SK Hynix Launches 192GB SOCAMM2 Memory for Nvidia’s Next-Gen AI Chips

SK Hynix Inc (KS:000660) has announced the mass production of its latest high-performance memory module, the 192GB SOCAMM2, specifically designed to support Nvidia’s upcoming Vera Rubin artificial intelligence platform. This next-generation memory solution is engineered to enhance AI server performance while significantly reducing power consumption, addressing one of the key challenges in modern data centers.

The SOCAMM2 module is tailored for Nvidia’s Vera Rubin architecture, which is expected to succeed the company’s Blackwell processor lineup. According to SK Hynix, the new memory technology is built to overcome persistent memory bottlenecks that occur during the training and inference of large language models (LLMs). As AI workloads continue to expand in complexity and scale, efficient memory solutions like SOCAMM2 are becoming critical for maintaining speed and reliability.

As one of the world’s leading memory chip manufacturers, SK Hynix plays a crucial role in supplying advanced memory components to Nvidia. The partnership highlights the growing importance of high-bandwidth memory in powering AI infrastructure, especially as demand for AI computing accelerates globally.

Following the announcement, SK Hynix shares rose by 2.1%, contributing to a 1% increase in South Korea’s KOSPI index. Meanwhile, competitor Samsung Electronics Co Ltd (KS:005930) saw its shares decline by 1%, reflecting mixed investor sentiment in the semiconductor sector.

Both SK Hynix and Samsung have benefited from the rapid growth in AI-driven demand for memory chips. Advanced memory solutions are essential for AI servers due to their ability to handle massive data processing requirements efficiently.

Nvidia first introduced the Vera Rubin platform earlier this year, with initial shipments expected in the second half of 2026. However, industry reports suggest that potential delays could arise due to capacity constraints among key suppliers. Despite these challenges, innovations like SK Hynix’s SOCAMM2 are expected to play a pivotal role in shaping the future of AI computing.