Is Samsung Really Building a Chip Factory in Gwangju? The Story Behind the Cheomdan 3 Site
Industry · Semiconductors
Is Samsung Really Building a Chip Factory in Gwangju?
The Cheomdan 3 site is reportedly locked in — the first new back-end facility in 35 years. Here's what's behind it, and what the hype is missing.
Why this news broke today
On the morning of June 11th, Kyunghyang Shinmun ran an exclusive: Samsung Electronics has effectively settled on the Cheomdan 3 district in Gwangju as the site for a new semiconductor plant. The story moved quickly — it was all over the regional feeds before noon, and people inside the chip industry were paying close attention too.
The basic facts are these. Samsung had been looking for a plot of around 496,000 square meters — roughly 150,000 pyeong in the Korean measure that construction people still use. Gwangju city put several options on the table: land near the Bitgreen industrial complex in Hampyeong, a former munitions depot site adjacent to Gwangju Airport, and Cheomdan 3. Cheomdan 3 won. The current plan is to secure about 50,000 pyeong first, then expand into neighboring Jangseong in South Jeolla Province to eventually build out a full semiconductor industrial zone.
Timing is not incidental here. The Jeonnam-Gwangju Integrated Special City is set to launch on July 1st. And sometime before the end of this month, there's a meeting scheduled at the presidential office between President Lee Jae-myung and the heads of Korea's major conglomerates — Samsung's Lee Jae-yong and SK Group's Chey Tae-won among them. The industry reads that as the moment when formal investment announcements will land. What we got today was the groundwork being laid before that official reveal.
The short version
- Samsung wanted: ~496,000 m² (150,000 pyeong)
- Phase one secured: ~165,300 m² at Cheomdan 3, with Jangseong expansion to follow
- Plant type: back-end semiconductor packaging, not front-end fab
- Historical context: first new back-end base since the Onyang campus opened 35 years ago
What back-end packaging actually means — and why it suddenly matters
For a long time, the chip industry treated packaging as the boring part. You do the hard, expensive work in the fab — printing microscopic circuits onto wafers — and then you hand it off to packaging engineers who wire everything up and stick it in a case. Front-end was where the prestige and the margins were. Back-end was closer to assembly work.
That logic has inverted pretty dramatically over the last few years. The clearest example is HBM — high-bandwidth memory — which is what Nvidia's AI accelerators run on. HBM is made by stacking multiple DRAM chips vertically, bonding them with tiny copper connections, and integrating the whole thing into a dense, high-speed package. That process is advanced packaging. You can make excellent individual memory chips, but if your packaging capabilities aren't up to it, you can't ship HBM. SK Hynix figured this out before Samsung did, which is a big part of why Hynix has been dominating the HBM market while Samsung has been scrambling to catch up.
The physics problem that packaging solves
There's a deeper structural reason this shift is happening. The industry has been shrinking transistors for fifty years, but that road has a physical end. You can only pack so much onto a single piece of silicon before quantum effects start working against you. So instead of just making chips smaller, the industry has increasingly turned to making them work better together — stacking them, connecting them with shorter, faster pathways, building what are essentially three-dimensional chip assemblies.
TSMC's CoWoS packaging platform has been enormously profitable off exactly this trend. The factory Samsung is planning in Gwangju would be a serious bet on the same idea — that the competitive edge in semiconductors is moving toward who can assemble the most sophisticated packages, not just who can etch the finest circuits. More on the broader industry dynamics at Electronic Times.
Why Gwangju, of all places
The obvious question. Samsung's entire semiconductor footprint — Pyeongtaek, Hwaseong, Yongin — sits in the Seoul metropolitan corridor. Decades of supplier relationships, component logistics, engineering talent pipelines: all of it is oriented around the capital region. So why go to Gwangju?
The capital region is running out of room
A semiconductor fab is one of the most resource-intensive industrial operations that exists. A single large facility can consume as much electricity as a mid-sized city. Pyeongtaek and Hwaseong are already dense with major plants; there are real constraints on how much additional power and industrial water the area can supply. The Honam region — Gwangju and the surrounding provinces — has capacity headroom, and it's widely considered the strongest part of Korea for renewable energy supply, which matters both for operating costs and for meeting the ESG commitments that Samsung's overseas customers increasingly care about.
Gwangju already has a packaging ecosystem
This is the less obvious but arguably more decisive factor. Amkor Technology, the world's second-largest outsourced semiconductor packaging company, runs a major operation out of Gwangju. Amkor's Korean business turns over roughly five trillion won a year. That means Gwangju isn't starting from scratch — there's already global-caliber back-end infrastructure, trained workforce, and institutional knowledge in the city. Layer on top of that the National AI Data Center, the Gwangju Institute of Science and Technology, and the AI Industry Convergence Complex that are already in the Cheomdan district, and Samsung gets a location with both the specific capabilities it needs and the kind of tech infrastructure it wants nearby.
The political dimension is real and worth naming
Anyone who leaves this part out is giving you an incomplete picture. Regional balanced development is a cornerstone of the current administration's platform, and Honam is a traditional stronghold. The fact that Samsung's Honam plant was reportedly a featured agenda item in government-led meetings on non-capital investment — and that the Cheomdan announcement is landing right before the integrated special city launch and the presidential business summit — is not a coincidence. That doesn't make the investment less real or less economically rational, but the political tailwind is clearly there, and it's part of why things are moving at this speed. EBN Industrial News has been tracking this angle closely.
What people are getting wrong about this deal
Local reactions in Gwangju have been enthusiastic, which is understandable. But there's a gap between what's actually being announced and what some of the coverage implies, and it's worth being clear about it.
A packaging plant is not a fab — and the difference is enormous
Amkor Technology, with operations in both Gwangju and Incheon, employs around 8,000 people in Korea. For the world's second-largest packaging company, that's not a massive number. The economic multiplier from a front-end fab — a real chip fabrication plant — is in a completely different category. When a fab comes to a region, dozens of equipment and materials suppliers follow. You get hundreds of high-level engineers needing housing, schools, and local services. The ecosystem compounds. A packaging facility is a real investment and a real employer, but its footprint on the surrounding economy is significantly smaller. Kim Yong-seok, a semiconductor professor at Gachon University, made this point bluntly: in terms of regional development, the impact of a packaging plant is inherently limited.
Logistics and talent are genuine problems
If Samsung's front-end fabs stay in Pyeongtaek and Hwaseong while the back-end moves to Gwangju, you're shipping wafers across the country between process steps. That adds cost, risk, and lead time. It's not disqualifying — global chipmakers operate distributed supply chains all the time — but it's a real operational consideration that will need to be managed carefully.
The talent question is trickier. Advanced packaging is a specialized discipline. Amkor's presence means there's some existing workforce, but Samsung's operation would need people at a different scale and technical depth. The government has signaled plans to support packaging-focused graduate programs and vocational training, and that infrastructure will need to actually materialize for this to work at the level Samsung is presumably planning for. Factory announcements are the easy part. Building a self-sustaining industrial cluster takes a decade or more. Relevant policy updates from the Ministry of Trade, Industry and Energy are available at motie.go.kr.
The honest ledger
- ✓ Existing packaging ecosystem — infrastructure already on the ground
- ✓ Power and water headroom — renewable energy supply particularly strong
- ✓ 35-year milestone — real signal of investment commitment to the region
- ✗ Not a fab — economic multiplier is much smaller than headlines suggest
- ✗ Logistics friction — geographic split from front-end operations adds complexity
- ✗ Talent gap — advanced packaging specialists are scarce; training takes time
Questions people are actually asking
When does construction start?
No official groundbreaking date has been set. The industry expects that a formal investment announcement will come at the end-of-June presidential summit with conglomerate heads. The site appears to be settled, but permitting, environmental review, and detailed negotiations still need to run their course. Realistically, don't expect shovels in the ground before 2027 at the earliest.
What exactly will this factory make?
It's a back-end packaging facility — its job is to take chips that have been manufactured elsewhere and assemble them into finished semiconductor products. The main focus is advanced packaging for AI-era applications, particularly HBM-type products where multiple memory chips are stacked and interconnected. It is not a front-end fab; Samsung will not be etching new circuits in Gwangju.
Amkor already does packaging in Gwangju. Why does Samsung need its own plant?
Amkor is an OSAT — an outsourced assembly and test company. Its business model is taking orders from multiple customers and packaging their chips. Samsung wants a captive facility for its own products. The strategic logic is vertical integration: keeping advanced HBM and next-generation packaging in-house rather than relying on a third party, which Samsung sees as critical for competing on quality and speed as the technology gets more complex. Gwangju being an established packaging hub is a reason to choose it, not a reason to skip it.
Will this affect Gwangju real estate?
Speculative interest around Cheomdan 3 is already picking up. That's predictable. The sensible framing, though, is that this is a medium-to-long-term story, not a confirmed catalyst with a fixed delivery date. There's no official investment announcement yet, construction is years out, and the population impact of a packaging facility is considerably smaller than what a full fab complex would bring. Pricing in the ceiling scenario before the news is even officially confirmed is how people get burned.
Is SK Hynix also coming to Gwangju?
SK Hynix has been looking at South Jeolla Province, but its situation is different. Hynix broke ground in April on its P&T7 advanced packaging facility in Cheongju, North Chungcheong Province — a 19 trillion won project. Adding a Honam back-end base on top of existing Yongin and Cheongju commitments is reportedly being discussed internally, but the pace is slower and nothing public has been confirmed. Samsung is clearly further along in this particular race.
