Samsung Slows High-NA EUV Rollout, Prioritizing Profitability Over Capacity Expansion
N.R. Finch
Samsung Electronics is slowing the production rollout of High-NA EUV lithography tools — the bottleneck has shifted from technical yield to the persistent losses in its foundry business. This means → even with the technology nearly ready, the financials have to work before the machines go live at scale.
What is High-NA EUV, and why does it matter?
High-NA EUV — the next-generation extreme-ultraviolet lithography tool, capable of printing finer chip circuits — is the key equipment for manufacturing at 2 nm and below. Each unit costs a fortune.
Right now, Intel is the only company that has put these tools into mass production, using them for its 18A process node (roughly equivalent to 2 nm) with plans to extend to 14A.
This means → whoever gets High-NA EUV running in volume production first gains a lead at the most advanced nodes — but "getting it running" is as much a financial question as a technical one.
Samsung bought the machines — why aren't they running?
Samsung has installed two High-NA EUV tools at its Hwaseong campus in South Korea, investing a combined ₩1 trillion (roughly $695 million).
Yet Samsung has not confirmed whether the tools are operating on a production line or producing finished chips. In plain terms = the machines are in the building, but they have not officially been switched on.
The reason is not technical: Samsung's 2 nm yield has reached roughly 55%, close to the 60% threshold widely seen as the floor for stable mass production — the technology is nearly there.
What is actually holding Samsung back?
Samsung's foundry division has been losing money since 2022. The market expects a break-even no earlier than Q4 2026.
Ramping expensive tools while still in the red would push up depreciation and operating costs, delaying the path to profitability.
This means → Samsung's problem is not "can we do it?" but "will doing it now make us lose even more?" Management has framed 2026 as the pivotal year to demonstrate both technical competitiveness and earnings power — the priority is executing on orders already booked, not accelerating capex.
Where do Intel and TSMC stand?
Intel is the most aggressive mover: it signed the industry's first High-NA EUV purchase agreement in 2023 and has since put the tools into production and shipped products to customers.
TSMC is equally cautious, planning to adopt High-NA EUV only starting at its 1.4 nm node.
This reflects an industry-wide pattern: outside Intel, major foundries are waiting for clearer economic-return signals before committing at scale.
What to watch next?
The key milestone: whether Samsung's foundry turns profitable by Q4 2026 as expected.
If the turnaround materialises, the pace of High-NA EUV capex can accelerate. If losses persist, the tools will most likely remain "bought but idle."
In plain terms = Samsung's High-NA EUV timeline is tied to its income statement, not its lab results.
Content is for reference only, not financial advice.