SK hynix has just released its strongest-ever Q2 financial report, but due to overly high market expectations, the market reaction has been very poor. Operating profit saw a significant increase but still fell below market expectations. Revenue also fell short of expectations, leading to a post-market price drop. The outstanding performance did not result in a higher premium, and the expectation gap has instead become the focus of trading.
The issue of the expectation gap was actually anticipated by the market not long ago. Two weeks ago, the Korean investment firm KIS lowered hynix's profit forecast, but the market's weakness turned out to be more severe than we had imagined.
Rumors are circulating in the market about the content of hynix's post-earnings conference call. The President's speech was actually not very innovative, and the questions from Wall Street institutions that followed are more worthy of attention. Because what they asked about is exactly what investors are most concerned about.
Q2 2026 Operating Performance (President Hong Hyun-joong)
In the second quarter, the strong demand from AI infrastructure investment expansion continued amid a tight supply environment, maintaining an upward price trend. DRAM and NAND once again saw significant price increases from the previous quarter, with server DRAM and enterprise SSDs, among other AI-related products, being the main drivers of this round of growth. In the second quarter, revenue was 79.3 trillion Korean won, up 51% QoQ, up 257% YoY, hitting a new historical high following the previous quarter.
DRAM
Under constrained supply conditions, the company expanded sales focusing on HBM and AI server DRAM products, achieving a high single-digit percentage growth in shipments QoQ, in line with previous guidance. Sales of server LPDDR products (including SOCAMM) saw significant growth; supported by the continuous strengthening of traditional DRAM prices, DRAM ASP rose by approximately 30% QoQ.
NAND
Building upon a low base caused by shipment contraction in the first quarter, coupled with expanded sales of enterprise SSDs, NAND shipments recorded a mid-double-digit growth QoQ in line with guidance. Revenue from enterprise SSDs doubled compared to the previous quarter, and revenue from large-capacity enterprise SSDs of 30TB and above more than doubled from the previous quarter. Driven by comprehensive price increases across all products, simultaneous price hikes in DRAM and NAND, and cost structure improvements, NAND ASP rose by 50% at the mid-tier level QoQ.
Profit and Loss
In the second quarter, operating profit was 60.5 trillion Korean won, representing a 61% increase QoQ and a 557% increase YoY; the operating profit margin increased by 5 percentage points to 76% QoQ, reaching an all-time high in both operating profit and margin. Depreciation and amortization in the second quarter were 4 trillion Korean won, with EBITDA at 64.6 trillion Korean won and an EBITDA margin of 81%. Net non-operating income was 62.2 trillion Korean won, including a 1.1 trillion Korean won exchange gain and a 63.3 trillion Korean won gain from asset disposal and valuation. As a result, pre-tax profit was 122.7 trillion Korean won, net profit was 93.9 trillion Korean won, and the net profit margin was 118%.
Financial Position
As of the end of the second quarter, cash and cash equivalents (including short-term investments) amounted to 88 trillion Korean won, an increase of 33.6 trillion Korean won from the end of the previous quarter; interest-bearing debt decreased by 0.7 trillion Korean won to 18.6 trillion Korean won. Net cash position expanded to 69.4 trillion Korean won, and the debt-to-equity ratio improved by 5 percentage points to 7% from the end of the previous quarter.
Market Outlook
AI technology is evolving into an Agentic form that can perform complex tasks on behalf of users for extended periods. With AI increasingly integrated into various services such as search, programming, and productivity tools, the scope of demand continues to expand. From a storage perspective, in addition to high-performance storage such as HBM required to enhance AI server performance and scale system, the demand for server DRAM supporting Agent-type services is also increasing; the role of high-performance enterprise SSDs is also expanding to more efficiently handle the continuous output generated by AI. This results in a structural shift in demand towards synchronized growth in AI storage and traditional storage.
Meanwhile, as AI models improve and software optimization advances, the computational power and cost for individual tasks continue to decline. The company believes that efficiency gains will not suppress overall infrastructure demand but will instead lower the price and usage barriers for AI services, thereby expanding the user base and application scope.
Major tech clients are increasing infrastructure investments due to rising AI service usage and a shortage of computing power; supported by the revenue and profit growth from AI services, their storage purchases appear to be continuing to expand. In fact, major clients are still requesting increased storage supply. In the PC and mobile application sectors, temporary sales adjustments have occurred due to storage shortages; however, with easing supply constraints and accelerated AI service adoption, these segments are expected to gradually regain growth momentum.
Under supply constraints, DRAM demand is expected to grow by the mid-teens, while NAND demand is expected to grow by the high-single digits. If subsequent supply constraints ease and pent-up demand is met, the market's growth trajectory could move even higher.
On the supply side, meaningful supply-demand balance improvement is unlikely to occur in the short term. This is due to the increasing complexity of advanced processes used in HBM and AI server storage, as well as the construction lead times required for new capacity additions. The supply-demand tightness is expected to persist for a considerable period of time.
Regarding medium- to long-term supply stability, the company is engaging in discussions with customers on Long-Term Agreements (LTAs). As of now, LTAs have been concluded with approximately 10 customers, including key customers, and discussions are ongoing with other major industry participants. These LTAs are more than mere quantity arrangements; they are strategic partnerships aimed at ensuring medium- to long-term supply stability and aligning with customers’ technology roadmaps for the development of next-generation storage products. The specific pricing structures vary based on customer and product characteristics, designed to address price fluctuations; simultaneously, financial mechanisms such as deposits are introduced to ensure contract fulfillment and enhance the visibility and reliability of customer's medium- to long-term demand plans. Building upon this foundation, the company will enhance investment and production operational efficiency to solidify the groundwork for medium- to long-term operational stability and sustainable growth.
Third Quarter Business Outlook and Technological Advancements
DRAM shipments in the third quarter are expected to increase by approximately 10% from the second quarter, with a focus on server products to actively respond to demand; NAND bit shipments are expected to decline by low single digits sequentially.
As AI models become increasingly complex and performance requirements for storage rise, the competitive landscape has expanded from the design of a single storage product to system architecture and packaging technologies. Relying on a product portfolio that includes DRAM and NAND, including HBM, and the ability to co-develop with customers, the company will lead storage innovation at the system level.
HBM4: Through ongoing product optimization, the company has achieved industry-leading energy efficiency and cost competitiveness while meeting the data processing speed required by customers, demonstrating distinctive technological capabilities. The company began mass production shipments in the second quarter and plans to ramp up in the second half of the year.
HBM4E: Samples have been sent to a major customer in the first half of the year. The product adopts a mature and mass production-stable optimized process, and the follow-up development is expected to progress smoothly. With a stable supply capability and cost competitiveness supported by high yield and excellent quality, along with industry-leading performance, the company will continue to maintain its leading position in HBM.
Traditional DRAM: In the second quarter, full-scale supply of SOCAMM2 products based on the 1c nm process has begun. The company will optimize the product line according to customer development progress and prepare samples to expand its customer base.
NAND: Accelerating the transition to advanced processes, focusing on high-capacity, high-performance products to match market demand. In the previous quarter, 321-layer products accounted for the highest proportion of NAND output, and the company plans to increase its share in domestic production capacity to around 50% by the end of the year as originally scheduled.
In a market environment of continued supply-demand imbalance, stable supply capability— the ability to deliver the required quantity to customers at the requested time— is now on par with technical prowess and has become a core operational competitive advantage. To respond to strong customer demand and medium- to long-term growth opportunities, the company will continue to advance its expansion investment in the short term to enhance supply response capacity: Advancing the mass production schedule of M15X ahead of time and increasing investment to rapidly expand capacity; D1 in Hwaseong Fab 1 is set to complete cleanroom line openings by early 2027. Due to the accelerated progress and expanded investment scale, capital expenditures for 2026 are expected to reach the higher end of the 4 trillion KRW range.
In the medium to long term, based on discussions with customers and market demand forecasts, the company will proactively establish future capacity reserve infrastructure. Recent new investment plans have been announced to enhance advanced packaging capabilities and the new NAND production site M17; simultaneously, a medium- to long-term plan for the construction of a new semiconductor cluster in South Korea has been revealed to address long-term demand after Hwaseong. The subsequent actual construction, equipment installation, and capacity expansion will be phased in consideration of factors such as customer demand visibility and investment efficiency. The company will strengthen its supply response capacity and financial soundness without delaying medium- to long-term growth opportunities while maintaining capital expenditure discipline.
About ADR Issuance
On July 10, the company's ADR was successfully listed on the NASDAQ market in the United States, representing the largest issuance by a foreign company in a U.S. IPO. This listing is significant not only for fundraising but also for confirming global market trust in the company's technological competitiveness and growth potential, while also expanding the company's connection points to the next-generation computing ecosystem. Building on this foundation, the company will strengthen strategic cooperation with key customers and partners, explore new business opportunities, and contribute to the semiconductor industry's development and AI system growth through continuous technological innovation.
Financial Strength and Shareholder Return
Driven by record-high profitability and cash generation, the company's financial capability has been further strengthened. At the same time, as structural growth opportunities in the AI era continue to expand, the scale of investment needed to capture these opportunities has significantly increased compared to the past. In this environment, the company prioritizes investments in growth opportunities that can generate high profitability and strategic value, while striving to build a financially resilient structure that can sustain stable operations even amid market volatility, and intends to continue sharing the outcomes with shareholders.
Despite the expected increase in future investment needs, the company believes that its significantly enhanced cash generation capability is sufficient to meaningfully expand shareholder returns while achieving future growth investment goals and maintaining financial strength objectives. The company is currently reviewing various additional shareholder return implementation plans from multiple perspectives.
Q&A Session
Question 1: J.P. Morgan, Jay Kwon
Q: Recently, some large tech companies have been considering leasing data centers, while more efficient AI models continue to emerge. As a result, the market is concerned that AI infrastructure investment may slow down or even decline. Based on the company's communication with customers, how does the company view the AI infrastructure investment trends of major CSPs? What does this mean for HBM, DRAM, and NAND demand?
A: The company understands the origin of these concerns. However, the company does not view these trends as a signal of a slowdown in AI investment but as a transition to increasing the utilization of the already large-scale AI infrastructure and accelerating monetization.
For major CSPs, AI competitiveness is closely tied to their core competencies in search, advertising, cloud services, software, and the like, so investments aimed at strengthening AI capabilities are likely to remain robust.
Likewise, the company does not believe that more efficient AI models will reduce infrastructure demand. After model and system efficiency improvements, the same infrastructure can support more users and services, thereby expanding AI's accessibility and adoption. The recent explosive growth in demand following the emergence of highly efficient AI models is evidence that efficiency improvements lead to broader AI adoption and usage, rather than a decrease in infrastructure demand.
This assessment is also supported by discussions with key customers about their medium- to long-term demand outlook. While the timing of individual projects may vary due to factors such as power supply and data center construction, the company believes that AI infrastructure investment will remain robust beyond next year, supported by AI competition among CSPs and the continued expansion of AI services.
Therefore, the overall storage demand will continue to expand: not only for HBM used for AI computing power, but also including server DRAM supporting Agentic AI, and high-performance, high-capacity NAND for AI service expansion and data growth.
Question 2: Hana Securities, Kim Rok-ho
Question: The company recently announced a significant mid- to long-term expansion plan. What is the long-term storage demand assessment that underpins this strategy? Does this include demand locked in through long-term agreements? Additionally, the increase in capacity has raised concerns in the market about potential oversupply. How does the company view this?
Answer: The company's mid- to long-term capacity strategy is based on structural growth in storage demand being driven by AI expansion, as well as ongoing discussions with core customers regarding longer-term demand.
Recent cooperation between the company and its customers is evolving from transactional relationships to more strategic long-term partnerships. The willingness of customers to sign long-term agreements and build partnerships is in itself evidence of the sustainability of AI ecosystem demand.
SK hynix's current capacity expansion is based on market demand visibility obtained through customer partnerships. Actual capital investments and capacity ramp-up will be phased based on demand visibility, investment efficiency, and other factors, with capacity expansion being flexibly aligned with confirmed customer demand. The company believes that the mid- to long-term investment plan will not immediately lead to oversupply.
Question 3: Meritz Securities, Kim Sunwoo
Question: The question is about Long-Term Agreements (LTA). Competitors have recently completed and disclosed LTAs. Although the company has briefly mentioned this in its performance briefing, could you further elaborate on SK hynix's LTA framework, such as contract duration and pricing structure?
Answer: The LTAs that the company is currently discussing with customers take various forms, with differentiated designs based on customer and product characteristics. The contract duration is usually around five years, and specific terms vary depending on the customer and product.
The pricing structure will also not be uniform. The company is exploring various pricing mechanisms with customers to better address price fluctuations, aiming to reduce the uncertainty brought about by short-term market volatility while enhancing the long-term operational stability for both customers and the company.
At the same time, considering the impact of demand fluctuations on the storage cycle, achieving effective procurement commitments is equally important. Therefore, in addition to long-term volume commitments, the agreement also includes mechanisms such as a deposit to strengthen contract performance and demand visibility, with specific terms varying depending on each customer's requirements and contract structure. This structure allows customers to develop more reliable long-term procurement plans while enabling the company to optimize investment and production planning based on improved demand visibility.
Regarding the percentage of sales covered by LTAs, the company is currently unable to provide specific figures but will maintain it at an appropriate level based on market conditions and customer demands: enhancing performance resilience on the downside and retaining flexibility to capture incremental demand and growth opportunities in a market upturn.
Building on long-term collaborations with major AI customers such as NVIDIA, the company has established a solid profit foundation in HBM. Looking ahead, the company will continue to strengthen its leading position in HBM and leverage the demand visibility and operational flexibility obtained through LTAs to achieve a balance between stability and profitability.
Question 4: Daiwa Capital Markets, SK Kim
Question: The question is about DRAM. The second-quarter DRAM ASP growth seems to be lower than market expectations. What are the reasons for this? What is the outlook for the second half of the year?
Answer: Based on customer demand and medium- to long-term product strategies, the company manages the sales mix between HBM and traditional DRAM. In the second quarter, the shipment of some high-value-added products was postponed to the second half of the year, and the product mix changes had an impact on the blended ASP; these factors are expected to gradually ease in the second half of the year.
With HBM4 shipments ramping up comprehensively and increasing shipments of 1z nm traditional DRAM, it is expected that bit growth in the second half of the year will be higher than the first half.
Furthermore, considering changes in customer demand and product mix, the growth in HBM4 sales and the rise in high-value-added product contributions will also have a positive impact on the blended ASP. Increased shipments combined with ongoing product mix enhancements will drive ASP and performance improvement in the second half of the year.
In terms of sales strategy, the company does not focus on short-term price fluctuations or short-term profits but takes into account demand visibility, long-term customer relationships, and the supply-demand situation of each product segment as a whole. This principle will remain unchanged, and the company will achieve stable and sustainable performance growth while seizing market growth opportunities.
Question 5: SK Securities, Han Dong-hee
Question: The question is about HBM. There is a market view that competitors have made rapid progress in HBM recently. Where does the competitiveness of the company's HBM4 lie? What is the key differentiating factor to maintain the leading position in the HBM market?
Answer: The competitiveness of HBM4 depends not only on whether it can deliver the required performance but also on whether it can achieve scaled supply through stable yields and consistent quality.
Since the HBM2 generation, SK hynix has continuously validated these capabilities. The competitiveness the company has built in terms of listing timing, product performance, mass production yield, quality, and customer trust cannot be replicated in the short term.
Building on this, the company has started mass production of HBM4 for core customers in the second quarter. Currently, the yield and quality of HBM4 are approaching the mature stage of HBM3, and the current focus is on steadily expanding production capacity.
As mentioned earlier, the company has also completed customer sampling of HBM4E. This product utilizes an optimized manufacturing process with verified technical maturity and production stability, following the roadmap smoothly with the goal of starting mass production in 2027.
The company's preparations go beyond this and also include forward-looking planning for next-generation technologies. In addition to hybrid bonding, the company is developing technology for heat dissipation for future products such as HBM5: this technology integrates heat dissipation elements into the package, expecting to reduce thermal resistance by over 30%, thereby enhancing system stability and operational efficiency in a high-performance, high-density AI environment. (The abbreviation of this technology is read as "IBM" in the transcript and is described here based on its functionality.)
As the AI market continues to expand and AI accelerators become increasingly complex in performance and packaging, the company believes that customers will place greater value on a partner with validated manufacturing capabilities, quality, and reliable supply. HBM is a high-value-added product, and if quality issues occur, it would bring significant costs to customers and have a wide-ranging impact on the entire system.
Based on early co-development experience with customers and long-term strategic partnerships, the company will continue to reliably deliver the right products at the right time and lead the transition to next-generation technologies to maintain its leading position in the HBM market.
Question Six: UBS, Nicolas Gaudois
Question: How is the 2027 HBM price negotiation progressing? Could you please elaborate on the contract discussions, including HBM4E and HBM4, as well as the expected pricing?
Answer: The company is currently in discussions with key customers regarding the 2027 HBM supply volume and pricing. With solid customer demand support, the negotiations are progressing well. The contract terms and pricing details with individual customers cannot be disclosed.
Over the past few months, traditional DRAM prices have risen significantly. This market environment may have a certain impact on the price discussions for HBM; however, the price of HBM is not determined solely by traditional DRAM prices.
Compared to traditional DRAM, HBM requires a much greater investment of resources, including more wafer starts, advanced manufacturing processes, TSV, and packaging capabilities. With each product generation, customer requirements for performance and quality continue to rise, and product development and certification are becoming increasingly complex.
Therefore, the company's price discussions will take into account a range of factors: traditional DRAM prices and market supply-demand dynamics, resources and opportunity costs related to HBM production, technological complexity, and the value the product creates for customers. The goal is to achieve a reasonable level of profitability that aligns with the differentiated value provided while driving a healthy AI ecosystem and sustainable growth.
With its accumulated technological leadership, cost competitiveness, stable manufacturing capabilities, and trust and collaboration with customers, the company will maintain the solid profitability of the HBM business through successful product generation migrations and ongoing customer value creation, solidifying its strategic partner position in the AI era for mutual growth with customers, focusing on long-term sustainable growth and profitability.
Question Seven: CLSA Securities Korea, Sanjeev Rana
Question: The question is about capacity expansion. In addition to the recent large-scale investment announced in Korea, the market is also discussing expanding production overseas to the U.S., Japan, and other countries. Could you please provide detailed information on the company's investment strategy and direction in Korea and overseas?
Answer: In the AI era, being merely technologically advanced is not enough; the ability to supply the required quantities at the right time has also become a key competitive element. Especially at this point of extremely tight supply, it is the supplier's responsibility to provide the necessary storage products to the ecosystem.
The company's medium- to long-term investment direction is: to invest in AI-driven storage as needed, while executing capital expenditure based on business feasibility and investment efficiency. In the medium to long term, it will ensure additional manufacturing capacity by maximizing the utilization of existing production sites and building new infrastructure where necessary.
Within South Korea, the company will continue to position Icheon and Pyeongtaek as the core production hubs for next-generation DRAM and AI storage, while enhancing Cheongju's manufacturing capabilities in NAND and advanced packaging. (The transcription lacks or misidentifies the names of the sites, based on the company's current production site description.) The announced large-scale investments are also part of this strategy, proactively ensuring the manufacturing foundation and infrastructure needed to support future demand.
Regarding future production sites, the company does not distinguish between domestic and overseas locations but focuses on a comprehensive decision-making approach that considers factors such as integrated power supply, water resources, manpower, supply chain and semiconductor ecosystem, and customer accessibility.
It is worth noting that as of now, no decisions have been made other than the announced investments. In the future, the company will continue to ensure production sites are responsive to customer needs at the right time and will enhance investment efficiency by leveraging existing assets and evaluating additional investments.
Question 8: Daishin Securities, Ryu Hyung-kyun
Q: The question is about NAND. As inference demand expands and KV cache offload demand rapidly increases, how is the role of enterprise SSDs changing? Could you explain the company's strategy on various product segments, including QLC SSDs for HDD replacement and high-performance SSDs based on the SLC mode? Competition in these segmented markets also seems to be intensifying.
A: As observed in the question, the AI market is transitioning from a training-centric environment to an inference-centric one, with NAND rapidly becoming a core component in AI storage hierarchy. Therefore, NAND demand centered around SSDs is rapidly increasing, and the company sees this trend continuing.
At the same time, the company believes that the AI storage market cannot be served by a single technology: requirements for latency, throughput, power consumption, capacity, and TCO vary depending on the customer. What customers demand may not be a specific technology or medium; the key is whether each type of workload's performance requirements can be reliably delivered.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
