Next Week Marks a Major Week for AI Semiconductors: Nvidia GTC and Lisa Su's Visit to Korea

Jensen Huang Set to Unveil Next-Generation 'Feynman'
AI Semiconductor Focus Shifts Toward Inference

Chairman Chey Tae-won Makes GTC Debut, to Discuss Collaboration with Jensen Huang

AMD CEO to Visit Korea on March 18, HBM4 Supply Talks Ex

Next week, industry attention is focused on both NVIDIA's annual technology conference, GTC 2026, and the visit to Korea by AMD CEO Lisa Su. As the leaders of the world’s top two companies in the global AI accelerator market move simultaneously, competition for dominance in the global semiconductor market is expected to intensify.

Jensen Huang, CEO of NVIDIA, is speaking at the NVIDIA Annual Developer Conference (GTC 2025) held last year. Photo by Yonhap News Agency

According to the semiconductor industry on March 12, NVIDIA will host GTC 2026 in San Jose, California, USA, from March 16 to 19 (local time). The highlight of the event is Jensen Huang's keynote speech scheduled for the morning of the 16th. At this event, CEO Huang is expected to unveil NVIDIA’s next-generation AI semiconductor development roadmap.

In particular, the market is closely watching the development progress of the next-generation platform "Vera Rubin," which will be equipped with the 6th generation high-bandwidth memory (HBM), "HBM4," from Samsung Electronics and SK hynix, as well as whether a physical prototype will be revealed. There is also speculation that the detailed specifications of the "Vera Rubin Ultra," expected to launch next year, may be disclosed. Another major focus is whether detailed information about the next-generation graphics processing unit (GPU) "Feynman," first mentioned at last year’s GTC, will be made public. Feynman, the successor to Vera Rubin, is known as a next-generation chip that improves training and inference efficiency for large-scale AI models.

Additionally, there is speculation about the possible announcement of an "inference-only chip" designed to boost AI model response speeds. This is interpreted as a sign that the center of the AI semiconductor market is shifting from a training-centric structure to inference infrastructure for implementing "Agentic AI."

The industry is also paying close attention to the potential meeting between Chey Tae-won, Chairman of SK Group, and CEO Huang. Chairman Chey is expected to attend GTC 2026 for the first time and discuss ways to expand the supply of SK hynix's HBM for NVIDIA's next-generation platform, among other collaboration opportunities. This meeting comes about a month after their last encounter on February 5 in Santa Clara, California.

Lisa Su, CEO of AMD, is delivering a keynote speech at The Venetian Hotel in Las Vegas, USA, on January 5, a day before the opening of CES. Photo by Yonhap News

Lisa Su, CEO of AMD—NVIDIA’s main competitor—will also be visiting Korea during the same week. On March 18, CEO Su is scheduled to meet with executives from Samsung Electronics and Naver.

There is particular speculation about a possible meeting with Lee Jae-yong, Chairman of Samsung Electronics. At this meeting, the expansion of HBM supply—a key component for AI accelerators—is expected to be a central topic. Last month, Samsung Electronics became the first in the world to mass-produce and ship the 6th generation HBM, HBM4. If AMD adopts this for its next-generation AI accelerators, Samsung Electronics is likely to emerge as a leading supplier.

A semiconductor industry official commented, "Currently, the demand for AI accelerators far exceeds supply—there is an absolute supply shortage," adding, "Given the urgent need for supply chain diversification, it is both natural and strategic for AMD, which has strong technological capabilities, to partner with Samsung Electronics given current market conditions."

This content was produced with the assistance of AI translation services.