Nvidia Weighs Lower Specs Amid HBM Shortage
Nvidia is facing an emergency in securing memory for its next-generation AI chip, 'Rubin Ultra,' scheduled for release early next year. The company plans to equip this chip with cutting-edge High Bandwidth Memory (HBM), but due to the supply shortage of HBM, Nvidia has resorted to altering its design.
The bottleneck of HBM supply failing to keep up with demand intensifies, prompting Nvidia to consider a reduction in HBM specifications. Initially, Nvidia planned to equip the chip with 12-layer HBM4E, but now they are discussing downgraded alternatives such as 8-layer HBM4E or 12-layer HBM4.
The limitation of DRAM production capacity is expected to continue until next year, and the wafer capacity that memory companies can allocate to HBM production is limited. Uncertainties regarding the verification schedule and mass production yield of the 12-layer HBM4E remain.