Samsung's HBM4 yield is said to be near 80% — a volume push, not a keynote
Seoul Economic Daily reported on September 20 that industry sources put Samsung's HBM4 yield around 80%, up from under 60% early in mass production. The same piece projects monthly HBM wafer input rising from about 180,000 this year to about 250,000 next year. Counterpoint figures cited there still show SK hynix ahead, 50% to Samsung's 33% in the second quarter.
What happened
On September 20, 2026, Seoul Economic Daily reported that Samsung’s yields have climbed to around 80%, according to industry sources. Those sources said yields sat below 60% early in mass production. The paper’s point is mechanical. stacks several DRAM dies, and a defect in one layer can scrap the finished stack. Higher yield means more sellable stacks from the same wafers and tools. This is industry sourcing, not a Samsung investor slide.
The next bottleneck in the piece is supply, not a missing keynote. Samsung is pushing HBM4 volume. Seoul Economic Daily says the company projected third-quarter HBM4 revenue at more than triple the prior quarter, and expects HBM4 to be more than 60% of HBM revenue in the second half. Three months after HBM4 mass-production shipments were announced in February, related revenue reportedly passed $1 billion. Some observers expect more than $10 billion by year-end. Those revenue lines are a company projection and observer color inside an industry-source story. They are reported, not audited results.
The capacity guide in the same article puts average monthly HBM wafer input up nearly 40%, from about 180,000 wafers this year to about 250,000 next year. On the back end, outsourced cleaning of glass carriers used in high-stack HBM is expected to rise from 20,000 to 50,000 sheets a month. The paper says cleaning volume is not a one-to-one map of HBM output. It reads the jump as a sign that high-stack work is expanding.
is still ahead. Counterpoint Research figures cited by Seoul Economic Daily put Samsung’s HBM revenue share at 33% in the second quarter, up from 21% in the first, against SK hynix at 50%, down from 58%. The gap narrowed from 37 percentage points to 17. Twelve-layer HBM4E samples were said to be in customer qualification, with reliability-stage yield said to be above 70%. Mass production still depends on that certification. The paper also describes HBM4 at about 2 million won per chip, more than double HBM3E. Treat the price and the yield percentages as industry-source direction, not a price list.
The piece links Samsung’s memory, foundry, and packaging work, including 1c DRAM and a 4-nanometre-class base die, as a reason yields could stabilize faster. That is analysis in the article, not a customer teardown.
Why it matters
HBM allocation still quietly sets AI accelerator schedules. If Samsung’s yield holds and wafer input rises, , , and custom-chip buyers get a second serious Korean source beside SK hynix. Mix can loosen. Pricing leverage can shift. Commodity DRAM can get tighter if more wafers are reserved for HBM.
For readers watching China, pair this with CXMT’s commodity DRAM ramp and with CXMT’s early HBM yields. Korea is still where advanced HBM yield is being argued in public. A Samsung volume push is a supply fact. It is not proof that memory inside China got cheaper overnight.
What to watch is whether the 80% band shows up in Samsung’s own quarterly comments, whether the 250,000-wafer guide survives into 2027 spending talk, how fast the share gap moves after that 17-point spread, and what HBM4E qualification actually produces. Keep every yield percentage labeled as an industry source until Samsung says it on the record.
Read this next to CXMT’s phone-memory platform and CXMT’s earlier HBM3E yields.