SK hynix Presents ‘New Spectrum’ for AI Infrastructure at ‘AI Infra Summit 2026’

SK hynix showcased its next-generation AI memory portfolio under the slogan 'New Spectrum' at the AI Infra Summit 2026, held September 15th-17th in Santa Clara, California, where the company doubled the size of its booth from the previous year. The event drew roughly 6,000 attendees, double the prior year's figure, and SK hynix's HBF (High Bandwidth Flash) session — a next-generation NAND solution applying TSV technology like HBM — drew the greatest visitor interest, alongside PIM demos featuring the AiM chip and AiMX accelerator card and a SALT-KV session with an eSSD-equipped server.

SK hynix took part in ‘AI Infra Summit 2026,’ held over three days on September 15th-17th local time in Santa Clara, California, US. The event served as a venue to examine the latest technology trends and discuss future directions across five areas: Data & Models, Compute, Data Movement, AI Data Center, and Physical AI. With global big tech companies eager to share their innovative technologies, the exhibition floor buzzed with energy. Without continued technological innovation in infrastructure fields such as servers and data centers, AI technology would struggle to sustain its current rapid growth. Underscoring the importance of AI infrastructure, this year’s event drew roughly 6,000 attendees — double the figure from the previous year. SK hynix likewise doubled the size of its booth compared to last year and shared an extensive next-generation memory portfolio, reaffirming its leading technological competitiveness in AI infrastructure. SK hynix Leads in Next-Generation Memory, Too: Presenting ‘New Spectrum’ for AI Infrastructure Under the slogan ‘New Spectrum,’ the SK hynix booth focused on showcasing the company’s next-generation AI memory solutions and its elevated brand presence. At the rear of the booth, SK hynix’s technology vision was presented in graphic form. To its left, a dedicated space introduced technologies such as PIM and SALT-KV, while an HBF structural model was placed in front, creating a layout in which every session could be taken in at a glance upon entering the booth. The HBF session consisted of an introduction video and an HBF structural model. As the AI paradigm shifts from training to inference, there is a growing need for a new memory tier between HBM , which offers high bandwidth, and SSDs, which can store large volumes of data. HBF High Bandwidth Flash is a next-generation NAND solution that applies TSV technology like HBM, simultaneously delivering the large capacity of conventional storage and high bandwidth. It is drawing attention as a core AI memory technology poised to open the inference-centric era of Agentic AI. Accordingly, the HBF session attracted the greatest interest from visitors at this exhibition. Many of the visitors stayed at the booth to watch the introduction video through to the end, while others spent considerable time in front of the product mock-up, closely examining the structure of HBF. TSV Through Silicon Via : A technology that connects chips vertically through electrodes that pass vertically through the silicon substrate. In the PIM session, SK hynix’s AiM chip, its AiMX AiM based Accelerator card, and a server equipped with the AiMX card were displayed side by side, along with a hands-on demo that let visitors experience an enhanced LLM service firsthand. PIM Processing-In-Memory is a technology that embeds computational functions within memory to reduce data movement between the processor and memory. It is regarded as a next-generation solution that eases the bottleneck in data transfer the Memory Wall while improving both computational performance and power efficiency. Through the demo, visitors took keen interest in how much PIM technology could improve AI service efficiency, and as they viewed the server equipped with the AiMX card, they could not hide their anticipation for the fast-approaching next step in AI technology. LLM Large Language Model : An artificial intelligence model that learns from big data to generate and understand natural, human-like sentences. By dramatically increasing the scale of its parameters, it acquires advanced language-processing capabilities such as complex context comprehension, reasoning, and translation. In the SALT-KV session, a server equipped with SK hynix’s eSSD was displayed, and a monitor set up beside it allowed visitors to see for themselves — through actual data — how the SALT-KV solution works and how data storage shifts within a real AI system. SALT-KV Semantic-Aware Lifecycle Tiering for KV Cache divides an LLM’s KV cache into context-based segments and comprehensively assesses their reuse value and storage cost. Based on this, it places data in the most suitable storage tier among HBM, DRAM, and SSD, raising utilization efficiency within limited memory resources. Notably, this exhibition also featured a demo in which visitors used an LLM model with the SALT-KV solution to create and run a game, drawing a steady stream of visitors eager to experience the solution firsthand throughout the event. “SK hynix Will Deliver Solutions Optimized to the Pace of AI Advancement” — Sharing Its Technology Vision in the Session Presentation as Well Beyond the exhibition, SK hynix also conveyed confidence in its session presentation that it can fully meet the increasingly diverse needs of customers in step with the pace of AI advancement. At the Data Movement track session held on the second day of the conference, Vice President Lim Eui-cheol Head of Solution AT took the stage under the theme ‘Beyond One-Size-Fits-All: PIM, HBF and More for the New Spectrum of AI Serving,’ sharing the research and development achievements of next-generation technologies currently underway. “As AI workloads grow more diverse — ranging from ultra-low-latency agent services to long-context services where cost efficiency is critical — the conventional GPU-HBM architecture alone is increasingly unable to meet new requirements,” said Lim. “In response to these changing workloads, SK hynix is preparing new solutions that span both hardware and software: PIM, HBF, and SALT-KV.” He also emphasized, “HBF, which combines high bandwidth and large capacity to improve the efficiency of long-context workloads, and PIM technology, which improves performance and efficiency in the fast-decoding workloads of premium services, are excellent alternatives that can move beyond existing limits. SALT-KV, too, which raises processing efficiency by taking into account the characteristics and reusability of the KV cache, holds ample potential to drive the evolution of AI systems.” Through its participation in AI Infra Summit 2026, SK hynix introduced next-generation AI memory solutions such as PIM, HBF, and SALT-KV, and shared its vision to lead innovation in AI infrastructure. SK hynix will continue to strengthen its technological competitiveness so that it can stay one step ahead in responding to the diverse demands of the AI era.