Danny Weber
Samsung is reportedly developing Gaia, a 4 nm Arm-based SoC for AI PCs and physical AI devices, with a focus on NPU performance and energy efficiency.
Samsung Electronics is preparing to return to the PC processor market, this time with a more ambitious bet. Insider Ice Universe, citing the South Korean outlet News1, reported that Samsung System LSI is independently developing a SoC under the code name Gaia. The chip is aimed not only at AI PCs, but also at the “physical AI” segment, including robotics and other intelligent hardware.
Gaia could become Samsung’s attempt to move beyond mobile devices and regain a foothold in the computer segment. According to the source, the processor is built on a 4 nm process and puts a strong emphasis on energy efficiency, one of the key requirements for new Windows devices with AI features. Special attention is being paid to the NPU: the neural processor reportedly has an optimized internal architecture for local AI workloads.
Samsung has already supplied Gaia prototypes to major global PC makers for performance testing. According to Ice Universe, mass production is planned for next year. If the tests go well, Samsung could offer partners an alternative to Qualcomm’s Arm chips and Apple’s solutions, especially as interest grows in computers with local AI capabilities.
This is not Samsung’s first attempt to enter the computer processor market. Back in 2012, the company used mobile chips in Google Chromebooks, but the project quickly ran into performance and software ecosystem limits, and by 2014 it had effectively faded out. The landscape is different now: Microsoft is actively pushing AI PC standards, Windows is becoming better adapted to Arm, and demand for energy-efficient devices keeps rising.
Samsung’s advantage may lie in vertical integration: the company can both design chips and manufacture them in its own fabs. That could help it control costs and make Gaia attractive to PC makers. Still, the success of this return will depend on three things: real-world performance, power consumption, and compatibility with the Windows ecosystem.
© A. Krivonosov