Samsung Electronics is facing a challenging moment in its bid to supply HBM3E (High Bandwidth Memory) to NVIDIA, primarily due to difficulties in meeting the stringent standards set by the tech giant. Despite being optimistic about overcoming these hurdles, Samsung admits that providing HBM3E to NVIDIA this year appears “virtually impossible,” and eyes are now set on a potential breakthrough in 2025.
The delay stems from Samsung’s inability to meet NVIDIA’s performance requirements for the chips, as noted by Korean media reports. This gap has allowed SK hynix, Samsung’s main competitor, to take the lead in the high-performance memory sector. SK hynix, leveraging advanced methodologies such as “MR-MUF,” has managed to set a high benchmark that Samsung has struggled to meet.
This setback is significant considering Samsung’s historical dominance in the HBM and NAND markets. Competitors like SK hynix and Micron are not only securing deals with major industry players but also pioneering new technologies. SK hynix has launched a groundbreaking 321-layer+ NAND solution, further overshadowing Samsung’s current efforts.
However, the future isn’t entirely bleak for Samsung. The company is preparing to begin its HBM3E supply by the first quarter of 2025, aiming to regain footing with major corporations, including NVIDIA. Additionally, Samsung plans to introduce the next-gen HBM4 process, potentially giving it a renewed competitive advantage. Despite this, Samsung must act quickly as SK hynix has already teamed up with TSMC for HBM4 production, a move that underscores the competitive pressure Samsung faces to capture industry attention and market share.
