SK hynix Launches Volume Production of HBM3E for NVIDIA’s Next-Gen AI GPU

SK hynix, a global leader in memory solutions, has announced the commencement of volume production for their latest HBM3E memory chips. This advanced AI memory product is designed for ultra-high performance and was provided to NVIDIA, a prominent customer, in late March for their up-and-coming Blackwell B200 AI GPU.

Industry-First Achievement

With the initiation of mass production, SK hynix has become the first in the industry to supply next-generation HBM3E memory, following its successful HBM3 memory used in various artificial intelligence (AI) GPUs available in the market. After a focused development period of seven months, this move propels SK hynix to being a forerunner in delivering top-tier AI capabilities globally.

Strategic Investments and Collaborations

The company's proactive steps, including a recent $1 billion investment in the future of HBM memory, illustrate its commitment to leading the industry. This also strengthens SK hynix's position as a comprehensive AI memory provider, backed by a significant partnership with NVIDIA. The volume production of HBM3E represents a key addition to its portfolio of cutting-edge AI memory products.

Busy with innovation, SK hynix also supplied NVIDIA last month with new 12-layer HBM3E memory samples, showcasing their constant push towards advancing memory technology.

Unprecedented Performance

SK hynix's new HBM3E memory stands out for its blistering speeds and efficient heat management. It can process an astonishing 1.18TB/sec of data, equating to the throughput of 230 Full HD movies, each 5GB, in just one second. Such performance metrics set new industry benchmarks and enable significant advancements in AI applications, where speed and data management are paramount.

HBM3E, SKhynix, NVIDIA