Samsung is reportedly developing a dedicated AI accelerator for PCs as demand grows for local generative AI workloads on laptops and desktops, according to KED. The chip is reportedly being developed by Samsung’s LSI division under the codename GAIA.
Samsung has reportedly supplied prototypes to HP in the U.S. and Lenovo in China for performance validation. Mass production could begin as early as 2027, with devices potentially arriving in late 2027 or early 2028.
GAIA is not intended to function as a full system processor like an Intel Core, AMD Ryzen, or Qualcomm Snapdragon X chip. Instead, it is described as a companion processor built on a 4nm-class node and designed specifically to accelerate AI tasks on PCs.
The reported chip would focus on PC-side generative AI workloads, including on-device language models, real-time translation, image generation, and other tasks that can be offloaded from the CPU and GPU. This would put Samsung into the growing AI PC market from a different angle than companies integrating NPUs directly into general-purpose processors.
Samsung is reportedly positioning GAIA as a memory-centric AI accelerator. The architecture is designed to place compute closer to memory, potentially reducing the need to move data back and forth through separate processors.
That approach could also align with Samsung’s work on processing-in-memory technology, which enables certain computations to occur within memory itself. Since Samsung controls its own DRAM production, the company could potentially pair custom AI logic with memory technology in ways that many chip competitors cannot.
The project could mark Samsung’s return to PC silicon more than a decade after Exynos chips powered early Samsung Chromebooks beginning in 2012. Since then, Samsung’s Galaxy Book laptops have generally used processors from Intel or Qualcomm.
If GAIA gains traction, Samsung could place its own silicon back into its laptops and potentially into PCs from other manufacturers. Testing by HP and Lenovo suggests the company is pursuing broader OEM validation rather than limiting the chip to Samsung-branded devices.
The move would also place Samsung into a more competitive position within the AI PC ecosystem. Intel, AMD, Qualcomm, and Nvidia are pursuing different approaches to local AI acceleration, while Samsung would enter the market with a dedicated NPU-style companion chip.
However, several key details remain unknown. Samsung has not publicly confirmed the project, and there are no reported performance figures, power specifications, architectural details, or comparisons with NPUs from AMD, Intel, Qualcomm, or other AI PC chip providers.
The broader question is whether AI PC workloads will become heavy and common enough to justify dedicated local AI silicon beyond the NPUs already included in modern processors. If local generative AI applications become mainstream, GAIA could give Samsung a new role in the PC market; if adoption remains limited, it may be another early bet on a still-developing category.
