Chinese DRAM maker CXMT has stated it will launch 4F2-cell DRAM within this year, according to a report by Asia Economic Daily (아시아경제). Samsung is developing the same cell architecture in parallel. Its CTO, Song Jaehyeok, confirmed the effort but gave no schedule.

The claim places CXMT on a compressed timeline for a technology node that leading memory makers have spent years qualifying. Skepticism is warranted. A stated launch year is not a volume ramp, and the report contains no capacity figures, no yield data and no named products.

What is 4F2 and why does it matter?

4F2 denotes a DRAM cell design occupying four times the minimum feature size (F), half the area of today's mainstream 6F2 cell. Shrinking the cell raises bit density per wafer and cuts cost per bit. That arithmetic explains why every major DRAM vendor pursues it.

The transition is not a simple lithography step. Moving from 6F2 to 4F2 requires a change in cell transistor structure, new capacitor alignment and redesigned wordline and bitline routing. Each element carries yield risk. Samsung and SK hynix have not announced mass production of 4F2 either, which frames CXMT's stated schedule.

What did Samsung actually commit to?

Very little, by design. CTO Song Jaehyeok told the outlet: "We're working hard." That is a confirmation of development, not a roadmap. Samsung did not provide:

• A target production year
• A pilot-line status
• Yield or reliability data
• A named customer or product family

Every promise in this story, on both sides, arrives without numbers attached. The single firm date belongs to CXMT's launch claim, and CXMT has not published supporting specifications.

How credible is CXMT's timeline?

The report offers no independent verification. CXMT has moved quickly from DDR4 into DDR5-class products, so a fast cadence is plausible. But a cell-architecture change of this scale typically moves from pilot to volume over multiple quarters, and a single-sentence launch pledge cannot be checked against fab output. A caveat belongs here: "launch" may mean pilot production, samples or paper announcement, and the report does not say which.

For context on scale: 4F2 halves the cell footprint relative to 6F2, a change comparable in structural difficulty to the move from planar to 3D NAND, which took the industry several years to yield at volume.

What is not said

Three omissions stand out. First, neither company specified a target DRAM generation, such as DDR5 or DDR6. Second, no capacitor architecture was named, a decisive variable for 4F2 feasibility. Third, the report does not state whether CXMT's claim refers to its own fabs or to partnership output. Until those gaps close, the year-end claim remains an assertion, not a specification sheet.