Samsung Electronics’ System LSI division is developing a generative-AI
accelerator for PCs codenamed GAIA and has supplied prototype samples to major
OEMs including Lenovo and HP for performance validation, industry sources said.
GAIA is built on a 4nm process and positioned as a storage‑centric AI
accelerator that places compute close to memory; Samsung is also advancing
integration with next‑generation DRAM processor-in-memory (PIM) technology that
can execute operations on stored data. Unlike GPU-based accelerators used mainly
for training and inference, GAIA is optimized for NPU architectures and targeted
at generative AI tasks on the PC; mass production could begin as early as next
year, the sources added.
