CNBC AI News, July 23 – Back in 2018, Nvidia CEO Jensen Huang reportedly visited Samsung with a collaborative proposition spanning three pivotal areas, according to sources familiar with the matter.
These areas included bolstering Samsung’s HBM (High Bandwidth Memory) production capabilities, forging a partnership on advanced process node development through Samsung Foundry, and synergistically advancing Nvidia’s CUDA software platform. Had these proposals been embraced, Samsung could have ascended to become a mission-critical partner for Nvidia in the burgeoning AI landscape.
However, Samsung reportedly declined Nvidia’s overtures. Huang apparently expressed his disappointment, allegedly stating words to the effect that Samsung lacked internal stakeholders empowered to engage in substantive long-term strategic discussions.
One potential factor influencing Samsung’s decision-making at the time could have been the ongoing legal challenges facing then-Samsung Electronics Chairman Lee Jae-yong, who was embroiled in multiple criminal investigations.
In the years since, Nvidia has emerged as a prime beneficiary of the AI boom, fueled by its cutting-edge AI chips. HBM is a critical component in these processors, with accelerators like the H200 requiring significant volumes of high-speed memory for AI inference tasks.
Following Samsung’s rejection, Nvidia pivoted and partnered with SK Hynix to co-develop HBM, ultimately securing SK Hynix as a key supplier for HBM3E memory.
This strategic alliance has proven immensely lucrative for SK Hynix, whose share price has surged over 220% in the past five years. Samsung, on the other hand, has found itself trailing SK Hynix in HBM production capacity and technological advancement.
Furthermore, Samsung’s decision has ostensibly impacted its standing in the AI chip foundry market. Currently, TSMC (Taiwan Semiconductor Manufacturing Company) dominates the production of Nvidia’s AI chips, including the latest B200.
While Samsung is now reportedly attempting to regain ground by vying for HBM4 supply contracts, its position in the AI chip manufacturing ecosystem has significantly diminished relative to TSMC.
Original article, Author: Tobias. If you wish to reprint this article, please indicate the source:https://aicnbc.com/5446.html
