Google is in talks with Samsung Electronics to manufacture a component of its next-generation AI processor, according to reports citing South Korean outlet 조선일보 (Chosun Ilbo). The component in question is memory for Google's Tensor Processing Units, the custom accelerators Google designs to power its AI workloads in its data centers.
The reports describe the arrangement as a production partnership rather than a design collaboration — Samsung would manufacture the memory that slots into Google's TPU hardware. According to the MSN report citing the original Korean-language coverage, the deal could provide a meaningful boost to Samsung's chip manufacturing business at a time when the South Korean giant faces intensifying competition in the memory and foundry markets.
Details remain limited: the reports do not specify which memory standard or generation is involved, the scale of the order, or a timeline for when production might begin. Both companies have not publicly confirmed the talks.
The story fits a broader pattern of U.S. tech giants diversifying their chip supply chains. Google has long invested in custom silicon to reduce dependence on third-party AI accelerators, and pairing its own TPU designs with Samsung's manufacturing capacity would deepen that vertical integration.
If confirmed, the deal matters because it signals that Samsung — which has struggled to close the gap with TSMC in advanced chip manufacturing — could secure a foothold in the fast-growing market for AI accelerator components, while giving Google another lever to control the hardware underlying its AI infrastructure.