SK hynix participated in the Consumer Electronics Show 2024 (CES 2024), the world's largest IT and consumer electronics exhibition, held in Las Vegas, U.S.A., from January 9th to 12th (local time), to showcase its ultra-high-performance memory technology, which is the core of future artificial intelligence (AI) infrastructure. At the show, seven SK Group companies, including SK hynix, decorated the amusement park-themed SK Wonderland pavilion and unveiled AI technologies at the SK ICT Family Demo Room, which was set up separately at the event.
Through its participation in CES, SK hynix emphasized that memory semiconductors are at the center of the upcoming AI era and expressed its ambition to showcase its competitiveness in the global AI memory market. In addition to HBM3E, an ultra-high-performance memory for AI, SK hynix introduced memory products based on Compute eXpress Link (CXL), a next-generation interface that integrates different interfaces such as CPU, memory, and storage to increase processing efficiency and capacity scalability; a prototype of Computational Memory Solution (CMS), a memory solution that integrates CXL-based computational functions; and AiMX, a prototype card for AI accelerators based on Processing-In-Memory (PIM), a next-generation intelligent memory with processor-like computational functions.
CES, where innovative ICT products are unveiled every year, is also known as an exhibition full of things to see and do. This year, a variety of products and ideas attracted visitors, and SK hynix also provided various experience opportunities for visitors through SK Wonderland.
SK Group companies, including SK hynix, organized SK Wonderland as a place where the dream of Net Zero becomes a reality, centered on the message of "where all the happiness in the world is gathered. SK Wonderland, which encompasses carbon reduction technologies, offered various programs to experience the everyday values and happiness that can be felt in a world where the climate crisis has disappeared. To this end, various media arts and attractions were implemented, including a sphere LED that plays 360-degree videos, a 15-meter-long media tunnel, and a future urban air traffic (UAM) video simulator.
In particular, SK hynix's AI Fortune Teller was a hit with visitors. Fortune Teller is a generative AI that creates tarot cards synthesized from visitors' faces through face recognition AI technology. In addition, the features and importance of HBM3E, an ultra-high-performance memory that is the core of AI infrastructure, were introduced in detail.
Introduced in the Fortune Teller, HBM3E is the highest-performance memory semiconductor in existence, developed by SK hynix in August 2023, with an operating speed of more than 1.18TB (terabytes) per second. It is being utilized as a key element in AI, cloud, and high-performance computing fields that require fast processing of massive amounts of data, and the company expects to begin full-scale mass production in the first half of this year to solidify its leadership in AI memory technology.
In addition, CIS, which converts images of objects into digital signals, is mainly used in smartphones and cameras, and has recently been actively utilized in AI-based autonomous driving, medicine, security, and other areas where image data is required.
SK hynix showcased its AI memory technology at the SK ICT Family Demo Room, which was set up separately at the CES 2024 venue. First, CXL, which has recently gained attention as an AI memory technology alongside HBM, received a lot of attention.
SK hynix believes that the CXL technology showcased at CES 2024 will be of great help to AI customers who want to process massive amounts of data more efficiently. SK hynix plans to commercialize DDR5-based 96GB (gigabyte) and 128GB CXL memory solution products in the second half of this year.
The company also showed a prototype of CMS, a CXL-based memory solution. CMS is a solution that integrates high-capacity memory with machine learning and data filtering computational functions, and is expected to be utilized in various data-intensive areas such as AI and big data analysis.
In particular, AiMX (AiM based Accelerator) stood out at the exhibition. AiMX is a card-type product used for generative AI accelerators specialized in Large Language Model (LLM), and is equipped with GDDR6-AiM (Accelerator-in-Memory) chip, a next-generation intelligent memory that can not only store data but also compute. Compared to existing GPU-centric generative AI systems, AiMX has the advantage of significantly reducing data processing time while consuming less power, and is expected to play an important role in the development of AI systems that need to process a lot of data.
