Aug 26 — Qualcomm EVP said HBC (high‑bandwidth compute) near‑memory architecture
can deliver higher effective bandwidth and lower power than HBM, providing an
alternative route to ease the AI-era memory‑wall bottleneck. He described HBC as
a flexible stacked solution that can be paired with other compute units to
optimize system performance. Qualcomm said hyperscalers have expressed strong
interest and it is working with major DRAM manufacturers to advance the DRAM
stacking technology HBC requires.
