Google Explores Samsung Partnership for Next-Gen AI Chip Amid Capacity Crunch

GOOGL is in talks with Samsung Electronics to manufacture a component of its next-generation AI chip, codenamed "Icefish," according to a report from The Information published June 11, 2026. The chip is Google's 10th-generation Tensor Processing Unit (TPU), designed to power AI workloads across its data centers. Under the proposed arrangement, Taiwan Semiconductor Manufacturing Company (TSM) would continue to fabricate the primary compute die using its forthcoming 1.4-nanometer process, while Samsung would produce a separate memory-interface component using its 2-nanometer process technology.

The push toward a split manufacturing strategy reflects intensifying pressure on TSMC's advanced-node capacity. Surging demand for AI chips, led by Nvidia's production volumes, has strained TSMC's most cutting-edge fabs, leaving hyperscalers like Google with less scheduling flexibility than in prior chip generations. Icefish represents a meaningful escalation of Google's custom silicon ambitions: Google is partnering with chip designer MediaTek on the design work, and if Samsung is confirmed as a manufacturing partner, Icefish would become an early real-world test of Samsung Foundry's 2-nanometer process at commercial scale.

For Google, the arrangement would reduce dependency on a single foundry while preserving access to TSMC's most advanced node for the performance-critical compute engine. Mass production is targeted as early as 2028, though timelines remain subject to change given that the chip is still in active development. The partnership signals a broader industry shift: as AI infrastructure spending accelerates, hyperscalers are diversifying their foundry relationships rather than competing purely on chip design.