Nanya Technology Rides DRAM Shortage to Blowout Quarter, But DDR5 Mix Stuck at 10% on Capacity Constraints
Q2 2026 earnings call, July 10, 2026 — Net income more than doubles as pure pricing power, not volume, drives results
Nanya Technology delivered one of its strongest quarters in years, with net sales of TWD 82.5 billion, up 68.2% quarter-over-quarter and a staggering 684.2% year-over-year. Gross margin expanded to 79.5% from 67.9% in the first quarter, while net income nearly doubled to TWD 50.2 billion, translating into a 60.8% net margin and earnings per share of TWD 14.66, up from TWD 8.41 in the prior quarter. Book value per share jumped to TWD 93.49 from TWD 62.25. The Taiwanese DRAM maker's first-half net profit now stands at TWD 76.25 billion, or TWD 23.38 per share.
President Pei-Ing Lee framed the results within a broader industry narrative: "The cyclicality has been mitigated in DRAM industry," he said, pointing to 11 profitable years out of the past 13 and cumulative profit of TWD 282 billion over that period. Management guided for continued operational improvement in the third quarter, though it was notably more cautious on the fourth quarter, saying it does not currently "see major opportunity of major change from Q4 versus Q3."
Pure Price, Not Volume, Behind the Beat
Bank of America's Simon Woo pressed management on the mechanics of the quarter, and the answers were revealing: shipments were flat quarter-over-quarter, and the cost structure barely moved. Every bit of the margin expansion came from a roughly 60% sequential increase in average selling prices. "The first question, the answer is yes," Lee confirmed when asked if ASP alone drove the result. This is an important distinction for investors — Nanya's earnings power in this cycle is almost entirely a function of pricing leverage in a supply-constrained market, not underlying demand growth or operating efficiency gains. That makes the sustainability of pricing the single most important variable for the stock going forward.
On the pricing outlook, Lee was directionally bullish but declined to quantify it. "The pricing trend may continue to go higher," he said, though he noted that longer-term agreements are becoming more stable while shorter-term contracts are seeing more pronounced increases. When pushed for magnitude, Lee repeated he "cannot give you a very specific magnitude."
Legacy DDR4 Still Dominates Despite Industry's DDR5 Shift
One of the more counterintuitive disclosures of the call was Nanya's product mix. Despite the industry-wide migration toward DDR5 across PCs, smartphones and servers, DDR5 represents only around 10% of Nanya's revenue, with DDR4 and low-power DDR4 combined accounting for 60-70%, and DDR3 making up the remainder. Lee explained this is a capacity allocation decision rather than a demand issue: "Right now, the reason why we're not doing more DDR5 is mostly because we don't have sufficient capacity to do more. On the other hand, in the market, customers can get DDR5 supply from more suppliers. But DDR4 and low-power DDR4 is more restricted." In other words, Nanya is deliberately concentrating its limited capacity on the legacy nodes where supply is tightest and its competitive position strongest, rather than chasing the more crowded DDR5 market — a rational but structurally limiting choice given DDR5's higher long-term ceiling.
New Fab and CapEx Timeline
Nanya's new fab remains on track for a first-phase ramp in 2028 at up to 30,000 wafers per month, with total planned capital expenditure of $16 billion for the full 45,000 wafer-per-month capacity, inclusive of construction. Woo extracted useful granularity here: construction costs will run "less than $3 billion, maybe $2 point some billion," with the remainder — the bulk of the $16 billion — going toward equipment, including EUV lithography tools slated for 2028 implementation, potentially applied at the 1C, 1D or 1E node depending on "equipment arrangement."
Despite a full-year 2026 CapEx budget of TWD 52 billion, only TWD 6.9 billion has actually been disbursed through the first half. Lee clarified this is a timing issue rather than a change in commitment: payments are only triggered once construction milestones pass engineering inspection and acceptance criteria, meaning "this number will increase a lot more in the upcoming quarters." Investors should expect a materially heavier CapEx cadence in the back half of the year.
Cash Pile Swells on Private Placement
Nanya's balance sheet strengthened dramatically, with net cash rising to TWD 198.4 billion from TWD 68.5 billion in the prior quarter, driven both by TWD 55 billion of operating cash flow and a TWD 79.5 billion cash injection from a private placement completed during the quarter. On capital returns, Lee indicated a dividend payout ratio of roughly "40% plus/minus" of net profit, while also confirming the company has been reserving cash each quarter for employee profit-sharing bonuses — a detail that explains the elevated SG&A and R&D expense lines, both of which management attributed partly to "higher employee profit sharing."
Government Support Lagging Behind Logic Peers
In a candid response to a webcast question, Lee acknowledged that Taiwan's government support for the DRAM sector trails what it has extended to logic players like TSMC. "DRAM industry is lagging behind logic industry, for sure," he said, adding that government structure across central, local and departmental levels is "quite complicated," with some units "very, very helpful" and others "not so helpful." Lee said Nanya would push for a unified government office to streamline support rather than navigating "10 departments to get one thing done" — a structural headwind investors should note as peers in Korea, the U.S. and China benefit from more centralized state backing.
NVIDIA Design Win Speculation Left Unconfirmed
A webcast question referenced market reports that Nanya's TSMC-supported LPDDR product had been designed into the memory supply chain for NVIDIA's next-generation Vera Rubin AI platform. Management declined to confirm or deny, citing customer confidentiality: "I must say that this is related to our customers' confidentiality, and Nanya has to respect our customer's trade secret." AI infrastructure and server-related products already contribute more than 20% of total revenue, and the company is also engaged in multiple customized DRAM and wafer-to-wafer bonding projects, though Lee offered no revenue timeline, saying progress "depends on which customer... may be more successful or taking longer time."
Structural Shortage Narrative, With a Caveat on Legacy Cyclicality
Management's core market thesis remains that AI-driven demand is structurally tightening supply across HBM, low-power DDR5 and DDR5, with multiyear long-term agreements increasingly aligning supply and demand beyond 2028. However, when asked about a competitor's warning on cyclicality, Lee did not dismiss the risk entirely, conceding that "especially in those low-ticket items... the ecosystem may need some help" and that business conditions are "tougher now" for non-AI, lower-end product lines. He maintained, however, that this pressure has not yet undermined the broader supply shortage narrative. On contract structure, Lee said Nanya would consider adopting stricter long-term agreement terms — including take-or-pay clauses, price floors and ceilings, and cash prepayments — in response to an analyst suggestion aimed at preventing counterparty defaults during future downturns, an indication that management is actively hardening its commercial terms against the next cyclical trough even as it enjoys the current upswing.
Nanya Technology Corporation Deep Dive: The Accidental Beneficiary of the AI Memory Squeeze
The Business Model: A Pure-Play Legacy DRAM Engine
Nanya Technology Corporation operates as a pure-play Integrated Device Manufacturer in the Dynamic Random Access Memory sector. Unlike diversified semiconductor giants, Nanya focuses exclusively on the design, manufacturing, and sale of memory chips. The company's product portfolio is heavily weighted toward legacy and specialty memory, specifically DDR4, LPDDR4, and DDR3, which collectively account for the vast majority of its output, while next-generation DDR5 currently represents roughly 10 percent of total revenue. Nanya monetizes these products by supplying a broad base of end customers across consumer electronics, networking infrastructure, industrial applications, and the automotive sector. The company operates its own 12-inch wafer fabrication facilities in Taiwan, allowing it to tightly control its manufacturing processes and product quality.
While Nanya lacks the sheer scale of the industry's dominant players, its business model is highly sensitive to supply and demand imbalances in the broader memory market. Because DRAM is a commoditized product, Nanya's revenue and profitability are dictated almost entirely by global bit demand and average selling prices. The company relies on its independent 10-nanometer class process technologies, currently transitioning to 1B and 1C nodes, to remain competitive on cost and performance. By eschewing the bleeding edge of logic and ultra-high-density memory, Nanya has historically carved out a sustainable, albeit highly cyclical, niche serving customers who require reliable, cost-effective memory solutions rather than the absolute highest bandwidth available.
Industry Dynamics: The HBM Vacuum and a Historic Supply Squeeze
The global DRAM market is currently undergoing a profound structural shift, creating an unprecedented windfall for Nanya Technology. The industry is dominated by an oligopoly known as the Big Three—Samsung, SK Hynix, and Micron—which collectively control approximately 90 percent of global DRAM revenue. Driven by the explosive demand for artificial intelligence infrastructure, these dominant incumbents have aggressively reallocated their wafer capacity away from traditional DDR4 and LPDDR4 production. Instead, they are prioritizing High Bandwidth Memory and high-density DDR5 to feed AI servers. Because High Bandwidth Memory requires significantly more wafer area to produce the same bit output as standard DRAM, this capacity shift has effectively cannibalized the supply of legacy memory.
This reallocation has created a massive supply vacuum in the exact segments where Nanya operates. With the Big Three deliberately underinvesting in legacy nodes, the market for standard DDR4 and LPDDR4 has experienced a severe shortage, leading to a historic pricing squeeze. The financial impact on Nanya has been staggering. In the first quarter of 2026, the company saw its average selling prices surge by more than 70 percent sequentially, even as bit shipments slightly declined. By the second quarter of 2026, Nanya reported a 684 percent year-over-year revenue increase to NT$82.55 billion, while net income skyrocketed by 1,324 percent to NT$50.19 billion. Most remarkably, gross margins expanded to an unheard-of 79.5 percent, up from a negative 20.6 percent in the same period a year earlier. Nanya, which holds a modest 1 to 2 percent global market share by revenue, has become the ultimate accidental beneficiary of the AI data center boom, profiting immensely from the scarcity left in its wake.
Competitive Advantages and Strategic Validation
Nanya's primary competitive advantage lies in its position as an independent, geopolitically secure supplier of essential memory components. As the Big Three vacate the legacy DRAM space, original equipment manufacturers are scrambling to secure reliable sources of DDR4 and standard DDR5. Nanya's proprietary 1B and 1C process nodes allow it to manufacture these chips with a competitive cost structure, without relying on licensed technology from larger rivals. Furthermore, operating out of Taiwan provides a critical layer of supply chain security for Western and allied tech companies seeking to avoid reliance on Chinese memory manufacturers amid ongoing geopolitical tensions.
This strategic importance was emphatically validated in April 2026, when Nanya completed a NT$78.72 billion private placement. The equity was subscribed by four major global technology players—SanDisk, Solidigm, Kioxia, and Cisco—who collectively took a 10.19 percent ownership stake in the company. This transaction is highly unusual for the memory sector and serves as a powerful endorsement of Nanya's long-term viability. By taking equity stakes, these NAND flash and networking giants are effectively locking in their supply of companion DRAM chips. Consequently, Nanya has secured multi-year long-term agreements extending up to three years, insulating the company from future cyclical downturns and providing a guaranteed baseload of demand as it scales its operations.
New Entrants and the CXMT Threat
While the current pricing environment is immensely favorable, Nanya faces a formidable structural threat from a rapidly rising new entrant: China's ChangXin Memory Technologies. Backed by aggressive state support, ChangXin Memory Technologies has capitalized on domestic Chinese demand to rapidly expand its footprint. By the first quarter of 2026, the company's global DRAM market share surged to approximately 8 percent, firmly establishing it as the world's fourth-largest DRAM maker and overtaking Nanya in sheer volume. The Chinese manufacturer is targeting a production capacity of roughly 350,000 wafers per month by the end of 2026, a scale that begins to rival the raw output of incumbents like Micron. The company has also secured massive commercial anchors, including a recent $2.94 billion supply agreement with Tencent for server DRAM, proving its chips are viable for hyperscale applications.
However, this ascent is heavily constrained by structural and geopolitical limitations. Due to stringent United States export controls, ChangXin Memory Technologies is barred from acquiring extreme ultraviolet lithography equipment. As a result, the company is forced to rely on older 16-nanometer deep ultraviolet tools and complex multi-patterning techniques. This technological ceiling means its DDR5 dies are roughly 40 percent larger than equivalent chips from industry leaders, resulting in significantly higher costs per bit and lower overall competitiveness on the open market. Furthermore, the broader DRAM market is currently so starved for supply that these rapid capacity additions have not crashed global pricing; the Chinese firm is reportedly pricing its chips only 5 to 10 percent below the incumbents. While this new entrant will undoubtedly dominate the domestic Chinese market and pressure commodity DDR4 pricing over the long term, its inability to access leading-edge equipment limits its threat to Nanya's higher-margin, internationally bound product lines.
Future Growth Drivers: Edge AI and Fab 5A
To capitalize on its current cash windfall and avoid being permanently relegated to legacy nodes, Nanya is aggressively investing in future growth drivers. The most significant of these is the development of customized Ultra Wide Input Output memory. While Nanya does not produce the complex High Bandwidth Memory used in massive data center graphics processing units, its custom DRAM is specifically engineered for edge AI and on-device AI inference. These applications require high bandwidth, low power consumption, and compact memory-logic integration, but do not need the extreme density of data center memory. By pairing this architecture with advanced wafer-on-wafer packaging, Nanya is building a differentiated, lower-cost solution for the edge AI market, a segment that is expected to see explosive growth as AI capabilities migrate from the cloud to consumer devices.
To support this technological leap, Nanya is embarking on a massive capacity expansion. In July 2026, the company outlined a preliminary capital expenditure plan exceeding NT$200 billion, or roughly $6.2 billion, for 2027. This figure is approximately four times the company's 2026 capital budget and represents a monumental bet on the future. The capital will primarily fund the equipment move-in and ramp-up of Nanya's new Fab 5A facility. Once fully operational, Fab 5A is expected to nearly double the company's total production capacity and facilitate a full transition to advanced 1B and 1C nodes, ensuring Nanya can produce higher-density DDR5 and customized AI memory at scale.
Management Track Record
Under the leadership of President Pei-Ing Lee, Nanya's management team has demonstrated exceptional tactical agility and a willingness to make bold capital allocation decisions. The memory industry is notoriously brutal during cyclical downturns, and management navigated the severe 2023 to 2025 slump by strictly controlling inventory and reigning in capital expenditures. When the market began to turn in late 2025, management exhibited remarkable pricing discipline. Rather than rushing to clear inventory, Nanya actively paused public price quoting to manage sales, correctly anticipating that the supply squeeze would send legacy DRAM prices parabolic.
This opportunistic strategy has paid off handsomely, transforming a company that posted gross losses just a year prior into a cash-generating machine with near 80 percent gross margins. Management's decision to immediately parlay this windfall into the NT$78.72 billion strategic private placement and the massive NT$200 billion Fab 5A expansion shows a clear understanding of the industry's stakes. Rather than simply returning all excess cash to shareholders, leadership is utilizing this rare, highly profitable window to permanently upgrade Nanya's technological capabilities and secure its relevance in the impending DDR5 and edge AI eras.
The Scorecard
Nanya Technology is currently experiencing a generational financial windfall, driven not by its own technological supremacy, but by a structural anomaly in the global memory market. As the industry's dominant players abandon legacy DRAM to chase the AI data center boom, Nanya has been left to harvest extraordinary margins in a severely supply-constrained market. The company's recent financial performance, highlighted by 79.5 percent gross margins and a 684 percent surge in revenue, demonstrates the sheer leverage it holds in this unique environment. Furthermore, the strategic equity investments from major global tech firms provide a robust floor for future demand, insulating Nanya from the typical boom-and-bust volatility of the commodity memory sector.
However, the long-term thesis requires careful monitoring of the competitive landscape, specifically the aggressive expansion of China's ChangXin Memory Technologies. While United States sanctions currently handicap the Chinese competitor's technological efficiency, its raw wafer output will soon rival the industry's majors, posing a latent threat to commodity pricing once the current supply squeeze normalizes. Nanya's ultimate success will depend on management's ability to execute the massive $6.2 billion Fab 5A expansion and successfully commercialize its customized memory for edge AI. For now, Nanya stands as a highly profitable, strategically vital supplier in a market desperate for its products, making it a compelling, albeit cyclical, asset.