizing logic foundry for DRAM peripheral circuits.
Notably, the expansion of 3D technology is expected to blur the technical boundaries between front-end and back-end processes, as well as between memory and foundry. Son Ho-young, vice president of SK hynix, stated, "3D technology is changing the technical boundaries between fabs and packaging," emphasizing the importance of creating new collaborative optimization and cooperation systems that go beyond existing domains alongside innovations in advanced packaging technology.
SK hynix plans to link the discussions from this forum to the business and technology challenges of each organization. The company aims to solidify its foundation to become a 'full-stack AI memory creator' that provides memory solutions optimized for the AI era, leveraging the competitiveness gained from HBM.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.