SEOUL, December 26 (AJP) - Samsung Electronics plans to introduce its first fully in-house graphics processing unit (GPU) as part of a new Exynos application processor scheduled for release in 2027, marking a major step in the company’s efforts to strengthen control over on-device AI performance.
According to industry sources on Thursday, Samsung’s System LSI division is developing a proprietary GPU architecture to be integrated into the tentatively named Exynos 2800. This would shift Samsung away from its long-running reliance on partner technologies such as AMD’s graphics architecture, which the company currently uses in its Exynos 2600 chip.
The push reflects the growing centrality of GPUs in the AI era. GPUs handle parallel processing for graphics, video, gaming and neural-network operations, and serve as auxiliary accelerators to on-device NPUs. Optimizing GPU behavior for Samsung’s own software and hardware is seen as increasingly critical as devices adopt more advanced generative-AI features.
Industry officials say the company’s U.S. semiconductor research organizations — including Samsung Austin Research Center and the Advanced Computing Lab — have recruited senior GPU engineers in recent years in preparation for full in-house development. Samsung has also sought to strengthen its design capabilities to compete with global GPU leaders such as Nvidia, AMD, Intel, Apple and Qualcomm.
By internalizing GPU technology, Samsung aims to extend its AI ecosystem beyond smartphones to emerging form factors such as smart glasses, autonomous-vehicle infotainment systems and humanoid robotics. The in-house GPU is also expected to support the company’s longer-term ambitions in ASIC development, where customized chips are designed for external clients.
A successful rollout would represent a milestone for Samsung’s System LSI business, bringing the unit closer to competing as a top-tier fabless player in advanced AI silicon design.
Copyright ⓒ Aju Press All rights reserved.