Nanya Technology, the Taiwanese DRAM maker long known for mature process nodes, is set to become the newest customer for ASML’s EUV lithography systems, according to Bits&Chips, a purchase that signals how the AI memory boom is dragging the industry’s second tier into the most expensive equipment the chip world makes.
Nanya has built its business on older-generation DRAM, the memory chips that go into consumer devices, industrial controllers and legacy products, and its manufacturing strategy has been deliberately conservative: buy proven equipment, serve steady customers, avoid the frontier. The move to EUV changes that posture. EUV machines, which use extreme-ultraviolet light to print the smallest features on a chip, cost hundreds of millions of dollars each and require years of process development to use productively. A company that has never run one does not buy it lightly, and the reported decision suggests Nanya sees its future in the advanced memory that AI workloads demand.
The context is a memory market transformed by artificial intelligence. High-bandwidth memory, the stacked, high-speed chips that sit beside AI accelerators, has become the industry’s most profitable product, and the three companies that dominate it, SK Hynix, Samsung and Micron, have poured record capital into capacity. The AI boom has also raised the floor for the rest of the market: even mainstream DRAM now benefits from the same demand wave, and customers are asking for more density, lower power and better performance across the board. For second-tier makers like Nanya, the choice is stark: follow the leaders into advanced equipment, or watch their product lines age out of the market.
The purchase also reshapes the math of the equipment industry. ASML’s EUV backlog has become a symbol of the semiconductor industry’s capital intensity, with lead times stretching for years and only a handful of customers able to pay. Each new EUV customer expands the machine maker’s order book, and Nanya’s entry, while small relative to the volumes that Samsung or TSMC take, is notable because it comes from the industry’s second tier. Equipment analysts read the deal as evidence that the memory recovery has broadened: when a mid-sized DRAM maker starts ordering the industry’s most advanced tools, the boom is no longer confined to the top three.
For Nanya, the decision is not without risk. EUV technology demands enormous engineering resources, and the company will need to hire, train and retain process engineers who can operate machines that most of the industry has used for less than a decade. The payoff, access to the advanced nodes that AI-era customers want, will take years to realize, and the machines themselves will consume a meaningful share of Nanya’s capital budget for the foreseeable future. The company’s conservative reputation was built on avoiding exactly this kind of bet, and the fact that it is making one now says as much about the market as about the company.
Nanya’s history makes the move all the more notable. The company was founded in the mid-1990s as part of Taiwan’s Formosa Plastics group, the industrial conglomerate that also controls memory maker Winbond, and it has spent most of its existence serving the commodity end of the DRAM market. It survived the industry’s brutal consolidation waves by staying small and staying behind the curve, a strategy that looked timid in good times and wise in bad ones. The EUV purchase does not abandon that history, but it does abandon its central premise: that Nanya could thrive while the industry’s most advanced machines belonged to other companies.
The broader pattern is familiar from earlier cycles: when the top of an industry booms, the spending eventually reaches the second tier. The memory industry has consolidated around a handful of players, and the gap between the leaders and the rest has widened with every technology generation. EUV was the line that separated them, the equipment so expensive and so complex that only the largest could cross. Nanya’s move, along with similar investments by other mid-sized memory makers, suggests that line is being crossed, and that the cost of staying in the memory business has gone up for everyone.
What it means for the memory market is more competition in the segments that matter most. Advanced DRAM, once the preserve of the top three, now has another entrant with a long runway, and customers in the server and AI markets will have one more supplier to negotiate with. ASML, for its part, gains another long-term revenue stream from a customer whose loyalty to EUV will only grow as it builds expertise around the machines. And for the industry as a whole, the message is that AI has reordered the priorities of even the most conservative chip makers: the future is advanced, and the price of admission keeps rising.
