OpenAI could soon have a custom processor built for phones and laptops, if unconfirmed reports out of Seoul are accurate.
South Korean outlet Greened.kr said on September 4, citing industry sources, that Arm has agreed to fund development of a next-generation edge AI chip and that Samsung Electronics will handle the engineering and production. Work is said to have started in late August after Arm approved the initial non-recurring engineering outlay.
The division of labor is unusual for both companies. Arm is contributing its AI accelerator architecture and register-transfer-level designs under a license that restricts use to this one customer program, rather than a general-purpose core. Samsung’s System LSI unit is responsible for the chip design, and the company’s foundry business would manufacture it at 2nm.
That structure gives Samsung two revenue streams from a single project, design fees plus wafer sales, and offers its foundry arm a showcase order at a node where it is still chasing TSMC. Winning a marquee custom chip could pull more AI companies toward Samsung’s 2nm capacity.
For OpenAI, the reported interest fits its push to run agents and other workloads on consumer devices, where dedicated low-power silicon can cut latency and keep data local. None of the three companies has publicly commented, and the end customer could still change before the design is finished.
