In the latest profit announcement, Micron has announced that the company will finally launch its first HBM2 DRAM for high-bandwidth applications later this year. These devices will be mostly GPUs and the flagship network processors that have turned to HBM over the last five years for their growing bandwidth needs.
HBM2-DRAM, Micron launches new memory for GPUs and networks this year
However, despite the fact that Micron remains at the forefront of storage technologies, the company has been noticeably absent from HBM so far. Previous efforts have focused on GDDR5X and another vision of fast, stacked memory with the Hybrid Memory Cube (HMC). HMC was first announced in 2011 as a joint effort with Samsung and IBM and was a similar type of stacked DRAM for bandwidth-hungry applications, featuring a low-bandwidth bus and ultra-high data rates to deliver memory bandwidth that far exceeded the then DDR3 bandwidth.
As a competing solution from HBM, HMC saw some benefit in the market, particularly in products such as accelerators and supercomputers. However, HMC lost the battle against the more widespread HBM/HBM2 and Micron decided to focus on GDDR6 and HBM in 2018.