Samsung Foundry Sees Surge in Inquiries from BYD, Google, AMD and Other Clients
N.R. Finch
BYD, Google, AMD and Tesla are increasingly turning to Samsung for advanced foundry services, driven by TSMC's persistent capacity crunch — a shift that signals the global chip-manufacturing supply map is being redrawn under pressure.
Why are these major clients suddenly knocking on Samsung's door?
TSMC is channeling advanced-node capacity to core clients — Nvidia, Apple, AMD, Broadcom — leaving little room for new orders.
This means → smaller Chinese chip designers and even some existing customers cannot secure additional TSMC capacity slots.
One Chinese automotive-chip executive told Nikkei Asia: Samsung's yields still lag TSMC's, but "capacity availability makes it an increasingly attractive option."
In plain terms = TSMC is a five-star hotel with no vacancies; Samsung is a four-star with rooms open — and right now, availability matters more than star rating.
Who is in talks, for what, and when does it land?
BYD is negotiating with Samsung over production of its next-generation autonomous-driving chips.
Google is evaluating Samsung to fabricate its next-gen Axion processor (expected around 2028) and is also considering shifting some TPU — tensor processing unit, a chip purpose-built for AI workloads — production to Samsung as early as 2028.
AMD is in discussions to have Samsung produce some future CPUs from 2028 onward; Tesla's AI6 chip will be manufactured at Samsung's Texas fab.
This means → the 2028 window is the pivotal moment — if orders from multiple giants land in the same timeframe, Samsung's external foundry revenue mix undergoes a structural change.
What problem does the "dual-supplier strategy" actually solve?
Multiple industry executives told Nikkei Asia that geopolitical considerations are pushing some U.S. clients to proactively pursue parallel foundry relationships.
Google is described as a leading advocate of the multi-foundry approach — Nikkei previously reported that MediaTek is helping Google work with both TSMC and Intel on advanced packaging. Qualcomm has long used both TSMC and Samsung.
Chinese chip designers broadly split orders between TSMC and Samsung, with the ratio varying by product.
In plain terms = put eggs in two baskets — one optimized for performance (TSMC), the other ensuring "we can still get chips made" (Samsung).
What is the barrier to running two foundries at once?
Different foundries use different manufacturing processes; adapting designs requires additional R&D spend and supply-chain coordination costs.
This means → the dual-supplier path is typically viable only for well-capitalized chip developers — smaller firms can rarely afford to maintain design teams for two process lines simultaneously.
This reflects a deeper shift: the foundry market is moving from "use whoever has the best technology" toward "add whoever can guarantee supply."
What is Samsung Foundry's real test?
Samsung's foundry arm previously served mostly its own consumer-electronics and appliance divisions; the sustained rise in external orders marks a substantive repositioning of the business.
The core variable has not changed: whether yields can close the gap with TSMC remains the decisive factor in converting these potential orders into firm commitments.
Put simply = clients showing up at the door proves Samsung has an opportunity — but turning that opportunity into long-term contracts still comes down to how many chips come off the line working.
Content is for reference only, not financial advice.