At the Hot Chips conference, both Samsung and SK Hynix agreed that GPUs, memory, wafer foundry, and packaging must be designed as a single system. From that point onward, their solutions began to diverge.
Samsung is transforming the base die into an active part of the system. It plans to move the memory controller from the GPU onto the logic base die, thereby freeing up 5–10% of the GPU’s area for more compute and potentially boosting performance by 10–20%. Selective AI operations can also run on the base die to reduce data movement. In the longer term, it is advancing zHBM technology, which stacks the GPU and DRAM directly on top of each other, expected to cut DRAM power consumption by about 70% while more than doubling bandwidth.
SK Hynix(), on the other hand, is focusing on higher stacking while managing the resulting heat and warpage issues. It has confirmed that HBM4 12-high stacks have entered production, while the 16-high version is in customer validation. Hybrid bonding is the path to 20-high and higher stacks, enabling thicker dies and tighter pitches. Its iHBM approach adds a dedicated thermal conduction path inside the hottest regions of the stack. The competition is no longer just about bandwidth and capacity. Base-die architecture and advanced packaging are becoming the decisive battlegrounds.
Samsung’s Custom Base Die Based on Logic Nodes
In its 2026 Hot Chips presentation, Samsung Electronics unveiled zHBM, the ultimate evolutionary form of high-bandwidth memory. The concept eliminates the 2.5D interposer and vertically stacks DRAM directly on top of compute chips such as GPUs, forming a true 3D integrated architecture. The goals are to reduce DRAM power consumption by 70% and increase bandwidth by 2.3× compared with HBM4E.
Read more
