According to Micron, the new 24GB 8H HBM3E will be a part of Nvidia's H200 Tensor Core GPUs, scheduled to ship in the second quarter of 2024. The three key improvements, performance, efficiency, and scalability, were key due to the surge in AI demand and the expanded workloads.

The new Micron HBM3E will offer a performance of over 9.2Gb/s, with a memory bandwidth of more than 1.2TB/s, which makes it suitable for extensive workloads with AI accelerators, supercomputers, and data centers. In addition to higher performance, the Micron HBM3E also has 30 percent lower power consumption and offers up to 24GB of capacity, making it quite scalable.

“Micron is delivering a trifecta with this HBM3E milestone: time-to-market leadership, best-in-class industry performance, and a differentiated power efficiency profile,” said Sumit Sadana, executive vice president and chief business officer at Micron Technology. “AI workloads are heavily reliant on memory bandwidth and capacity, and Micron is very well-positioned to support the significant AI growth ahead through our industry-leading HBM3E and HBM4 roadmap, as well as our full portfolio of DRAM and NAND solutions for AI applications.”

Micron developed this industry-leading HBM3E design using its 1-beta technology, advanced through-silicon via (TSV), and other innovations that enable a differentiated packaging solution. Micron, a proven leader in memory for 2.5D/3D-stacking and advanced packaging technologies, is proud to be a partner in TSMC’s 3DFabric Alliance and to help shape the future of semiconductor and system innovations.

Micron has also announced it has started sampling 36GB 12-High HBM3E, which should push memory bandwidth well over 1.2TB/s, with more details to be unveiled at the NVIDIA GTC on March 18.