Nanya Technology
Integrated device manufacturer (IDM): designs and fabs its own DRAM at a New Taipei (Taishan/Fab-3A) 300mm site; sells standard DDR/LPDDR, MCP/eMCP, KGD and Elixir-brand modules to OEMs, module makers and distributors. Commodity, ASP-cycle-driven; single owned fab with a new fab under construction (equipment move-in targeted Q1 2027).
Earnings, margins, COGS & capex
Nanya is a highly cyclical, single-product commodity DRAM house whose earnings are dominated by ASP, not volume. After a brutal 2023-24 trough (losses), the 2025-26 legacy-DRAM shortage -- triggered by Samsung/SK Hynix/Micron cutting mature-node output to prioritize HBM/DDR5 for AI -- sent DDR4/DDR3/LPDDR4 prices vertical. Q1 FY2026 ASP rose >70% QoQ on a mid-single-digit QoQ bit-shipment DECLINE, driving gross margin to ~68% and a record NT$26.1B net profit. The paradox management flagged: legacy DRAM margins currently exceed HBM, and DDR4/LPDDR4 tightness is even slowing Nanya's own shift to DDR5.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~32¢ is cost of goods and ~7¢ operating expense, leaving ~61¢ of operating profit (~53¢ net).
Revenue trend
Margins
up from 49.0% QoQ; negative through much of 2023-24
up from 39.1% QoQ
from a net loss in Q1 FY2025
vs NT$3.58 in Q4 FY2025 and NT$2.13 for full-year FY2025 (net income NT$6.60B); vs a Q1'25 loss
COGS structure
Wafer manufacturing at a single owned New Taipei 300mm fab -- silicon wafers, high electricity/utilities, photomasks, and heavy depreciation on prior fab/equipment investment. As a mature-node maker Nanya is not gated on leading-edge EUV; cost structure is fixed-cost-heavy, so operating leverage is enormous on the way up (68% GM at peak) and brutal on the way down (losses at trough). Node roadmap: 1B nm (3rd-gen 10nm-class) in production, qualifying 16Gb DDR5/LPDDR4; 1C/1D nodes and custom-AI projects in development.
Capex
2025 capex NT$13.4B actual (below a NT$19.6B cap; portion deferred to 2026). 2026 guide ~NT$52B -- ~3-4x prior year -- split ~70% new-fab construction / ~30% wafer equipment. New fab (New Taipei) equipment move-in targeted Q1 2027; management flags a potential large capacity step-up within 2-3 years of the ramp.
Latest earnings
Blowout vs the prior quarter and vs a year-ago loss -- record revenue and EPS; management guided Q2'26 to improve further with high margins characterized as sustainable over the next few quarters.
2026 capex ~NT$52B; FY2026 bit-shipment growth target ~15% YoY; new-fab equipment move-in Q1 2027; Q2'26 expected up on Q1 with sustained high gross margin. Honest tension: DDR4/LPDDR4 scarcity is capping the DDR5 ramp (DDR5 ~10% of revenue).
- Revenue
- NT$49.09B (+582.9% YoY, +63.1% QoQ)
- Gross margin
- 67.9%
- Net income
- NT$26.06B (net margin 53.1%)
- EPS
- NT$8.41
- ASP QoQ
- >+70%
- Bit shipment QoQ
- -mid-single-digit %
- Net cash
- ~NT$68.2B (ex-placement)
Growth drivers
- Legacy-DRAM shortage supercycle — the big-3 (Samsung, SK Hynix, Micron) reallocating capacity to HBM/DDR5 for AI, starving DDR4/DDR3/LPDDR4 supply and handing pure mature-node makers historic pricing power.
- ASP, not bits — Q1'26 growth was ~all price (ASP +70% QoQ) on flat-to-down shipments; the cycle is a pricing event.
- NT$78.72B (~US$2.5B) private placement completed 8 Apr 2026 (~10.19% of enlarged share count) subscribed by SanDisk, Solidigm (SK Hynix), Kioxia and Cisco, plus 3-year supply agreements -- customers pre-buying allocation and de-risking Nanya's fab expansion.
- New fab (equipment move-in Q1 2027) adding a material volume lever for the next cycle.
- DDR5/LPDDR5 mix-up and 1B/1C/1D node migration lowering cost-per-bit and opening server/AI-adjacent sockets.
Bull & bear
A largely-depreciated pure-play riding a genuine structural shortage: the big-3 have vacated mature nodes for AI/HBM, so DDR4/DDR3/LPDDR4 -- the bulk of Nanya's book -- are in a supply-starved melt-up where Nanya prints ~68% gross margins, ~NT$68B net cash, and is adding capacity into its own scarcity, all backed by Formosa and a NT$78.72B customer-equity placement.
- Operating leverage is extreme -- a largely-depreciated fab means peak-cycle margins (68% GM, 53% net) that few semis touch; Q1'26 net profit alone (NT$26B) nearly 4x all of FY2025 (NT$6.6B).
- The shortage is supply-structural, not a demand spike: Samsung/SK Hynix/Micron reallocating fabs to HBM leaves a durable mature-node gap Nanya is positioned to fill.
- Balance sheet is net cash ~NT$68B plus a NT$78.72B (~US$2.5B) strategic placement -- Nanya funds the new fab from strength, and SanDisk/Solidigm/Kioxia/Cisco 3-year deals pre-sell the volume.
- Management says legacy-DRAM margins currently BEAT HBM -- Nanya captures AI-cycle pricing without the EUV capex war.
- Optionality: new fab (Q1'27 move-in) adds a fresh volume lever, and DDR5 mix-up is still ahead -- volume layered on top of price.
This is commodity memory at the top of a violent price cycle -- single product, single fab, ~1.6% share, no HBM, and growth that is 100% ASP on FALLING bit shipments. History says these margins mean-revert hard (Nanya lost money in 2023-24), and Nanya is committing ~NT$52B of peak-cycle capex plus a new fab straight into the setup for the next glut, with CXMT and the big-3 able to flood supply.
- Growth is a pricing event, not a franchise: ASP +70% QoQ on -mid-single-digit bits -- when supply normalizes, the same leverage runs in reverse to losses.
- Structurally sub-scale: ~1.6% of a market where Samsung/SK Hynix/Micron hold ~90% and set the cycle; Nanya is a price-taker, not a price-maker.
- No HBM, ~10% DDR5 -- shut out of the secular AI-memory growth tier; the franchise is legacy nodes a stronger rival can re-enter at will.
- CXMT (China) is scaling mature-node DRAM specifically in Nanya's lane (~8% share, global #4) -- a low-cost structural share threat that persists after this cycle.
- Capacity-at-the-top: ~NT$52B 2026 capex and a new fab ramping by ~2027-28 is textbook memory over-build risk if demand cools or the big-3 return to DDR4/DDR5.
- Cyclically-adjusted, trough earnings are near zero-to-negative; valuing it on peak EPS is the classic memory value trap.
What it is worth
Peer-relative (vs MU/Samsung/SK Hynix) plus a cyclically-adjusted reverse read -- memory is valued on mid-cycle, not peak, earnings.
ASPs roll in 2026-27 as capacity (Nanya's own + CXMT + big-3 re-entry) floods legacy DRAM -- margins collapse toward the 2023-24 loss regime while ~NT$52B of capex is committed; earnings evaporate and the stock de-rates toward book/net-cash support. Not financial advice.
One more strong upcycle year, then normalization as DDR5 supply broadens and CXMT/big-3 add capacity -- value converges to mid-cycle earnings x a below-market cyclical multiple, cushioned by net cash.
Shortage persists into 2027 + new-fab volume layers on -- earnings hold near peak, net-cash-backed, and a sustained-margin narrative supports a re-rate on higher normalized EPS.
At ~NT$450-455 (~NT$1.55T / ~US$49B market cap) on a ~NT$196B (~US$6.2B) annualized peak run-rate, Nanya trades at ~8x annualized peak sales but only a low-teens (~13-14x) P/E on ANNUALIZED PEAK EPS (Q1'26 NT$8.41 x4 ~ NT$34). The low P/E is the classic memory trap: the peak multiple is optically low precisely because the market prices mean-reversion. On mid-cycle/normalized earnings the multiple is far higher, and on trough earnings (2023-24 losses) there is no P/E. The price embeds a belief that the legacy-DRAM shortage persists and the new fab adds volume before ASPs roll. Net cash (~NT$68B) plus the NT$78.72B placement is real downside support; the swing factor is entirely the ASP cycle.
SWOT
Strengths
- Pure-play mature-node DRAM with enormous operating leverage — ~68% gross margin at the top of the cycle on a largely-depreciated single fab.
- Fortress balance sheet post-turnaround — ~NT$68B net cash plus NT$78.72B (~US$2.5B) fresh strategic placement, funding fab expansion without dilution stress.
- Formosa Plastics Group backing (Nan Ya Plastics is controlling shareholder) -- capital and staying power through troughs.
- Customer-anchored expansion — SanDisk/Solidigm/Kioxia/Cisco 3-year supply deals + equity subscription de-risk the new fab.
- Not gated on EUV/leading-edge -- avoids the capex arms race Samsung/SK Hynix/Micron fight in HBM and sub-10nm.
Weaknesses
- Commodity single-product, single-fab concentration — earnings are a pure ASP bet with brutal downcycles (losses in 2023-24).
- No HBM and minimal DDR5 (~10% of revenue) — absent from the highest-growth, AI-driven memory tiers; the bulk of revenue is DDR4/DDR3/LPDDR4 legacy.
- Technology follower — relies on a node roadmap (1B/1C/1D) trailing the leaders; process heritage traces to licensed Micron technology.
- Bit shipments declining even as revenue soars -- Q1'26 growth is entirely price, which mean-reverts.
- Tiny global share (~1.6% of DRAM revenue) vs a Samsung/SK Hynix/Micron oligopoly controlling ~90%.
Opportunities
- Extended legacy-DRAM shortage as the big-3 stay HBM-focused — structurally tighter mature-node supply into 2027.
- New fab adding capacity to capture volume when its own supply is the constraint, not demand.
- Specialty/niche DRAM, LPDDR for edge/industrial/automotive/networking, and custom-AI memory projects diversifying beyond pure commodity.
- DDR5 mix-up as the ramp finally accelerates once DDR4 tightness eases -- richer ASP and server exposure.
- Western customers actively seeking a non-China, non-big-3 second source -- Nanya's structural role.
Threats
- The cycle turns — legacy-DRAM ASPs are notoriously mean-reverting; a demand air-pocket or big-3 re-entry into mature nodes collapses margins.
- CXMT (ChangXin, mainland China) scaling mature-node DRAM aggressively (~8% share, global #4) -- the direct structural threat to Nanya's commodity niche on price.
- Big-3 (Samsung/SK Hynix/Micron) could flood DDR4/DDR5 supply once HBM demand normalizes, crushing pricing.
- New-fab capex (~NT$52B) commits into a peak — classic memory-capacity-at-the-top risk if the ramp lands into a glut.
- Geopolitical/Taiwan-Strait and export-control exposure; FX (NT$/US$) swings on reported earnings.
Moats, dependencies & bottlenecks
Moats
erodes as CXMT builds comparable low-cost mature capacity Cost advantage is real at peak but not proprietary; it is depreciation + scale, not a defensible process lead.
Nan Ya Plastics as controlling shareholder lets Nanya survive troughs that would kill a standalone commodity player.
Customer allocation lock-in (3-yr supply deals + NT$78.72B equity placement) tied to the current shortage SanDisk/Solidigm/Kioxia/Cisco anchoring supply is a cycle-driven relationship, not a structural switching-cost moat.
Western supply-chain diversification gives Nanya strategic relevance, but it is a positioning edge, not pricing power.
Dependencies
Earnings are almost entirely ASP-driven; a price roll-over swings Nanya from ~68% GM to losses. The single biggest determinant of the stock.
Big-3 capacity discipline (Samsung/SK Hynix/Micron staying HBM-focused) Competitive-structural Nanya's whole thesis rests on the leaders NOT re-entering mature nodes; they could reverse this.
AMAT, Lam, KLA, Tokyo Electron) New-fab ramp and 1C/1D node migration depend on tool delivery and pricing amid an industry-wide tool crunch.
Nanya's DRAM process lineage traces to licensed Micron 1x/1y tech and the former Inotera JV; it is a technology follower on the node curve.
Single-country manufacturing; NT$/US$ moves and Taiwan-Strait risk feed directly into reported results.
Advantages
- Highest-torque way to play a mature-node DRAM shortage -- pure-play, no HBM dilution of the signal.
- Peak-cycle profitability and net-cash balance sheet fund expansion without dilution.
- Strategic second-source relevance to Western memory/storage/networking customers.
- Low leading-edge capex burden relative to the big-3's EUV/HBM arms race.
Weaknesses
- Commodity, single-product, single-fab, sub-scale (~1.6% share) price-taker.
- Zero HBM and minimal DDR5 -- absent from AI-memory secular growth.
- Earnings 100% ASP-cyclical with a proven history of trough losses.
- Direct structural threat from CXMT's low-cost mature-node expansion.
Bottlenecks
- DDR4/LPDDR4 scarcity is itself capping the DDR5 ramp — Nanya can't free legacy capacity fast enough to shift mix (DDR5 stuck ~10%).
- Single-fab capacity ceiling until the new fab's Q1 2027 equipment move-in -- upside is supply-constrained, not demand-constrained, right now.
- Node roadmap (1B -> 1C/1D) trails the leaders, limiting cost-per-bit competitiveness and exclusion from HBM/high-end.
- Capex funding + execution on a fab that adds material output -- a build-into-peak timing risk.
Top signals & trends
Top signals
Bullish while rising · TrendForce reported ~+81% QoQ industry DRAM revenue in 1Q26 (conventional DRAM contract prices +~93-98% QoQ); the single most important read on Nanya's earnings.
Bearish if they re-enter · Any Samsung/Micron/SK Hynix signal of returning DDR4/DDR5 supply is the top downside catalyst.
Bearish structural · China low-cost DRAM (~8% share, global #4) is the durable threat to Nanya's niche beyond this cycle.
Bullish while accelerating · e.g. Apr'26 revenue NT$25.49B (+40.3% MoM, +717% YoY) -- monthly cadence is a fast tell on ASP momentum.
On-schedule ramp = volume lever; slippage or landing into a glut = risk.
Bullish if rising · Break above ~10% signals the higher-ASP transition is finally underway.
Trends
Positive (for now) · Big-3 vacating mature nodes is the entire tailwind -- legacy DRAM margins currently exceed HBM.
Structural DDR4/DDR3/LPDDR4 tightness into 2026-27 is the core bull driver; also the core reversion risk.
Negative (structural) · The durable long-term overhang on commodity DRAM pricing and Nanya's share.
Western storage/networking players (SanDisk/Solidigm/Kioxia/Cisco) prepaying allocation and taking equity.
Negative (cyclical) · Industry-wide capacity additions at the top historically seed the next glut.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Lithography (DUV for mature nodes); tool supplier for the node roadmap.
Deposition/etch/CMP wafer-fab equipment for the new fab and 1C/1D.
Etch and deposition tools for DRAM manufacturing.
Process control / inspection / metrology.
Coater/developer and etch equipment.
Controlling shareholder and parent -- capital and group support (also PCB/materials adjacency).
Strategic 3-yr supply agreement + participant in the NT$78.72B (~US$2.5B) private placement -- reportedly its first equity stake in a memory maker.
Storage OEM; 3-yr supply deal + largest placement subscriber (~4%) securing DRAM allocation.
Enterprise SSD maker; supply agreement + placement participant.
Networking OEM; 3-yr supply agreement + placement backer.
PC / server / networking / industrial OEMs + memory-module makers Broad base buying standard DDR/LPDDR, MCP/eMCP, KGD and Elixir-brand modules via distribution.
US-listed pure-play DRAM/NAND, #3 globally (~22% share); the closest liquid comp and source of Nanya's legacy process heritage. In HBM/DDR5 where Nanya is not.
#1 DRAM (~38% share); sets the commodity cycle. Could flood mature-node supply.
#2 DRAM (~29% share), HBM leader; capacity-allocation decisions drive the shortage Nanya monetizes.
Taiwan niche/specialty DRAM + flash peer; overlaps in mature-node/specialty memory.
Taiwan foundry-DRAM; competes in mature-node commodity/specialty DRAM.
Mainland-China, private; fast-scaling low-cost mature-node DRAM (~8% share, global #4) -- the key structural threat in Nanya's exact lane. Context only, not a buy/own call.