Joining SK hynix presentation. There will be a Q&A session. If you have a question, please press star one. Please note that presentations will be interpreted simultaneously and Q&A session consecutively. With that, we are now ready to begin.
Good morning, afternoon, and evening. This is Park Seong-Hwan, Head of IR at SK hynix. Welcome to SK hynix 2024 first 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 that all outlooks presented by the company are subject to change depending on the macroeconomic and market circumstances. With that, we will now begin the SK hynix earnings release conference call for first quarter 2024.
Mr. Kim 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 first quarter of 2024. In the first quarter, while demand was relatively soft from PC and mobile, AI server continued to see strong demand driving improvement in memory markets. DRAM and NAND prices rose more than what we have expected in the beginning of the quarter, which makes us believe that memory market is clearly entering into a full recovery phase. As a result, our first quarter revenue reached KRW 12.4 trillion, an increase of 10% sequentially and 144% year-on-year. DRAM bit shipments decreased by 18% sequentially, which is in line with our guidance, as product sales increased for AI servers while conventional DRAM sales saw seasonal weakness.
ASP increased by over 20% compared to the previous quarter, with prices rising across all product lines for two consecutive quarters. While NAND also was impacted by seasonality, bit shipment was maintained at a similar level compared to the previous quarter, with expanding sales of eSSD that is improving in demand. ASP increased by over 30% sequentially, driven by significant price increases across all products. Operating profit in the first quarter was KRW 2.89 trillion, a KRW 2.54 trillion improvement from the previous quarter, with operating margin of 23%. The performance was driven by increased sales from rapid price improvement, sales strategy focusing on profitability, ongoing cost-cutting efforts, and reversal effects of previously recorded inventory valuation losses. Particularly notable was NAND profits, which turned into profits thanks to higher mix of premium eSSD products as well as steep ASP increases since fourth quarter last year.
Following DRAM's turnaround in the third quarter of last year, we believe that this quarter's NAND turnaround marks the beginning of a full-fledged growth phase of our performance. Depreciation and amortization expenses for the first quarter was KRW 3.19 trillion, slightly lower compared to the previous quarter due to reduced investment in the previous year. EBITDA was KRW 6.07 trillion, and the EBITDA margin was 49%. Non-operating loss net of gain in the first quarter was KRW 0.51 trillion. This includes net interest expense of KRW 0.32 trillion and net foreign currency-related loss of KRW 0.22 trillion, including translation loss of foreign currency-denominated debts due to the rise in U.S. dollar. As a result, our net pretax loss was KRW 2.37 trillion, net profit was KRW 1.92 trillion, and the net profit margin was 15%.
Consolidated cash balance at the end of first quarter was KRW 10.3 trillion, up by KRW 1.4 trillion from year-end last year. Interest-bearing debt was KRW 29.5 trillion, remaining at a similar level to that of previous quarter. Debt-to-equity ratio and net debt-to-equity ratio at the end of first quarter was 53% and 35% respectively, both improving from previous quarter's levels. Next, I will discuss market outlook and our company plans. The memory market is now believed to enter a full recovery cycle with improving profitability in the industry due to strong AI demand and a favorable supply-demand environment. In the second half of the year, demand from traditional applications such as PCs, mobile devices, and general servers is expected to improve, leading to stable growth in memory demand.
Meanwhile, despite the gradual utilization recovery in the industry, prioritization of premium products like HBM will lead to a production limitation on general DRAM products, which will eventually accelerate inventory depletion across the industry once the demand for conventional market improves. Favorable pricing environment is expected to continue throughout the year, and thus the memory market in 2024 is projected to reach the revenue size comparable to that of past peak cycles. Looking at specific applications, the PC market was somewhat soft in the first half of the year, but demand is expected to recover toward the second half with the end of Windows 10 support and the introduction of AI PCs, propelling replacement demand especially from enterprises. Additionally, the need for high-performance, high-density memory support for Windows upgrades and AI PCs is expected to drive continued growth in memory contents.
The smartphone market exhibited a slower demand recovery than expected, with only some new flagship products with AI capabilities seeing growth. However, the introduction of new AI features along with upcoming product launches in the second half of the year is expected to stimulate consumer replacement demand, driving overall unit and content growth of smartphones. For the server market, the strong demand for AI servers is expected to continue as generative AI is advancing from generating text-based responses to generating images and videos, and the focus of AI technology is shifting from training to inferencing. Furthermore, gradual replacement demand is expected to arise for servers in cloud data centers that were heavily invested in during 2017 and 2018, considering the depreciable time for servers. In NAND, which has seen relatively weaker impacts from AI demand, demand recently started growing for high-density eSSDs.
The advantages of NAND as a storage solution, such as faster data transfer speeds, lower power consumption, and the ability to store large capacities in small physical spaces, are gaining traction in the AI market. As a result, demand for high-performance, high-density NAND is increasing, and we anticipate further adoption of high-density ESSDs in AI servers and data centers going forward. In the second quarter, DRAM bit shipment is expected to grow by 18% sequentially, driven by the higher sales of HBM3E products. For NAND, while ESSD sales are expected to increase due to strong demand, overall company's NAND bit growth is expected to be flat sequentially, as we are planning to prepare for more advanced NAND demand improvement. In March, SK hynix started mass production and supply of HBM3E using the leading-edge 1B nanometer process, maintaining our competitive edge as a leading player in the HBM market.
Supply of HBM3E will increase according to customer demand this year, and we also plan to expand our customer base by leveraging the increased production capability compared to that of last year. SK hynix has signed an MOU with TSMC for the development of the next-generation HBM product, HBM4, and cooperation in next-generation packaging technology. We will utilize TSMC's leading logic process to produce the base die of HBM4, which is scheduled to start mass production in 2026 and supply customized HBM products that meet a wide range of customers' requirements, including performance and power efficiency. We will also cooperate to optimize the combination of our HBM and TSMC's CoWoS technology to strengthen our position as a total AI memory provider. In addition, our DDR5 products have been able to capture the early market growth backed by our stable technology and quality since the first-generation product.
Already, over 45% of our PC bit sales and over 60% of our server bit sales are from DDR5 products. Our timely support of 128 GB and higher density modules have also contributed to our strong DDR5 sales. In addition to our product lineup, we are also planning to introduce 32 Gb DDR5 products based on 1B nm process in the near future to support demand for high-density server DRAM. For NAND, which successfully turned into profit in the first quarter, we plan to focus on optimizing product mix and lineup to sustain this growth. Along with high-performance 16-channel eSSD that has contributed meaningfully to our Q1 sales, we plan to support growing demand for high-density eSSDs with industry-leading QLC-based ultra-high-density eSSD products provided by Solidigm. Going forward, we are also planning to proactively support demand for on-device AI solutions by launching PCIe Gen 5 CSSD for PCs this year.
Utilization rate is gradually recovering in the industry as demand is improving, driven by AI memory. However, as demand is focused on advanced node products, upgrade investments are essential in order to increase wafer production. This year, DRAM and NAND production growth is expected to be constrained as a result of conservative investments of last year, as well as higher capacity allocation to HBM, which has meaningfully larger die size compared to that of conventional DRAM. As a move to proactively support growing AI memory demand as well as conventional DRAM demand, we decided to invest in M15X as a new DRAM production facility with the target to open by end of 2025. The progress of establishing Yongin Semiconductor Cluster is also on track. M15X in Cheongju and the new facilities in Yongin, Korea, will be the foundation of the company's mid to long-term growth.
Furthermore, we have decided to construct an advanced packaging production facility for AI memory in West Lafayette, Indiana, USA, in order to strengthen our leadership in AI semiconductor technology and customer collaboration. We will also collaborate with local institutions for research and development. The establishment of the Indiana facility will involve an investment of about $3.87 billion, and we plan to mass produce next-generation AI memory products, including HBM, from 2028. Through these efforts, we will be able to supply a variety of customized memory products that can meet the increasingly complex needs and expectations of customers and take a leading role in strengthening the global AI semiconductor supply chain. Due to rising needs to meet the rapidly growing demand for HBM together with new investment decisions for M15X, the company's CapEx this year is expected to be somewhat higher than the amount that was planned initially.
However, these investment decisions were made to meet the elevated levels of customers' demand to support not just the HBM products but potentially for conventional DRAM. Therefore, we try to contribute to stable and sustainable growth of the memory market. We will continue to make prudent investment decisions focusing on investment efficiency and financial soundness. Next, let me share our ESG management activities and performance. In February, we signed a renewable power purchase agreement for a 100-MW renewable energy product based on solar power to comply with RE100. Through this agreement, we aim to expand the reduction of Scope 2 greenhouse gas emissions and diversify our domestic renewable energy sources, which have been heavily focused on green premiums. We plan to continue collaborating with key players in the renewable energy ecosystem to secure competitive renewable energy sources in the future.
Additionally, for the first time for a semiconductor company, we announced a mid to long-term roadmap to utilize recycled materials in production. We aim to raise the weight proportion of recycled materials in our products to over 25% by 2025 and over 30% by 2030. To achieve this, we have developed strategies to increase the recycled portion of metals such as copper and tin, as well as plastic used in packaging to protect finished semiconductor products. We will strengthen the certification procedures and quality evaluation for recycled materials directly purchased by the company and also encourage our suppliers to join the efforts to obtain validation from publicly trusted external organizations to verify the use and proportion of recycled materials. We will continue our efforts in carbon reduction and various eco-friendly activities based on the company's ESG initiatives to contribute to the transition to a carbon-neutral society. Thank you.
With that, we are now ready to take your questions.
[Foreign language]
Now, Q&A session will begin. Please press star 1, that is star one, if you have any questions. Questions will be taken according to the order you pressed star one. For cancellation, please press star two, that is star two, on your phone.
[Foreign language]
The first question will be presented by Dongwon Kim from KB Securities. Please go ahead with your question.
[Foreign language]
Thank you very much for taking my questions. I have two questions, one each for NAND and DRAM. The first question is about the NAND market. We see that the NAND market recently has been seeing increased demand for high-capacity storage SSD led by enterprise SSD. What does the company believe are the main reasons for the higher demand for high-density SSD, and what is the strategic direction for NAND products? And the second question is that all memory companies recently are focusing on HBM capacity expansion, which will inevitably affect existing memory capacity. What is the likelihood of non-HBM shortage occurring in the second half of the year?
[Foreign language]
Regarding the first question about the demand for the high-density eSSD, it appears that the demand spike for the high-density eSSD appears to be primarily driven by the demand coming from on-prem AI data center customers as the AI market continues to expand.
[Foreign language]
To be more specific, we see that AI technology is shifting from training more to inference, with many individual companies now tapping into AI and seeking to build on-premise AI servers for reasons of technological security and customization.
[Foreign language]
This has led to growing demand for high-density NAND solutions that are faster than existing storage solutions as they need to process heavy workload, which is characteristic to unstructured data. They also consume less power, which offers better TCO attractiveness.
[Foreign language]
This is a positive development for NAND vendors as it is new demand created by the expanding AI market and may be a sign of structural change.
[Foreign language]
For the longer term, the growth of AI and its use by individual enterprises appears to be leading to real demand for high-performance, low-power storage solutions that highlight the advantages of NAND storage. I s also a customer need to solve data center space constraints, and accordingly, demand is increasing for 30 TB, even 60-128 TB high-capacity enterprise SSDs that greatly exceed the capacity of existing enterprise SSD products. In addition to such high-speed and power savings, there are also customer demands to address data center space constraints, which is driving demand for much, much higher density eSSD than existing products for 30TB and even 60-128TB.
[Foreign language]
Such ultra-high-density eSSD deployment requires QLC-based products rather than what is the mainstream today or the TLC-based. So we plan to address the upside demand with a 60 TB or higher density eSSD solution based on QLC, which is unique to Solidigm.
[Foreign language]
Just as we have demonstrated our technology and competitiveness in HBM and high-density DRAM products, we will do the same in ultra-high-density eSSD, working with our customers to strengthen our position as a memory leader.
[Foreign language]
Moving on to the second question about the DRAM supply and demand in the second half of the year.
[Foreign language]
While strong AI-driven demand continues this year, demand for existing applications is also expected to improve overall, year over year. But it is likely to be concentrated in the second half of the year.
[Foreign language]
At the same time, on the supply side, suppliers are gradually normalizing utilization but are prioritizing capacity for HBM, for which demand is surging and visibility is high.
[Foreign language]
But for the HBM, it requires more wafer capacity than regular DRAM because its die size is roughly double that of regular DRAM. Thus, the wafer capacity for regular DRAM this year is expected to be limited.
[Foreign language]
If demand improves in conventional applications like PC, smartphone, and general servers in the second half of the year, customers and memory vendors could sell down their inventory. And if demand exceeds expectations, there may be supply shortage for these products.
[Foreign language]
The company seems to have a bit insufficient clean room space. The M15X that you announced yesterday, and the Yongin fab mass production time point, and what is the production scale to some extent? It would be appreciated if you could let me know. And you said that the product to be produced at the Indiana back-end facility is the next-generation HBM. Is there also a possibility of linkage with the US factory that TSMC is currently building? It would be appreciated if you could tell me about such mid-to-long-term plans. That's all.
[Foreign language]
The next question will be presented by Sung Kweon Kim from Daiwa Securities. Please go ahead with your question.
First of all, congratulations on the good performance. I also have two questions. The first is, this was briefly mentioned earlier, but about the inventory level. So what is the current inventory level of customers and the company by application? And when does the company believe would be the timeline for normalization of the inventory level? And the second question is, now we see that the demand recently is rapidly growing, but compared to this, there seems to be not enough clean room space. So this was also announced yesterday about the M15X, but what is the timeline and scale of mass production at M15X and Yongin Fab? And also, during the presentation, it was mentioned that the Indiana Advanced Packaging Facility will be producing next-generation HBMs. And is there any possibility of connecting with the TSMC's US plant?
[Foreign language]
First of all, about the inventory level. The memory inventory levels that have been built up during the downturns of 2022 and 2023 have been shrinking since the second half of 2023 due to aggressive production cuts by suppliers.
[Foreign language]
In addition, supply of general or the conventional memory products will be limited this year as we respond to increased demand for AI-related products such as HBM. As a result, customers are believed to be strategically managing their inventory in light of current memory inventory level and expected demand improvement in the second half of the year.
[Foreign language]
In Q1, we responded to real demand amidst the weak seasonality. As a result, customer inventory at the end of Q1 appears to be little changed from the previous quarter. But inventory might have slightly fallen for ESSD products as demand began to pick up.
[Foreign language]
Customers' inventory reduction is likely to be limited in the first half of the year due to continued memory price increase and demand for inventory buildup in preparation against shrinking supply of legacy products. But customer build demand, which is the demand for parts for finished products, the customer build demand will improve moving into the second half of the year, which is likely to normalize industry-wide inventory level.
[Foreign language]
Our finished memory product inventory at the end of Q1 was down for both DRAM and NAND as sales continued to outpace production despite conservative sales in Q1.
[Foreign language]
Given that there will be production ramp-up mainly for leading-edge technology products this year, inventory of legacy products, which currently account for the bulk of inventory, will start decreasing at a faster pace in the second half of the year, likely to fall to a tight level by year-end.
[Foreign language]
Thank you for your question, and I would like to respond to the second part of your question.
[Foreign language]
Now, the company decided that additional clean room space was necessary to keep up with the rapidly growing demand for AI memory and conventional DRAM as well.
[Foreign language]
Thus, the decision was made to invest in M15X to maintain our current position in the AI memory market and to proactively address the growing demand for DRAM.
[Foreign language]
We believe that we can secure a clean room at M15X relatively quickly because we can utilize the existing infrastructure at the Cheongju site. If construction begins now, the company could potentially open the fab by the end of 2025.
[Foreign language]
Also, the M15X has the advantage of being adjacent to M15, where we are expanding the TSV capacity, allowing the M15X to optimize production of HBM.
[Foreign language]
At Yongin, site preparation work is currently underway and is on track to open the first fab in 2027.
base, and was decided to be built on the U.S. mainland where major customers in the AI field are concentrated and advanced post-process field technology research is also actively progressing. We are targeting production operation in the second half of 2028.
About the Indiana Advanced Packaging Facility that will be built in the mainland U.S., where the majority of AI clients are located, and when there is robust R&D on advanced packaging technology, the purpose is to strengthen our AI semiconductor technology leadership and customer collaboration base, and it is targeted to start production in the second half of 2028.
[Foreign language]
As demand continues to grow for ultra-high performance memory like HBM on the back of AI-driven growth and the importance of advanced packaging growth, the company will continue to strengthen collaboration with not only key customers but also with the various partners across the supply chain to maintain our differentiated competitiveness in the next-generation customized HBM market so as to defend our leadership in the AI memory solution market.
[Foreign language]
Now, to make sure that the company responds to the changing demand in a timely manner, we will be basing the decisions on our investment timeline size as well as the mass production scale and speed on our demand forecast.
[Foreign language]
We will take the next question.
[Foreign language]
The next question will be presented by Subin Lee from CGS-CIMB Securities. Please go ahead with your question.
[Foreign language]
Also, congratulations on the good performance. I also have two questions. Now, first is that the company earlier mentioned that the supply of HBM3E will be expanded. So is it going to be mainly for the 8-layer product? What is the timeline for qualification and mass production of the 12-layer product? And negotiations for HBM3E volume and price are presumably underway this year. And can the high margin that we saw in HBM3 be sustained in HBM3E in the face of increasing competition? And the second question is about the capacity. Now, what is the company's plan for the HBM capacity expansion in 2025? Your competitor has already mentioned that they have received orders for most of their capacity in their last earnings call. I wonder whether the company is in a similar situation.
Is there a possibility that the HBM supply is going to outpace demand after 2025 because of the competitive capacity expansion?
[Foreign language]
Thank you for the questions. This is my response to the first question about the HBM3E.
[Foreign language]
The customer demand for the HBM3E this year is primarily for the 8-layer. We are getting ready to complete the 12-layer HBM3E development in the third quarter of this year and go through customer qualification afterwards to meet the customer's required timeline. We will then be ready to ensure reliable supply next year in time for growing demand.
[Foreign language]
For the HBM3E price, it reflects a premium over HBM3 due to higher performance, higher density, and increased cost.
[Foreign language]
Ramp-up for mass production of HBM3E is also underway as planned, thanks to our industry-leading EUV productivity and maturity of the 1B nanometer technology that is breaking records all around.
[Foreign language]
Overall, the progress so far indicates that we could reach the level of yield comparable to HBM3 in the near future, which will also help quickly stabilize the cost.
[Foreign language]
Now, in terms of profitability, HBM3E is expected to be a more competitive product in terms of profitability, thanks to the company-wide commitment to build up production capacity with proactive investment and undertake productivity enhancement such as yield and quality improvement.
[Foreign language]
Now, let me move on to the second part of your question, which was on the plan for HBM capacity expansion in 2025.
[Foreign language]
Now, it was explained in our earnings call last October that we already sold out on this year's increased HBM capacity as the decision itself was made in consultation with our customers.
[Foreign language]
We had also revealed that we are in discussion with various AI players and potential customers to expand our business areas for the mid to long term.
[Foreign language]
What we are seeing recently is that there is investment increase by CSPs in AI servers, and also there is additional demand to improve AI service quality. We see that they are fast driving HBM demand, and as a result, demand visibility in HBM is becoming clearer than it was just half a year ago.
[Foreign language]
Regarding the concerns that the HBM market may face oversupply as vendors expand their capacity, well, the HBM market is set to keep rapidly growing in 2024 and beyond as data and model sizes grow for reasons such as increasing parameters and modalities and also the growing use cases among end users. There is need for higher AI performance and more AI service providers also appearing.
[Foreign language]
Based on our product competitiveness and large-scale mass production experience, we are now in discussion about long-term projects through 2025 and beyond with quite a number of existing and potential customers.
[Foreign language]
About the capacity for 2025, it is currently under discussion with our customers, taking into account the lead time for equipment, and we will be ready to grow with our customers in line with the AI market growth through timely investment and capacity expansion.
[Foreign language]
[Foreign language]
The next question will be presented by Sunwoo Kim from Meritz Securities. Please go ahead with your question.
[Foreign language]
[Foreign language]
Thank you very much for taking my questions. As a follow-up to the previous question, then I have a question about HBM, where the company is currently displaying overwhelming performance, and another question regarding the recent financial performance. Now, the first question about the HBM. There have been press reports that the next-generation HBM configurations would be relaxed. So if this is true, then how will this affect the company's future technology roadmap? And also, can the company provide us with an update on the hybrid bonding development? What is the timeline for the application of this technology? And the second question is, we see that the financial performance in Q1 was outstanding. And what was the size of the reversal of inventory valuation loss in Q1? And also, what is the current outstanding balance in the valuation provision?
Also, it is true that we have seen rapid price increase for NAND in the previous quarter, and it appears as if that there will be further rounds of significant increase in NAND prices in Q2 as well. So then following that, what would be the likelihood of further reversal in the inventory valuation loss? And also, even without such one-off factors as the reversals, is it still possible to maintain main profitability? So I do realize that these are a lot of questions, but I ask for understanding.
[Foreign language]
Thank you very much for your questions. Now, allow me to first respond to your question about the next-generation HBM packaging technology.
[Foreign language]
Now, if the packaging height configuration for HBM is relaxed, then we do expect the adoption of hybrid bonding technology to be pushed back. Given the complexity of the technology, there may be some productivity and quality issues, especially in the early days of adoption. For the sake of reliable supply and cost of HBM, it would be better to wait until the technology reaches sufficient maturity.
[Foreign language]
The company will continue to apply the proven and competitive Advanced MR-MUF process to our 16-stack HBM products, which will allow uninterrupted supply of efficient and competitive products.
[Foreign language]
Having said that, it is also true that the hybrid bonding is believed to be an important packaging technology to enable high-density, high-stack HBM products. The company is committed to proactively acquiring this technology to maintain our HBM market leadership.
[Foreign language]
Again, the hybrid bonding technology is highly complex, requiring control of the flatness of the bonding layer and the bonding strength. Particle control also needs to be done at the nanometer level. To ensure robust quality, we are conducting collaborative research in various fields such as metrology and inspection.
[Foreign language]
Once again, thank you very much for your questions. Allow me to respond to the question about the reversal of inventory valuation loss and the NAND profitability.
[Foreign language]
Now, this was also mentioned in the previous quarter, but as was the case in the first quarter, as was the case in the previous quarter, there was reversal of inventory valuation provision mainly for NAND products following a sharp rise in ASP. The total reversal was approximately KRW 900 billion, which is slightly higher than the previous quarter.
[Foreign language]
Looking ahead, as memory ASP continues to rise and inventories are expected to decline, we expect further reversals to be recognized, but to a gradually lesser extent.
[Foreign language]
In the first quarter, NAND returned to profitability amidst soft demand driven by an improved product mix led by eSSD products, faster than expected price rise, and a reversal from inventory valuation loss.
[Foreign language]
In the second quarter, on anticipation of the favorable pricing environment and strong demand growth for our competitive high-density eSSD products to continue, we expect profitability to continue, even excluding such one-time factors as reversals from inventory valuation loss. In particular, we see that performance from Solidigm is fast improving. Also, this is driving further growth in the revenue for the high-density eSSD. So we believe that overall, the NAND performance is going to continue to improve.
[Foreign language]
[Foreign language]
The next question will be presented by Nicolas Gaudois from UBS. Please go ahead with your question.
Good morning, and thanks for taking my questions. First one is, could you clarify when you expect to return to full utilization rates from a wafer size perspective for both DRAM and NAND? And would that be with wafer capacity broadly comparable to those of 2H22 or, in fact, lower due to process tech migration? And second question is, there have been some discussions about floating gate technology that you are running in Dalian Fab, reaching its limits at 144 layers, which you currently have. What are the options in that case going forward? And have you started moving Solidigm SSD products to your original charge trap technology? Thank you.
[Foreign language]
First of all, thank you very much, Mr. Gaudois, for your questions.
[Foreign language]
I will respond to the first part of your question.
[Foreign language]
Over the past two years, all memory providers, I would say, have undergone bold investment cuts and aggressive capacity cuts in response to the downturn. They are now gradually restoring utilization to meet the growing demand for new products.
[Foreign language]
As you would know, DRAM and NAND are different in terms of the demand and supply dynamics coming from AI, which warrant a different approach to utilization rate as well.
[Foreign language]
First, for DRAM, we are working to expand supply mainly with the high-demand HBM, which is showing strong demand. But due to the HBM's large chip size, the increase in finished product production is highly limited despite the increase in wafer input.
[Foreign language]
The expectation now is that utilization rate will recover at a slightly faster pace than planned in the second half of the year when improved demand for conventional DRAM products becomes more apparent.
[Foreign language]
However, as you have rightly pointed out, even if we do go back to 100% utilization, wafer production capacity is bound to decrease as we migrate to higher-level technology. This means that capacity in the DRAM industry is not likely to reach the historical peak by the end of this year.
[Foreign language]
For NAND, it is also expected to benefit from AI. We have yet to see meaningful improvement in demand for general applications. Given that it does not have the same supply constraint as HBM, we will be more cautious in deciding on utilization recovery.
[Foreign language]
For the company, we are focusing on products with clearly improving demand, like high-capacity eSSDs, in our decisions for increasing utilization and production. This hopefully will gradually bring down the production cut-induced cost burden at the fabs making the high-demand products.
like high-capacity enterprise SSDs, we are gradually recovering the utilization rate. While continuing mass production of the 144-layer and 192-layer products at Dalian Fab, we plan to respond to the upside in demand for high-capacity enterprise SSDs.
As was just mentioned, we also intend to respond to the rapidly growing demand for high-density ESSDs from the Solidigm side. We are also gradually restoring utilization for products where demand is improving, like the high-density ESSDs, which has led to increased utilization for the 144 layers.
[Foreign language]
From a long-term perspective, the utilization plan for the Dalian Fab is to be determined based on a comprehensive set of factors such as eSSD demand, geopolitical situation, and fab space operation plan at the headquarters.
the acquisition of Solidigm, we have been continuously preparing enterprise SSD products that combine SK hynix's unique NAND technology and then Solidigm's enterprise SSD solution capabilities. In the future, Solidigm is expected to secure the capability to flexibly respond according to supply and demand and according to customer demand by supplying not only floating gate-based enterprise SSD but also charge trap-based enterprise SSD.
Per your question on whether SK hynix CTF NAND is going to be used for Solidigm SSD products, what we can tell you is that since the acquisition, we have been preparing enterprise SSD products that combine SK hynix's unique NAND technology with Solidigm's eSSD solution capabilities. So Solidigm is expected to supply charge trap-based eSSD in addition to floating gate-based eSSD to flexibly respond to customer demand.
[Foreign language]
[Foreign language]
The last question will be presented by Dong Hee Han from SK Securities. Please go ahead with your question.
[Foreign language]
Thank you for taking my questions. I have three. One on finance, another on investment, and the third one is on HBM. So then it was mentioned that the CapEx will increase this year compared to the plan at the start of the year. So what is the expected increment for the year? And also it appears as if the free cash flow is expected to increase significantly this year despite the increased investment due to higher-than-expected price growth. So then what does the company believe will be the extent that you will be able to reduce borrowings as a result of this? And another question is about the custom HBM. So why is there such a need for custom HBM, and what is the company's product strategy going forward?
[Foreign language]
Thank you very much for your questions. This is my response to your first question about finance.
[Foreign language]
Now, to ensure that we flexibly respond to rapidly changing markets, we hold regular meetings regarding the market situation to review our production and investment plans.
[Foreign language]
More recently, we continue to revisit the size of our investment in light of the improved demand for HBM since the beginning of the year and decided to make additional Fab investment to ensure flexibility in our medium-term utilization.
[Foreign language]
And as a result, as was explained during the presentation, investment in 2024 will be slightly higher than planned at the start of the year. But the increment investment will be mostly for expanding production of high-margin products with clear demand visibility and for securing infrastructure for the medium term, meaning that any impact on the near-term supply and demand of conventional products will be limited.
[Foreign language]
We will prioritize our investment decisions around our competitive products, focusing on essential investment needed to maintain market leadership and respond to new opportunities. We will also balance investing in the future, investing in future growth with ensuring financial soundness given our current level of cash generation.
[Foreign language]
Thank you. Allow me to respond to the second part of your question about custom HBM.
[Foreign language]
Now, the needs for custom HBM are mainly driven by customer demand to utilize optimum HBM to maximize product utility in AI systems that demand very high performance.
[Foreign language]
Custom HBM is characterized by differentiation of the base die. The base die connects to the GPU or ASIC at the bottom of the HBM package to control the HBM. We plan to supply differentiated HBM by adapting the base die and packaging technology to meet a wide range of customer needs.
[Foreign language]
To this end, we recently announced that we have signed an MOU with TSMC, which has state-of-the-art foundry process capabilities, to develop HBM4 and collaborate on next-generation packaging technologies. We believe that this will help enable us to deliver a range of custom HBM products.
[Foreign language]
In addition to preparing differentiated products, we also plan to optimize some of our business operations as order-based supply becomes more prevalent, to position ourselves as a total AI memory provider that responds effectively to customer needs.
[Foreign language]
Thank you very much. With that, we conclude the earnings conference call for the first quarter of 2024. Thank you very much.