Sk Hynix Inc S/Gdr 144A

Good afternoon and evening. This is Park Sung Hwan, head of IR at SK Hynix. Welcome to the SK Hynix 2024 Second Quarter Earnings Release Conference Call. Allow me to introduce the executives present here today. We are joined by CFO Kim Woo Hyun, head of DRAM Marketing Kim Kyu Hyun, and head of NAND Marketing Kim Seok. Let me issue a disclaimer. All outlooks presented by the company are subject to change depending on the macroeconomic and market circumstances. With that, we'll now begin the SK Hynix earnings release conference call for second quarter 2024. Mr. Kim Se-ho will first present the earnings, followed by the company's future plans and market outlook, and a Q&A session with the attending executives. Good morning, everyone. Allow me to first introduce the company's performance for the second quarter of 2024. In the second quarter, AI memory product demand continued to hold very strong, while demand for conventional memory products also grew as customers tried to build inventory ahead of potential tightness in the future, which led to another quarter of meaningful price increase for both DRAM and NAND. As a result, our second quarter revenue reached 16.4 trillion won, an increase of 32% sequentially, and 125% year-on-year, marking the highest quarterly revenue in our company's history. DRAM saw a significant increase in sales of general server products, along with the expansion of HBM3e sales, leading to bid growth of low 20% sequentially above guidance of mid-10%. ASP rose mid-10% sequentially due to higher mix of premium products such as HBM and ServerDRAM, while prices rose across all products for three consecutive quarters. Particularly notable was HBM sales which drove overall revenue growth by growing over 80% from the previous quarter and over 250% year-on-year. For NAND, although sales of enterprise SSD and mobile products increased, total bid shipments decreased by low single digits sequentially due to reduction in sales of discrete products and client SSDs, where demand recovery remains relatively muted. Meanwhile, ASP increased by mid to high 10% as prices rose across all products. Due to revenue expansion from rising memory prices, higher mix of value added products, and favorable FX impacts, operating profit for the second quarter increased by 2.58 trillion won from the previous quarter to 5.47 trillion won, with an operating profit margin of 33%, up 10 percentage points from the previous quarter. Amidst rapid improvements of conventional DRAM profitability due to rising prices, the significant increase in sales of HBM3 products contributed to the growth in DRAM profits And margins also improved as a result of continuing ASB increases since fourth quarter of last year, expanded sales of enterprise SSDs, and ongoing focus on profit-oriented management. Depreciation and amortization expenses for the second quarter amounted to 3.12 trillion won, with sequential decline continuing due to reduced capital spending last year, EBITDA reached 8.59 trillion won, with an EBITDA margin of 52%. Non-operating loss net of gain for the second quarter was 0.42 trillion won, including net interest expense of 0.28 trillion won, and net foreign currency related loss of 0.6%. 0.16 trillion won. As a result, net pre-tax income was 5.05 trillion won and net profit for the quarter was 4.12 trillion won and net profit margin was 25%. Consolidated cash and cash equivalents, including short-term investments, at the end of second quarter was 9.7 trillion won, down 0.6 trillion won from the end of previous quarter. Meanwhile, interest bearing debt was 25.2 trillion won, down 4.3 trillion won from a quarter ago. As a result, debt to equity and net debt to equity ratio recorded 42% and 26% respectively, meaningfully improving compared to a quarter ago. Next, I'll discuss market outlook. Continuing from last year, demand for AI memory products is seeing another year of growth. Along with HBM, demand for high density server DRAM and enterprise SD are also rising. An increase in contents for PC and smartphone is also expected in order to support AI features on the device. Although memory suppliers are increasing production utilization this year, they are prioritizing to support HBM and enterprise SSD, which are showing faster demand growth than conventional memory products. In particular, When memory suppliers allocate more wafer capacity for HBM, production capabilities for other DRAM products can be constrained as HBM consumes more wafer capacity due to larger die size. As a result, DRAM prices are increasing continuously when demand for traditional end markets has not yet fully recovered. In the second half of this year, Strong demand from AI servers are expected to continue, as well as gradual recovery in conventional markets with the launch of AI-enabled PC and mobile devices. Looking at demand by application, the demand recovery in the PC market is trending relatively weaker than initial expectations. High inflation and cost burden due to rising memory prices are impacting not only downstream demand, but also product lineups. Customers are preparing for the launch of high spec products with AI features to create new demand, while also responding to the divergent demand trends by offering more affordable low spec products. As a result, high spec devices are expected to drive memory demand in the second half And in the longer term, increase in memory content and wider adoption of low power memory are expected to support optimal operation of AI PCs. Similarly, the smartphone market also showed more moderate demand recovery in the first half, but the release of new AI enabled flagship and foldable phones are expected to help improve that demand in the back half of the year. Smartphones adopting AI functions require higher content than existing high-end models. The further releases of these products that are scheduled in the second half to next year are expected to drive demand. In the server market, AI servers are driving demand growth with exponential increase in workload needs, as generative AI evolved to multimodal models, and also with the ever-increasing AI-related spending from big tech companies. Request from customers for additional HBM supply this year, and efforts to expand COAS capacity in the supply chain, also indicate strong demand for AI servers. In addition, general purpose server demand is also expected to gradually improve along with the impending replacement cycle for data center servers that were invested six to seven years ago, especially with growing needs to adopt new energy efficient platforms. Now, I'll discuss our plans. Third quarter DRAM bid shipment is expected to grow by low single digit percent sequentially, driven by further expansion of our HBM sales volume, where strong demand is evident. NAND bid shipment is expected to decrease by mid single digit percent sequentially, despite increase in ESSD sales volume, given yet soft end demand in conventional applications and relatively high customers' inventory. Net revenue growth, however, is expected to continue with rising prices and optimized product mix. HBM3 sales increased significantly in the second quarter as demand gained momentum. In the third quarter, HBM3e bid shipments are expected to be greater than that of HBM3, resulting in HBM3e accounting for more than half of total HBM sales volume this year. HBM3e 12 high product samples have already been shipped to major customers. and is on track to start mass production in this quarter, involving shipping to customers by fourth quarter. With a full product portfolio ranging from HBM2E to HBM3E 12 high products, we plan to maintain our competitive advantage in the HBM market. Our DDR5 products have been successful in securing early market leadership with stable technology and quality since first generation products. Currently, we are the sole supplier in the industry offering two 56 gigabyte products, maintaining our leadership of DDR5 products. In the second half of the year, We are planning to release 32 gigabit DDR5-based high-density server DRAM, as well as MCR-DIMM, which offers superior bandwidth compared to existing RDIMM to target high-performance computing market. We're closely collaborating with major server companies to successfully launch products that satisfy customers' requirements. For NAND, we achieved notable results in terms of revenue and profitability in the second quarter. We are planning to focus more on strengthening product mix and availability of product lineups for future improvement. We are currently expanding sales of high-density enterprise SSDs, where demand is increasing, and we are also raising production utilization for some of our NAND fabs to support growing demand. With the demand increase for enterprise SSDs this year, we have increased our enterprise SSD sales by 50% in the second quarter compared to previous quarter, and we expect our annual ESSD sales to grow by nearly four times compared that of last year. We plan to leave the market in the second half of the year with 60 terabyte ESSD, the only available ultra-high density product in the industry. We also plan to release 128TB enterprise SSDs in the beginning of next year to maintain our continued edge in the ultra-high density eSSD market. In addition, we plan to support upcoming AIPC demand by offering high performance and low power PCIe Gen5 client SSD that was developed in June. Construction for M15x, which was recently decided to support the growing demand for AI memory, as well as Yongin Cluster, which will be our next generation production site, is progressing as planned. Amid increasing infrastructure investment compared to past years, investment needs are also rising to meet demand of conventional DRAM as well as HBM, which requires more wafer capacity than regular DRAM. Therefore, this year's CAPEX level is expected to be higher than what we expected in the beginning of the year. Nonetheless, Our investment plans will be prudently decided based on end demand and profitability, and will be executed within the generated operating cash flow. We will continue to pursue financial soundness by executing investments efficiently, and this quarter's meaningful reduction in debt levels was in line with these efforts. Next, let me share our ESG management activities and performance. We issued sustainable report 2024, highlighting our ESG management activities and achievements during the year of 2023, as well as future commitments. In 2023, We achieved company-wide renewable electricity adoption rate of 30% level for the second consecutive year. Also, we received international standard certifications from the British Standards Institution for our compliance and anti-bribery management system based on our compliance framework. These certifications recognize our achievements in complying with domestic and international standards and proactive prevention of related risks. This quarter, we announced a joint declaration for greenhouse gas reduction with 28 members of the ECHO Alliance, composed of material component equipment suppliers. The greenhouse gas emissions of the participating companies account for approximately 50% of our Scope 3 emissions from purchased raw materials last year. We plan to implement various collaboration programs such as operating ESG funds and participating in government supported renewable energy projects to support our partners in achieving their greenhouse gas reduction targets. Thank you. With that, we are now ready to take your questions.

The first question will be presented by Nicola Godoa from UBS.

Yes, good morning and thanks for taking my question. You commented earlier on having seen some upside for conventional server DRAM demand. And I think we've seen the same in terms of factual server builds downstream. How did you see this translating into server DRAM demand upside on your side? Could you effectively quantify it versus what we expected earlier? And if we look at non-HBM server DRAM bid demand growth, could you quantify this as well for your expectations for 2024 and your initial view for 2025? Thanks.

Yes, thank you for your question.

Thank you very much for that question.

While the ramp of investment in AI servers that began last year continues, we're also seeing an improvement in demand for general purpose servers from some customers.

This is not only because of the impending replacement cycle for cloud data center servers, which were heavily invested in 2017 and 2018.

but also because of AI data center deployments that are driving demand for general-purpose servers.

AI servers with high power consumption have been rapidly increasing, and the cost of operation of data centers has been greatly reduced. Therefore, there is a movement to upgrade existing general servers to new server platforms with greatly improved power efficiency.

In particular, the rapid growth of AR servers with high power consumption has led to a move to upgrade existing conventional servers to new server platforms with significantly improved power efficiency to reduce operating costs and free up power across the data center.

As AI technology expands from learning to storage, the demand for AI servers will be maintained. Given that the demand for replacement servers will begin in earnest, the growth rate of server DRAMs except for HBMs this year and next year is about 20% or so.

Server DRAM excluding HBM is expected to see mid-20% level of demand growth this year and next year as AI technology expands from training to inference, driving robust AI server demand along with replacement demand for general purpose servers.

Next question please.

The next question will be presented by Peter Lee from Citigroup. I am Ita Cho from Citi Group.

Thank you very much for this opportunity to ask my question. I have a question about investment. With the recent announcements of increased investment and capacity utilization by memory companies, Concerns have been raised about supply growth beyond 2025. So, what is your company's view and direction regarding this matter?

Thank you for your question. Based on past experiences, you may think that the increase in investment in supply companies will soon cause supply overflows. However, the current investment and growth are due to the different trends in the market structure and production characteristics Thank you very much for that question.

So based on past experience, you may be concerned that increased investment by suppliers may lead to oversupply, but the current investment and ramp-up of production is centered on HBM, which has a much different market structure and production characteristics than conventional DRAM, So it's not really reasonable to assume that increased investment will automatically lead to oversupply.

Yes, as you know, considering the die-size penalty of HBM and relatively low productivity, even if investment increases, bit increase is only limited. So I think that these production increase restrictions will be aggravated as the HBM generation is upgraded.

As we know, given the die-size penalty of HBM and the level of clean room required, BIT growth is limited even as investment increases, and the limited production growth is expected to be exacerbated as HBM generations are upgraded. Furthermore, HBM is basing its investment decisions on at least a year's worth of customer contract volume, so increased investment in HBM means increased product orders.

The demand for HBM is rapidly increasing due to the deep competition in the AI industry, so the supply shortage is still continuing even with the expansion of the capacity of suppliers.

Demand for HBM is growing rapidly due to increased competition in the AI industry, resulting in a supply shortage even as suppliers expand their capacities.

The demand for HBM is growing rapidly due to increased competition in the AI industry, resulting in a supply shortage even as suppliers expand their capacities. Currently, demand is concentrated on AI servers due to the nascent AI market.

But as AI technology begins to be applied to various different applications, new types of memory products such as processor and memory PIM and wide IO and mobile are expected to emerge and generate new demand.

We expect the memory industry to evolve from a structure of mass production of a small number of products to a low-volume production of a wide variety of different products and to become an on-demand industry that provides long-term supply of the products that the customers want.

We will work closely with our customers to increase the visibility of demand and focus our investments on products with strong customer demand so that we can contribute to the stable growth of the market.

Next question, please.

The next question will be presented by Jae Kwon from JP Morgan.

Thank you for the question. I would like to ask you more specifically about our investment direction. You said that you are investing in sales cash this year. Could you please explain a little more about our company's investment direction this year and next year?

I'm Jae-gun from J.P. Morgan. Thank you very much for this opportunity to ask my question. I have a question about specific investment directions. You said in your presentation that you will be investing largely within the operating cash flow that is generated. So, can you elaborate more on what SK Hynix's investment direction for this year and next year is?

Yes, I will answer your question. Thank you for the question. So let me answer your question.

Thank you very much for that question. So our investment this year is higher than planned at the beginning of the year due to higher than expected demand for HBM products. and our decision to invest in fabs to secure clean rooms for the mid- to long-term.

Also, in 2025, not only HBM, but also general memory demand will increase. In order to do this, we need a considerable amount of investment in infrastructure such as M15X and Yongin Cluster.

In addition, in 2025, demand for not only HBM but also general memory is expected to increase further, requiring significant investment into infrastructure such as N15X and Yongin cluster to prepare for this. Therefore, we expect the scale of investment to increase compared to the historical average.

As a leading investor in the AI market, when considering the future demand growth, infrastructure investment is necessary for future growth-based acquisition. The long-term investment size of the company is reflected in this and is being produced, and the annual investment plan will be flexible by reflecting the market demand at that time.

Given our position as a leader in the AI market and expected growth in demand, infrastructure investments are essential to secure future growth. Our long-term investment plan reflects this, but we have the flexibility to execute our annual investment plan according to the market demand at the time.

Even if we exclude the increase in investment this year, our company's business cash flow is rapidly improving,

Even excluding this year's increased investment, our operating cash flow is improving rapidly, and we will continue to ensure that our financial strength is supported.

Next question.

The next question will be presented by Donghee Han from SK Securities.

Yes, hello. This is Han Dong-hee from SK Securities. The part I would like to ask is the KEPA operation strategy in HBM and general DRAM. Considering the demand for HBM and the demand for on-device AI next year, the supply of DRAM seems to be quite tight. I wonder what strategy you will use to distribute KEPA to HBM and general DRAM among our limited efforts. Thank you.

I'm Handong Yi from SK Securities. My question has to do with your strategy on managing the capacity for HBM and general DRAM. So there are many views that DRAM supply will be tight until next year as demand for HBM continues to be strong and demand for on-device AI starts to appear in earnest from next year. So compared to HBM, whose price is set on an annual basis, the price of general DRAM is adjusted quarterly according to supply and demand and profitability is improving rapidly. How does the company plan to manage the capacity between HBM and general DRAM to maximize DRAM profitability?

Yes, I will answer your question. As you said, due to the rapid growth of HBM demand, even though supply companies are increasing their operation rate, the additional capacity for the general DRAM production has been maintained at a significantly reduced level.

Let me answer your question. As you have mentioned, the available capacity for general DRAM production has remained at a reduced level since the production cutbacks, even though suppliers are increasing their utilization rates due to the rapidly growing demand for HBM. While overall capacity is expected to increase next year due to the increased industry investment, a significant portion of it will be utilized to ramp up production of HBMs, so the tight supply situation for conventional DRAM is likely to continue.

If the recovery of the demand for general DLM is accelerated here, the price of the general DLM, which is determined by the unit of return, cannot be excluded from the possibility of increasing compared to the HBM, which is determined by the price of the annual contract.

Furthermore, if the demand recovery for conventional DRAMs accelerates, it is possible that the profitability of conventional DRAMs, which are priced on a quarterly basis, may be higher than that of HBM, which is priced on an annual contract basis.

Unlike HBM's general DRAM, HBM has many new aspects. We will consider various aspects such as growth, stability, the status of the party within the market, and the relationship with the customer, and seek to maximize the DLM profitability from a long-term perspective rather than a long-term one.

As HBM has many new aspects that make it different from conventional DRAM, we will seek to maximize DRAM profitability over the long term rather than the short term, taking into account the growth and stability of HBM's revenue, our position in the market, and our relationship with the customers.

Next question, please.

The next question will be presented by Sungkyu Kim from Daiwa Securities.

Hello, thank you for the question. First of all, congratulations on your good performance. I have a question about HBM. In the case of HBM3 12-stage products, it is said that the difficulty of packaging is higher than 8-stage. Thank you very much for this opportunity to ask my question.

First of all, let me congratulate you on your excellent performance. I have a question on HBM. It is said that the packaging technology of the 12i product is said to be more difficult than the 8i product. So when do you expect to reach the material yield stage? And when do you think the HBM3e, 8i, and the 12i products will cross over each other? Those are my questions.

Yes, thank you for your question. As I mentioned earlier, HBM3e 12i Thank you for the question.

As I have noted previously, the HBM-3E-12HI samples were provided to customers in May, and we are currently preparing for mass production. And we plan to start mass production this quarter and start supplying customers in Q4.

Demand for the 12Hi product is expected to increase in earnest from next year, and we expect the supply of HBM3E 12Hi to exceed that of 8Hi in the first half of next year. Compared to the 8th stage, the technical difficulty of the 12th stage product has increased, but we have already mass-produced the HBM-3 12th stage, and the successful development of the 8th stage product and mass-production and wrap-up are being carried out, so we believe that we can shorten the time required to secure stable quality and demand.

Although it is true that the technical difficulty involved for the 12-high product is higher than the 8-high product, we believe that our experience in mass-producing the HPM3 12-high and the successful development and production ramp-up of the 8-high product will shorten the time required to ensure stable quality and yield.

Next question, please. The next question will be presented by Juni Lee from Goldman Sachs.

Thank you for the question. I also have a question related to HBM. The product release interval between HBM's main customers has been shortened from 2 years to 1 year, so it seems that HBM's product development period is also being shortened. I am curious if this change in the industry is advantageous or disadvantageous to the existing process companies, and I would like to know if you can predict Thank you very much for this opportunity to ask my question.

I also have a question related to the HBM. So the HBM's product development period seems to be shortening as major GPU customers have shortened their product release intervals from two years to one year. Do you think these industry changes are favorable or unfavorable for SK Hynix as HBM market leader? And does this change the mid- to long-term HBM TAM outlook?

Thank you for the question. The AI market is expanding much faster than expected, and as the growth rate of related industries increases, some customers are responding to the growth rate of the market as the release of new products is approaching.

Thank you for the question. As the AI market is expanding faster than expected and related industries are growing faster, some customers are responding to the market growth by accelerating their new product release cycles.

If the release cycle of the GPU is faster, the development cycle of the HVM used in the GPU is needed, but this can be a burden on the development of DLM companies.

Faster GPU release cycles require shorter development cycles for the HBMs employed in those GPUs, and this can be a burden for the DRAM companies.

However, it seems that it will be an advantageous environment for leaders who can respond to these customers' demands. In particular, due to the nature of the market where Time-to-Market is the most important market,

However, the environment is likely to be favorable for industry leaders who can respond to these customer demands, especially as time to market is of the essence, and it is important to minimize risk by partnering with industry leaders who have the right combination of technology competitiveness, mass production experience, and the scale.

We also believe that shorter time to market for new products will help to stimulate market demand

This is expected to expand the size of the AI market and positively impact the generation of demand for HBMs.

Next question, please.

The next question will be presented by Ricky Seo from HSBC.

Thank you for your question. I have a question about HBM4. It seems that there are two stages, stage 12 and stage 16. When do we expect stage 12 and stage 16 to be launched? There are various stories about whether or not to apply hybrid bonding. When can we apply hybrid bonding? Thank you for this opportunity to ask my question.

I also have a question about the HBM4. So in the HBM4 product range, there is the 12 high product and the 16 high product. And there is a lot of, so my first question is, when will the HBM4 12 high and the 16 high product Yes, thank you for the question.

HBM4 is expected to be shipped from the second half of next year, starting with the 12 high products, and we plan to apply the advanced MRMUF technology for mass production. The hybrid bonding technology that you have referred to is a technology that we are researching to reduce the height of the packaging by joining the pads of the copper wiring directly to each other

without micro bumps between the chips so that we can prepare for the ever-increasing number of stacks.

Currently, the packaging technology applied by each HBM supply company is different, and the capacity and resources of each company are all different, so the effect of hybrid bonding can be different for each company.

The effectiveness of applying the hybrid bonding technology may vary from company to company due to the different packaging technologies currently used by the HVM suppliers and their different capabilities and resources.

In order to apply hybrid bonding, the technology needs to be more advanced. Through cooperation with customers and partners, thorough quality verification at the system level,

Furthermore, the mass production application of hybrid bonding requires further technological advancement and thorough characterization and quality verification at the system level in collaboration with customers and partners.

The HBM-416 Hi product is expected to be in demand in 2026, and we plan to develop it in preparation for that demand.

And we are reviewing both the advanced MRMUF and the hybrid bonding methods to select the most optimal method to meet customer demands. Next question, please.

The next question will be presented by Sunwoo Kim from Merits Securities.

Hello, I'm Sunwoo Kim from Merits Securities. I would like to ask a question about NAND Bicross. In the first and second quarter, we are showing a sharp increase in the price. On the other hand, our exit volume seems to be a little weak. In the second quarter, the guidance seems to have been lowered a little, and in the third quarter, it has been guided by a small decrease in QOQ. It seems that this trend is suppressing the exit at the price priority, and please tell us how long this trend will continue. Thank you.

Hello, I'm from Merit Securities. Thank you very much for giving me this opportunity to ask my question. I have a question related to the bid growth of NAND. You have very strong performance for the first quarter and the second quarter. However, in comparison to that performance, your shipment level seems to be rather weak. The second quarter shipments have been below guidance, and for Q3 as well, it is expected to be down QQQ. So how long do you think this trend will continue going forward?

Yes, thank you for your question. Currently, except for enterprise SSDs that show a clear increase in demand, the demand for general applications such as PC and mobile is still showing a slight recovery rate. So, as the market situation improves by optimizing investment production and maintaining a profit-oriented sales strategy,

With the exception of enterprise SSDs where demand for NAND is clearly growing, general applications such as PC and mobile, demand is still showing a modest recovery, and we are responding to market conditions by maintaining a strategy of optimizing investments and improving profitability.

For example, last year, NAND invested at a minimum,

For instance, we are ramping up production on a limited base this year, which was reduced last year due to non-investment minimization and production capacity cuts. We're focusing this year on select applications where demand is growing.

While our quarterly NAND shipment growth may appear limited this year, we are committed to maintaining our market share by revenue

through our profitability-focused product mix and sales strategy based on actual demand?

In the third quarter, we will expand our sales focus on enterprise SD products, but considering the demand environment of some of the applications and the stock price of the customers, we are planning to increase the number of drop-offs in the entire NAND as I mentioned earlier, which is in the middle of the one-digit number compared to the previous quarter.

In the third quarter, while expanding sales for enterprise SSD products, we plan for a mid-single-digit shipment decrease in some applications due to low demand and customer inventory conditions.

However, in the fourth quarter, the production bit will increase on the inside, and some demand environments will be improved on the outside, and I expect that my overall output will be converted to growth tax. But in the fourth quarter, we expect shipments to return to growth as more bits are produced internally and the external demand environment improves.

Looking ahead to next year, we will continue to optimize investments and improve profitability, positioning the business to be profitable even in the face of future market fluctuations.

Next question, please. The next question will be presented by Seokhwan Shin from Daesin Securities.

Thank you for the question. I would like to ask a question related to the supply evaluation room. First of all, I would like to ask about the size of my usual return, which was reflected in the second quarter. In addition, if the price of the semiconductor continues to rise in the first half and the second half, please tell us about the possibility of additional returns. Thank you.

I'm Shin Seok-hwan from Daeshin Securities. Thank you very much for this opportunity to ask questions. My question has to do with the reversal of inventory valuation allowance. So what was the size of the reversal of the inventory valuation allowance in the second quarter? And if prices continue to rise in the second half of the year, is there a possibility of further reversals?

Yes, regarding the reversal, Let me take the question on inventory evaluation and allowance reversal.

So, our DRAM sales volume in the second quarter exceeded our guidance, and our finished goods inventory level at the end of the second quarter decreased from the end of the previous quarter as sales of DRAM and NAND continued to exceed production volumes.

The price rise of DRAM and NAND continued, and there was an increase of about 3 billion in the second quarter.

The DRAM and NAND prices continue to increase, resulting in a reversal of inventory valuation allowance of approximately 300 billion KRW in the second quarter.

In the second quarter of this year, it is expected that the market environment of DRAM and NAND will be superior, so there is a possibility of an additional return, but as much as the amount of VAT has already been returned to the upper part,

As the pricing environment for DRAM and NAN is expected to remain favorable going into the second half as well, there is the possibility of further reversals taking place. But as the majority of the inventory valuation allowance has been recognized during the downturn has been reversed, further reversals are expected to be minimal.

Next question, please.

The next question will be presented by Minsook Choi from Korea Investment and Securities. I would be grateful if you could share with us your plans to run HBM CAPA next year.

I'm Chaemin Sook from Korea Investment and Securities. Thank you very much for this opportunity. I have a question related to HBM. Given that HBM demand is growing rapidly and SK Hynix's leadership in the market is widely recognized, You seem to be taking a conservative stance when it comes to the capacity increase. So why are you not increasing HBN capacity more aggressively? And what are your plans for HBN capacity to meet demand next year?

Thank you for your question. Unlike some of us, AI demand continues to rise and grow.

Thank you very much for that question. Contrary to some concerns, AI-driven demand has continued to exceed expectations and is growing this year.

This year, we are planning to quickly expand the supply of HBM-3 based on increased TSB CAPA and 1B nano conversion investment.

We have previously announced that we will more than double our TSV capacities from last year's level to meet the growing demand this year, and we expect to rapidly expand the supply of HBM-3E based on the increased TSV capacities and 1B nanometer of conversion investment this year, resulting in HBM revenue growth of over 3%, 100% compared to last year.

That is our expectation.

We are very carefully crafting our investment plans and capacity-related plans by reflecting the demand in the downstream and also the supply chain situation.

Next year, most of the CEPA will be completed, and we expect more than twice the growth rate this year.

Most of next year's CAPA has already been largely agreed with the customers, and we expect shipment growth to be more than double compared to this year.

Next year, HBM-3 will be the main product, but we will also be preparing for HBM-4 to meet customer demand stably.

Next year, the HBM3E12 High product will be our flagship product, and we will also finalize preparations for the HBM4 to ensure a stable response to customer demand.

The last question, please. The last question will be presented by Youngho Ryu from NH Investment and Securities.

Hello, I am Ryong-ho from NSO. Thank you for the opportunity to ask a question. I would like to ask a question related to NAND. As you said before, the demand for enterprise SST is showing a strong trend. In fact, I think it is showing a strong trend compared to what we expected at the beginning of the year. I wonder if there is any difference in the level that we expected at the beginning of the year. How do you see the future of the ESS market and what strategy do you have?

Yes, thank you for the question. Last year,

Last year, the demand for general purpose servers slowed down significantly due to focus on AI-oriented servers within limited IT investment budgets.

And so as a result, enterprise SSD demand was weak.

This year, however, investment in general purpose servers is increasing compared to last year,

and AI-related demand, which was previously limited to DRAM, is spreading to NAND storage, and so ESSD demand is growing at a level that is significantly, significantly exceeding expectations at the beginning of the year, mainly for high-density products.

Because the power consumption of the AI server is large, the preference for low-power storage products will grow. In line with this growth rate,

Due to the high power consumption of AI servers, the preference for low-power storage products is expected to grow, and we plan to respond to this growth with a diverse lineup of enterprise SSD products to meet customer demand.

We are currently the only company in the industry to offer QLC-based enterprise SSD products over 60TB

and the quality of our products is highly competitive. We plan to maintain our high-density product leadership by introducing 128TB and later on 256TB products later this year.

Also, our Enterprise SSD product has excellent QoS competitiveness based on a high level of understanding of the data center workload. The quality of service. The response time of reading and writing data Our eSSD products also have very strong QoS advantage based on our deep understanding of data center workloads.

QoS stands for Quality of Service, which means that the response time to read and write data is short and reliable, which in turn means less time waiting for the SSD to respond, and this can increase the efficiency of the GPU.

In the future, the AI server will also be a very important factor in power efficiency. In that respect, high-capacity ESSD products with good performance and stability may be attractive to customers in terms of TCO, 향후 데이터 센터의 고용량 스토리지 시장에서의 성장은 더욱 기대되고 있습니다.

In the future, power efficiency will be a critical factor for AI servers, and high-density ESSD products with strong performance and reliability will be very attractive to customers in terms of TCO. So we're looking forward to further growth in the high-density storage market for data centers.

저희는 이러한 경쟁력을 바탕으로 올해 ESSD의 매출은 As I mentioned before, it will grow by 4 times over the previous year, and the sales of ESSD products will be about half of the entire NAND. We will continue to strengthen our position as the top two players in the enterprise SSD market with high growth.

Based on these competitive advantages, we're going to increase our sales volume to fourfold that of the previous level, and we will expand the proportion of ESSD sales to half that of NAND this year and strengthen our position as the top two player in the high-worth ESSD market.

With this, we would like to conclude the SK Hynix 2024 Q2 Earnings Release Conference Call.

Thank you very much.