News report
Samsung and ASML Expand High-NA EUV Work for Future DRAM and 12-Inch Photomasks
Samsung plans High-NA EUV for future DRAM high-volume manufacturing by 2028 and has joined the industry push toward 12-inch photomasks. Here is what is confirmed and what remains a roadmap.
On this page
- What Samsung and ASML announced on September 8, 2026
- High-NA EUV raises the numerical aperture from 0.33 to 0.55
- Why the current High-NA optical design creates a mask-field tradeoff
- The 12-inch photomask effort is an industry transition, not a finished production standard
- Samsung’s 2028 DRAM target arrives earlier than the 12-inch-mask production roadmap
- High-NA is already beyond the laboratory, but logic use does not prove DRAM readiness
- What this could change for memory manufacturing — and what it does not tell us yet
What Samsung and ASML announced on September 8, 2026
Samsung Electronics and ASML announced an expanded strategic collaboration on September 8, 2026 focused on High Numerical Aperture extreme ultraviolet lithography and supporting manufacturing technology. The two concrete items are separate but related: Samsung says it plans to introduce ASML High-NA EUV into future DRAM high-volume manufacturing by 2028, and Samsung is joining the industry effort to develop a larger 12-inch photomask platform for High-NA EUV.
Those are roadmap commitments, not evidence that Samsung is already mass-producing DRAM with High-NA EUV. Samsung did not disclose a production yield, a measured cost reduction, a specific future DRAM process node, or a retail-memory launch tied to the announcement. The useful way to read the news is therefore as a manufacturing-direction signal with explicit target dates, not as proof that the promised production economics have already been achieved.
High-NA EUV raises the numerical aperture from 0.33 to 0.55
ASML’s current EXE platform is its High-NA generation of EUV lithography. The company increases projection-system numerical aperture from 0.33 on its NXE EUV platform to 0.55 on EXE. In lithography, higher numerical aperture allows a smaller printable feature for the same exposure wavelength, which is why High-NA is intended to extend geometric scaling beyond the practical reach of today’s 0.33-NA EUV systems.
ASML states that its EXE:5000 can reach 8 nm resolution and can print, in a single exposure, features 1.7 times smaller than its NXE systems. ASML also says that can correspond to 2.9 times higher transistor density and can reduce the need for multiple patterning. Those figures are manufacturer technical claims about the tool platform; they are not independent Core Tech Tips measurements and they do not directly translate into an equivalent DRAM-density or cost gain for a future Samsung product.
Why the current High-NA optical design creates a mask-field tradeoff
High-NA EUV uses larger anamorphic projection optics. ASML explains that this lets the EXE platform retain traditionally sized reticles, but the anamorphic design exposes a field that is half the size of the field used by the 0.33-NA NXE platform. For designs that exceed that half-field, manufacturers can need stitching: exposing adjacent fields that together form the intended larger pattern.
That is the connection between High-NA EUV and the new photomask initiative. The larger-mask effort is not being presented as a prerequisite for every early High-NA layer. ASML and its partners are instead describing a transition in which High-NA can begin with current 6-inch masks while the ecosystem develops a larger mask format intended to remove some of the field-size and stitching constraints later.
The 12-inch photomask effort is an industry transition, not a finished production standard
Samsung says the semiconductor industry has relied on the 6-inch photomask format for decades and that it will join the industry initiative to advance 12-inch photomasks. In a parallel September 8 announcement, ASML and TSMC said their collaborative initiative targets a 12-inch mask pilot line by 2031 and full lithography-system readiness for advanced-node production by 2033. That timeline makes clear that the larger-mask ecosystem is still development work rather than an immediately deployable manufacturing standard.
Samsung and the initiative describe larger masks as a way to increase scanner productivity, lower chipmaking costs and remove stitching constraints. Those are expected benefits stated by the participating companies. They should not be read as measured savings already demonstrated in Samsung DRAM production, because the supporting mask infrastructure, standards and production equipment are still being developed toward later milestones.
Samsung’s 2028 DRAM target arrives earlier than the 12-inch-mask production roadmap
The dates matter. Samsung’s stated target is to bring High-NA EUV into future DRAM high-volume manufacturing by 2028, while the ASML-TSMC large-mask initiative targets a pilot line in 2031 and full advanced-node lithography-system readiness in 2033. That means Samsung’s 2028 DRAM plan cannot simply be treated as synonymous with a mature 12-inch-mask production flow.
A reasonable interpretation is that early High-NA DRAM adoption can proceed with the existing mask ecosystem and the current EXE optical approach, while larger photomasks are a later attempt to improve how the technology handles field size and productivity. That sequencing is consistent with ASML’s description of High-NA adoption progressing first with current 6-inch masks and later being further supported by 12-inch masks.
High-NA is already beyond the laboratory, but logic use does not prove DRAM readiness
There is useful evidence that High-NA EUV itself has moved beyond a pure research-tool phase. Intel Foundry and ASML said on September 7, 2026 that Intel had processed more than one million wafers across tool certification, research and development, and volume production, with High-NA used on select layers for a subset of Intel Core Ultra Series 3 processors. That demonstrates real production exposure for the technology in logic manufacturing.
It does not prove that Samsung can achieve its 2028 DRAM target on schedule. DRAM has different layouts, process integration, yield constraints and cost targets from Intel logic products. The Intel result is therefore evidence that the scanner platform has entered real manufacturing use, while Samsung’s future DRAM integration remains a separate roadmap that still has to be executed.
What this could change for memory manufacturing — and what it does not tell us yet
For DRAM, the strategic value is straightforward: finer patterning can help extend geometric scaling as conventional patterning becomes more complex. If High-NA can replace some multi-patterning steps with fewer exposures, manufacturers may reduce process complexity and cycle time. Larger masks are aimed at another part of the same problem by improving usable exposure-field economics as the industry pushes High-NA into broader production.
The announcement does not justify a forecast for consumer RAM prices, HBM availability, future DIMM capacities, or a specific Samsung memory-product launch. Those outcomes depend on far more than lithography, including process yield, die design, packaging, fab capacity, product mix and market demand. For now, the confirmed development is narrower but still important: Samsung has put a 2028 date on planned High-NA DRAM high-volume manufacturing and joined the industry effort to build the larger-mask ecosystem intended to support High-NA at scale later in the decade and beyond.
Sources
Primary and technical sources
These sources support the reporting and analysis above. Current stories are updated when later evidence materially changes the facts.
01 Samsung Semiconductor Global Newsroom
Samsung Electronics and ASML expand strategic collaboration for next-generation semiconductor manufacturing02 TSMC
ASML and TSMC announce initiative for industry transition to large-format photomasks for High-NA EUV03 ASML
TWINSCAN EXE:5000 High-NA EUV lithography system04 Intel
Intel Foundry and ASML accelerate industry readiness for High-NA EUV
Related
Continue from here
Useful next steps selected from the same technical reference and publication system.
Technical guide
Windows Page File Explained: Virtual Memory and Commit Limit
Understand what the Windows page file does, how it extends the system commit limit, how paging differs from RAM use, and why crash dumps can depend on it.
Tool
DDR Memory Latency Calculator
Convert DDR data rate and CAS latency cycles into CAS timing in nanoseconds.
Compatibility & upgrades
Laptop RAM Upgrade Compatibility: SODIMM, Soldered Memory, DDR4/DDR5, Capacity, and Mixed Modules
Check whether laptop RAM is actually upgradeable, then verify SODIMM versus soldered memory, DDR generation, slots, capacity, speed, and exact-model limits before buying.
Tool
DDR Memory Bandwidth Calculator
Calculate theoretical peak DDR memory bandwidth from transfer rate, bus width per channel, and active channel count.