Samsung Tells Customers It Will Close All 8 Inch Foundry Lines by the End of 2030

Samsung Electronics has manufactured chips on 200mm wafers at its Giheung campus for more than two decades, a business that started with 180nm mass production in 2000 and expanded to 130nm in 2002, 90nm in 2006 and 70nm in 2011. The three fabs, known internally as S6, S7 and S8, became the quiet backbone of South Korea's fabless sector, supplying the analog, power and display chips that sit inside phones, televisions and cars. For most of that period the company treated the business as mature but useful, a way to keep older equipment productive while leading edge logic moved to 300mm wafers.

What the lines make is broader than the node names suggest. Giheung produces display driver ICs, power management ICs, microcontrollers and CMOS image sensors, mostly at processes between 65nm and 180nm. The customer list reads like a roll call of Korean design houses: LX Semicon, ABOV Semiconductor, Dongwoon Anatech, Silicon Mitus and Imagis.

The wider 200mm market has been tightening throughout 2026. TrendForce reported on November 3, 2026 that capacity cuts at Samsung and TSMC, combined with growing demand for AI related power chips, are expected to push average 8-inch utilisation among the world's top ten foundries to nearly 90% in 2026, up from around 80% in 2025. TSMC is reportedly moving around 80% of its annual 8-inch wafer capacity of roughly 5 million wafers to affiliate Vanguard International Semiconductor in phases. Global 8-inch capacity is expected to keep shrinking through the first half of 2027.

Samsung had signalled its retreat in stages. In January 2026 it decided to decommission only S7 by the end of that year while keeping S6 and S8 running. On the April 2026 first-quarter earnings call, Kang Seok-chae, vice president of the Foundry Business, said the company would sequentially shut 8-inch lines for power semiconductors, display driver ICs and image sensors. What customers did not have, until the notifications reported this week, was a final date for the whole portfolio.

Key Facts

The Elec reported on November 2, 2026 that Samsung has decided to exit the 8-inch wafer foundry business completely and has notified customers of the plan. According to the outlet, the company told customers it will shut every related production line by the end of 2030 and redirect workforce and equipment investment toward advanced 12-inch processes. The report described it as the first time Samsung has specified an end date, having previously said only that it would gradually scale down 8-inch production.

East Asia Brief reported on November 3, 2026 that the notice reached fabless customers on November 2 and that the phased exit closes legacy lines at Giheung, which currently runs roughly 250,000 8-inch wafer starts per month across three dedicated fabs. S7 stops on December 31, 2026, removing 50,000 wafers per month and pushing total capacity below 200,000. The remaining S6 and S8 lines will process contracted inventory until final closure on December 31, 2030.

TrendForce reported on November 3, 2026, citing The Elec, that the exit covers mature nodes from 65nm to 180nm and follows Samsung's decision to stop accepting new 8-inch orders at the start of 2026.

Alternative capacity is scarce. East Asia Brief put DB HiTek at about 150,000 wafers per month across two sites and SK keyfoundry at about 90,000 wafers per month in Cheongju, with utilisation above 88% in the prior two quarters. The briefing added that transferring validated mask sets takes 12 to 18 months and that 300mm mask sets cost about three times as much as 200mm tooling, a pair of numbers that explains why requalification is not a quick fix.

Digital Watch, published by the Diplo Foundation, reported on November 2, 2026 that TrendForce estimated the combined Samsung and TSMC cuts could drive global 8-inch wafer output down about 2.4% year on year in 2026 while demand rises, likely prompting price increases and worsening shortages. The analysis also noted that some fabless firms are securing extra volumes from Samsung and building inventory, a move that adds capital and storage costs while only delaying the search for a new source.

Analysis

The straightforward reading is margin defence. East Asia Brief reported that 8-inch operating rates have fallen to roughly 50% in downturns, and that price competition from Chinese state backed foundries has eroded profitability. Samsung has a cleaner place to put its money: 300mm capacity, sub-3nm gate all around nodes and advanced packaging at Pyeongtaek and Taylor, Texas. What this really means is that Samsung has concluded a mature contract business with cyclical utilisation and price deflation cannot earn an acceptable return, even when the same tools are in demand elsewhere in the industry.

The awkward part is the timing. The 200mm market is not collapsing, it is tightening. With utilisation heading toward 90% and TSMC consolidating its own 8-inch output at Vanguard, Samsung is walking away from a segment that is about to become scarce rather than abundant. That is a defensible portfolio decision and a difficult one for customers, because the same shortage that makes their products more valuable also makes replacement capacity harder to book. A fabless firm that loses a qualified line in a tight market is not simply renegotiating a price, it is re-entering a queue.

The bigger picture here is that 200mm has turned into a strategic chokepoint for analog, power and display silicon, precisely the parts that AI infrastructure needs in volume. Power management chips are not glamorous, but every accelerator board and every server rack depends on them, and they are built on mature processes rather than leading edge ones. Samsung's exit, paired with TSMC's transfer to Vanguard, concentrates the remaining supply in fewer hands at the moment demand is rising, which gives the surviving operators pricing power and gives buyers a planning problem.

For Korean fabless companies the constraint is geographic as well as technical. DB HiTek and SK keyfoundry are the obvious domestic substitutes, but both are reportedly already heavily allocated. Taiwan has little spare 8-inch capacity, and outsourcing to Chinese foundries can run into end customers' restrictions on Chinese made chips. Moving a design to 12-inch advanced nodes raises manufacturing cost and requires redesign and requalification, and analog and high voltage parts are the hardest to port because they depend on specialised process options rather than raw density. Digital Watch reported that South Korean industry figures have called for Samsung to delay the closure, with some arguing that government intervention may be necessary. On the evidence available, that request looks less like special pleading and more like an accurate description of how little slack exists.

Why It Matters

The first effect is price. If global 8-inch output falls about 2.4% in 2026 while demand rises, the adjustment has to happen through allocation and pricing rather than volume, and the buyers with the least leverage are small and mid sized fabless firms. Those companies do not have the volumes to reserve capacity years ahead. A 12 to 18 month mask transfer window is survivable for a company with one product family and painful for a company with ten.

The second effect is structural. Samsung's decision shifts a meaningful share of mature node manufacturing toward specialist foundries and toward China, which is exactly where Western and Korean customers are least comfortable. That tension between cost and provenance is the digital sovereignty argument in concrete form: a chip does not have to be advanced to be strategic, and a power management IC can matter as much to a supply chain as a logic die.

The third effect is internal to Samsung. Cleanroom space, electrical infrastructure and process engineers at Giheung will move to 300mm fabs, according to East Asia Brief, and capital will follow the advanced nodes. The equipment may be sold, scrapped or left idle, but the people who know how to run a high voltage analog line are the part that is hardest to rebuild if the decision is ever revisited.

Next Up

The first hard checkpoint is December 31, 2026, when S7 is due to stop. Watch whether S6 and S8 continue to accept contracted volume without interruption, and whether Korean fabless firms announce alternative sources or pre-buying arrangements before the end of the year. DB HiTek and SK keyfoundry utilisation in the next two quarters will show how much of the displaced demand can actually be absorbed at home, and any expansion announcement from either would be the clearest signal that the domestic gap is being addressed.

The longer test arrives on December 31, 2030, but the decisions that matter will be made long before then. Requalification cycles of 12 to 18 months mean customers need to commit to new processes well ahead of the final shutdown, and the phase in of TSMC's capacity transfer to Vanguard will determine how much room is left in Taiwan. For now the market has a date, a shrinking pool of qualified 200mm capacity and a customer base that has to choose between paying more, redesigning or waiting.