Samsung Accelerates Exit from 8-Inch Foundry by 2029, Outsourcing Some Capacity to UMC
nashnova research
Samsung now plans to shut all 8-inch wafer lines by end-2029 — roughly a year earlier than expected — and has begun outsourcing orders to UMC, widening a looming capacity gap in mature-node foundry.
Why is Samsung pulling the timeline forward?
Samsung DS chief Jeon Young-hyun initially pushed for shutdown by end-2028 to stem losses faster.
Customers said they needed more migration time, so the target moved to end-2029 — still a year ahead of the previously expected 2030.
This means → Samsung split the difference between "fix profitability now" and "don't strand clients," but the direction is clear: the old 8-inch lines are going.
Two forces are driving the exit: rising maintenance costs + relentless pricing pressure from Chinese foundries on mature nodes.
Orders outsourced to UMC — who gains, who feels the squeeze?
Samsung has farmed out some 8-inch orders to UMC (Taiwan, 2303.TW). Volume and contract terms are undisclosed.
UMC's Q3 utilization hit 92%, with wafer shipments up 7.3% quarter-on-quarter — a record of over 1.2 million 12-inch-equivalent wafers.
In plain terms = UMC is already running near full. Samsung's outsourced orders are incremental upside, but there is a ceiling on how much more UMC can absorb.
Samsung has also sold roughly 300 8-inch tools from its shuttered Giheung S7 fab to DB HiTek. Equipment removal began in early November.
What chips actually run on these 8-inch lines?
The main products are display driver ICs (DDIs — chips that control screen output), power management ICs (PMICs — chips that regulate device power), and image sensors.
All are mature-node chips: they don't need cutting-edge process technology, but virtually every consumer electronic device depends on them.
This means → the shutdown doesn't touch "frontier compute." It hits the foundational supply chain of everyday electronics.
Why is customer migration so difficult?
Moving the same product to a different foundry typically takes six months to a year for process adjustment and qualification.
Redesigning for a different node and ramping to volume takes at least two years.
Alternative capacity is scarce: DB HiTek and SK Keyfoundry in Korea are largely booked by existing clients; migrating to 12-inch lines raises costs and requires design changes; restrictions on China-made chips close off some sourcing options.
In plain terms = customers aren't reluctant to move — there are simply very few places to move to, and most are nearly full.
What does this mean for the mature-node foundry market?
Average 8-inch utilization across the global top-ten foundries has recovered to 88%, forecast to reach 90% in the second half.
Industry outlook: strong demand for AI-related power chips + continued retirement of aging lines → 8-inch fabs are projected to run near full capacity through 2027.
This reflects a broader shift: 8-inch supply is shrinking while demand is flat to rising — the supply-demand balance is tilting.
Whether existing suppliers can absorb the capacity gap Samsung leaves behind is the key pricing variable for the mature-node foundry market.
市场有风险，内容仅供研究参考，不构成投资建议。