Nanya's Shortage Extends Through the Second Half A record, all-price quarter, and guidance that the pricing strength holds rather than recedes

Company snapshot

Nanya Technology (2408 TT) is a Taiwan-listed DRAM manufacturer headquartered in Taoyuan. It designs and produces commodity and specialty DRAM, spanning DDR4, DDR3 and LPDDR4/4X with a ramping DDR5 and LPDDR5/5X portfolio, on its own wafer fabs, and sells to consumer, PC, server, networking and industrial customers. It is small relative to the global memory majors, and its economics turn heavily on DRAM contract pricing because most of its manufacturing base is fully depreciated. The company is building a new fab to add capacity from 2028.

Results snapshot

1. A record quarter, and the surprise was the forward message

Nanya reported its second-quarter 2026 results at its investor conference on 10 July 2026, with net sales of NT$82.55bn, up 68.2% sequentially and 684.2% year on year, a record quarter and the first above NT$80bn. Gross profit reached NT$65.62bn and net income NT$50.19bn, for earnings per share of NT$14.66. A year earlier the company was loss-making at every line, so the year-on-year moves are shown as swings rather than percentages.

Nanya does not issue numeric revenue or earnings guidance, so the print is measured against consensus. The quarter's monthly revenues had already been reported to the market, so the surprise was in the forward message rather than the top line. Consensus had modelled third-quarter revenue near NT$78bn, about 5% below the second-quarter figure, on the assumption that the price cycle would begin to ease in the second half.

Management's commentary pointed the other way. It said second-quarter operational results should improve further into the third quarter and remain sustainable for several quarters. The first half now stands at NT$131.64bn of revenue and NT$76.25bn of net income, and the second-half question that had divided the market is the central subject of this report.

Net sales reached a record NT$82.55bn, up 68% quarter on quarter and almost eightfold year on year.

2. Revenue rose on price, not volume

The quarter was a price event. Average selling prices rose more than 60% quarter on quarter while bit shipments were flat. Against the year-earlier quarter, ASP rose more than 500% and shipments grew a high-twenties percentage. The exchange rate was broadly neutral sequentially and a low-single-digit tailwind year on year.

That split matters for the thesis. Growth carried by price rather than volume is the signature of a supply shortage in legacy DRAM, as the global majors redirect wafer capacity toward high-bandwidth memory for AI accelerators and leave less output for the DDR4, DDR5 and LPDDR products that Nanya sells. It is the same pattern the company reported in the first quarter, when ASP rose more than 70% and shipments fell. The second-quarter print confirms the mechanism a second time.

3. Depreciated fabs turned the price surge into a 79.5% margin

Gross margin reached 79.5% in the second quarter, up 11.6 points from 67.9% in the first. Operating margin reached 73.7% and net margin 60.8%. The company attributed the NT$32.3bn sequential rise in gross profit mainly to higher ASP, with operating expense up NT$1.6bn as employee profit-sharing and sales and research activity expanded alongside earnings.

The margin level is a function of the cost base. Most of Nanya's capacity is fully depreciated, so cost of goods sold rose only modestly while revenue expanded, and each increment of price fell through to profit. Cost of goods sold was NT$16.93bn, or 20.5% of sales, against 32.1% a quarter earlier. EBITDA margin reached 77.3%.

The forward question is durability. A margin this high is a cyclical peak that rests on DRAM contract pricing holding, and our forecasts assume gross margin normalizes over the coming years as new industry capacity arrives and the company's own new-fab depreciation begins to load. Near term, management's guidance and the multi-year supply agreements point to the level holding through the second half.

Margins moved from a loss a year earlier to a 79.5% gross margin as ASP levered the depreciated cost base.

4. The capacity response is funded, with bit growth still disciplined

Full-year 2026 capex is guided to about NT$52bn, roughly four times the NT$13.4bn spent in 2025, with around 30% earmarked for wafer-fab equipment. First-half capex was only NT$6.9bn, so the guidance implies a heavy second-half ramp as equipment moves in. The plan was approved by the board in March 2026.

The larger commitment is a new fab. Management set out a first-phase ramp toward 30,000 wafers per month by 2028 within a total plan of US$16bn for 45,000 wafers per month including construction, and said its 1C, 1D, 1E and EUV development is on schedule. The capacity response the market had been waiting for is now funded and dated. Even so, management guides 2026 bit shipments up only a high-teens percentage, so near-term supply stays disciplined and the pricing environment is not being met with a volume flood.

Full-year 2026 capex is guided to about NT$52bn, roughly four times 2025, funding the new-fab build.

5. The first half points to a bigger second half

On the Zero One model, the first half represents 42.9% of the full-year revenue forecast of NT$307bn and 44.1% of the NT$173bn net-income forecast. With management guiding the third quarter to improve further, the implied second half of NT$175bn in revenue sits above the first half, consistent with that message. The forecast was re-anchored on this print; the prior model, set to consensus in April, carried a full-year revenue figure of NT$247bn that the first-half actuals had already overtaken.

Our full-year net income of NT$173bn is in line with current consensus of about NT$173bn, and the earnings-per-share figure of NT$51.7 is close to the street's NT$51.8. The main point of difference is 2027, where our forecast carries a lower net margin than consensus as normalization begins earlier in our model. That is a deliberate view rather than a stale number, and it reflects the same peak-then-normalize path our margin section describes.

The forecast re-anchors full-year 2026 revenue to about NT$307bn, up from a NT$247bn April baseline that the first half had already overtaken.

6. Management expects the pricing strength to extend

Nanya does not publish numeric guidance, so the forward read comes from management's qualitative message. The company said second-quarter operational results should improve further into the third quarter and are sustainable in the next few quarters. It described the memory market as structurally changed by AI, with supply tightness expected to persist over the next several quarters and multi-year long-term agreements aligning supply and demand.

On demand, management said artificial-intelligence and general-purpose servers continue to drive strong demand for high-bandwidth memory and registered DIMMs, which constrains supply for smartphones, PCs, automotive and consumer electronics and drives price adjustments across those end markets. Revenue from AI infrastructure and servers is now more than 20% of the total. Taken together, the guidance is the clearest statement yet that the company sees the pricing strength extending rather than fading in the second half.

7. AI is redirecting memory capacity, and legacy DRAM benefits

The structural case is AI redirecting the memory industry. As the majors commit capacity to high-bandwidth memory, general-purpose DRAM supply tightens, and Nanya's legacy portfolio benefits from the resulting scarcity. Management framed this as a durable shift rather than a normal cyclical peak, supported by multi-year agreements and greenfield expansion aligned to demand beyond 2028.

The product roadmap is broadening toward AI-adjacent memory. The portfolio spans DDR5, LPDDR5 and 5X, DDR4, LPDDR4 and 4X, DDR3 and LPDDR3, and management said it supports customized AI and workload-optimized products and AI-infrastructure solutions. Higher-content, higher-performance DRAM in enterprise SSDs, SmartNICs and baseboard management controllers is part of the same demand story.

The balance sheet was recapitalized during the quarter. A NT$78.7bn private placement funded the capacity build and lifted net cash sharply. Who subscribed, and whether the placement carries the same allocation-style logic as the equity stakes memory customers took in the first quarter, is a question the release did not answer and one worth pursuing.

8. Net cash near NT$200bn raises the capital-return question

Nanya ended the quarter with NT$216.8bn of cash and NT$198.4bn of net cash, up from NT$68.5bn a quarter earlier. Operating cash flow was NT$55.01bn and capital expenditure NT$4.05bn, for free cash flow of NT$50.97bn. The NT$78.7bn private placement was the largest single financing item and the main driver of the cash build.

The company is strongly cash-generative even before the placement, with free cash flow covering the current pace of investment several times over. That position gives it room to fund the new-fab build from internal cash and the placement proceeds. With net cash near NT$200bn as the cycle peaks, capital-return policy is a reasonable question for coming quarters, though management did not address it in this release.

9. Pricing is the risk that reverses the story

Pricing is the dominant risk. Because growth is carried by ASP rather than volume, the earnings power reverses if DRAM contract prices roll over. The multi-year supply agreements and AI-driven tightness push that risk out, but they do not remove it, and the turn in a memory cycle can be fast.

China supply is the specific supply-side risk. Mainland capacity expansion, flagged by peers earlier in the quarter, could add legacy DRAM output into 2027 and cap pricing. Management's message of persistent tightness implicitly discounts this near term, but the release carried no explicit assessment of Chinese supply.

The margin level normalizes in our forecasts. A 79.5% gross margin is a cyclical peak; as industry capacity arrives and the new fab's depreciation loads, margins glide down in the outer years. Execution on the NT$52bn capex program and the US$16bn new-fab plan is the other risk to watch, since a large build into a peak carries the usual timing risk if demand softens as capacity arrives.

Financial Forecasts

The three statements below are drawn from the Zero One Investment Research financial model, re-anchored on the second-quarter print.