SK Hynix has unveiled AI-optimized memory, including HBM4 16-layer 48GB, LPDDR6, and 321-layer QLC eSSD, demonstrating its presence as a leading provider of differentiated memory solutions tailored to the needs of the AI era.

SK Hynix announced on the 6th that it will unveil a number of next-generation AI memory innovation technologies at CES 2026 and begin its full-scale attack on the global AI infrastructure market.

SK Hynix will operate an exclusive customer exhibition hall at the Venetian Expo in Las Vegas, USA from the 6th to the 9th (local time) and showcase its AI-specific memory portfolio, including HBM4, and future system solutions.

This year's exhibition theme is 'Innovative AI, Sustainable Tomorrow'. SK Hynix is focused on expanding collaboration with global customers, presenting a vision of creating a sustainable future through memory technology optimized for the AI era.

SK Hynix stated, “We will create new value across the AI ecosystem through close communication with our customers.”

The most notable product at this year's CES is the next-generation high-bandwidth memory HBM4 16-layer 48GB.

This is the successor model to the existing HBM4 12-layer 36GB, which achieved the industry's highest speed of 11.7 Gbps, and development is progressing smoothly in line with customer schedules.

SK Hynix also exhibited the HBM3E 12-layer 36GB, which will be the core of the AI semiconductor market this year, and unveiled the latest AI server GPU module of a global customer equipped with the product, showing how it is used in an actual system.

In addition to HBM, the company also introduced the SOCAMM2, a low-power memory module dedicated to AI servers, highlighting its product portfolio that can meet the rapidly increasing demand for AI servers.

SK Hynix also unveiled LPDDR6, a next-generation mobile DRAM optimized for on-device AI implementation.

Compared to existing products, it can significantly improve data processing speed and power efficiency, further enhancing the AI performance of smartphones and edge devices.

In the NAND sector, a 321-layer 2Tb QLC product for ultra-high-capacity eSSD was introduced to meet the expansion of AI data centers.

This product, which achieves the highest level of integration to date, significantly improves power efficiency and performance compared to the previous generation, securing competitiveness in AI data center environments that require low power and high performance.

SK Hynix also set up an AI system demo zone where users can see at a glance the organically connected structure of memory solutions for AI systems.

HereThe company will demonstrate future technologies such as △customer-tailored HBM structure cHBM △PIM-based generative AI accelerator card AiMX △memory direct operation technology CuD △CXL memory operation integrated solution CMM-Ax △data-aware memory CSD.

In particular, cHBM, which has a high level of customer interest, attracted attention by implementing its internal structure as a large-scale exhibit.

This visualizes a new design method that integrates some of the computational functions previously handled by GPUs and ASICs into HBM, as competition in the AI market shifts from simple performance to inference efficiency and cost optimization.

“As AI innovation accelerates, customers’ technological needs are also rapidly evolving,” said Kim Joo-sun, President and Chief Marketing Officer of SK hynix AI Infra. “We will contribute to the development of the AI ecosystem based on differentiated memory solutions and close collaboration with customers.”