JPMorgan Raises Nanya Technology Price Target to NT$710
Miles Bennett
JPMorgan upgraded Nanya Technology from neutral to overweight and lifted its target price from NT$230 to NT$710 — the thesis is not a one-quarter price hike but a re-rating from second-tier DRAM cycle stock to AI server bypass-memory supplier.
How much did the earnings forecasts move?
JPMorgan raised Nanya's FY2026–2028 EPS estimates by 48% to 118%, with revenue projected to climb from NT$66.59 billion in 2025 to NT$532.86 billion in 2028.
Adjusted EPS forecasts: NT$58.43 for 2026, NT$89.27 for 2027, NT$103.47 for 2028.
The valuation multiple is set at 8× FY2027 expected EPS — on par with Korean DRAM leaders Samsung and SK hynix. This means → JPMorgan believes Nanya should no longer trade at a second-tier discount but be priced on a first-tier DRAM framework.
Why have "legacy specs" become a scarce asset?
In Nanya's 2025 revenue mix, consumer electronics accounted for roughly 55%, low-power products about 19%, PC systems about 21%, and data-center/server applications only about 5%. This is not an HBM-driven company.
Its edge lies in retaining DDR4, LPDDR4, and DDR3 product lines. As Samsung, SK hynix, and Micron shifted resources toward HBM and DDR5, legacy-spec supply shrank. In plain terms = the old products big players no longer want to make are still in demand — fewer suppliers, higher prices.
The financials confirm this: Q3 2025 ASP rose over 40% quarter-on-quarter with shipments up over 20%; Q4 ASP climbed another 30%-plus with shipments up about 10%, producing single-quarter profit of NT$11.09 billion.
Where does the server business start?
Nanya's server revenue is still at a low base, but the company has mapped out a roadmap: expanding RDIMM — a server-grade memory module — sales, advancing SOCAMM modules, pursuing TSV-related opportunities, and developing custom high-bandwidth products.
On process nodes, 1B-node DDR5 and DDR4 are in mass production; 1C-node DDR5 pilot products entered trial production in Q3 2025; 1D-node DDR5 pilots are expected to begin trial production in Q2 2026.
This reflects a clear strategy: fund new-product qualification with legacy-product cash flow. In Q1 2026, DRAM revenue hit NT$49.07 billion, nearly all of total revenue — extreme product purity means strong price leverage, but also zero profit cushion from NAND or packaging businesses.
Can the NT$710 target hold?
JPMorgan's model rests on three assumptions that must all hold: ① DDR4 and DDR5 prices rise together in the upcycle and legacy products keep pricing power; ② legacy capacity is absorbed by SSD, networking, automotive, and other long-tail demand; ③ server revenue steadily ramps from a low base.
At the current price of NT$409.5, the stock trades at roughly 7× FY2026 EPS and 4.6× FY2027 EPS — optically cheap. But only if 2027 earnings are not a cycle peak. In plain terms = if the price surge is just another cyclical high rather than a structural shift, the low multiple is a "peak-earnings trap."
The next four quarters of server revenue share, gross-margin trajectory, and long-term contract progress will determine whether this re-rating sticks. If pricing discipline, customer contracts, or yield ramp-up falters on any front, Nanya will be repriced from a server supply-chain asset back to a cycle stock.
Content is for reference only, not financial advice.