The first half of the year is now in full swing.

Reflecting on the semiconductor market over the past six months, it appears relatively serene on the surface, yet a closer look reveals numerous adjustments made by leading chip companies.

They have decisively "sliced away" and exited certain product areas.

What implications does this process hold for the semiconductor industry? And which companies stand to gain?

01

Chip Titans Abandoning These Chips

Three Storage Giants Plan to Cease DDR3 and DDR4 Production

News about major DRAM manufacturers such as Samsung, SK Hynix, and Micron planning to halt production of DDR3 and DDR4 memory has long been circulating among insiders.

In February of this year, this news was officially confirmed, driven by increasing demand for advanced memory technologies like HBM (High-Bandwidth Memory) and DDR5, alongside declining profitability for DDR3 and DDR4 memory.

It is anticipated that starting in summer 2025, the market supply of DDR3 and DDR4 will significantly decrease due to the exit of these major manufacturers, potentially leading to actual supply shortages.

In fact, Samsung ceased DDR3 production in the second quarter of 2024 and has continued to reduce DDR4 capacity. SK Hynix's DDR4 proportion also steadily declined last year, falling below 20% in the fourth quarter, with major manufacturers gradually shifting their focus to DDR5 and HBM.

Notably, these storage manufacturers have previously announced the launch of NOR Flash. With the continued reduction in DDR3 and DDR4 production, the competitive landscape in the niche storage market has significantly improved, offering Chinese Taiwan and mainland China storage manufacturers an opportunity to seize market share vacated by these manufacturers' production cuts and exits.

Currently, domestic chip manufacturers providing DDR3 and DDR4 include GigaDevice, Beijing Actions, Dongxin Semiconductor, and others. According to Huayuan Securities research reports, in 2023, GigaDevice's DDR3 4Gb and 2Gb products achieved mass production, and in 2024, they contributed to revenue through bulk shipments. Additionally, the company's DDR4 8Gb product successfully taped out and provided samples to customers in 2024.

Beijing Actions' niche DRAM portfolio includes multiple types such as DDR, DDR2, LPDDR2, DDR3, DDR4, LPDDR4, with capacities ranging from 16M, 32M, 64M, 128M to 1G, 2G, 4G, 8G, 16G, and other specifications.

Dongxin Semiconductor's DDR3(L), LPDDR1, LPDDR2, and other products have achieved mass production and continue to develop new products. Among them, the company's LPDDR4x product has entered mass production.

Samsung Plans to Exit the MLC NAND Business

On May 26, Korean media outlet TheElec reported that Samsung Electronics informed customers that MLC NAND flash memory production would soon cease, planning to accept the final MLC chip orders next month.

The report stated that while notifying customers of the final MLC NAND production schedule, Samsung Electronics also informed some customers of plans to raise MLC prices, prompting them to start seeking new alternative suppliers.

It is said that LG Display is also looking for alternative MLC NAND suppliers to Samsung, primarily including 4GB eMMC for large-size OLED panels.

Prior to this, LGD had been using storage solutions from Samsung, Kioxia, and ESMT, with ESMT's eMMC using Samsung MLC NAND for packaging, and Kioxia using its own MLC NAND for supply. This means that if Samsung Electronics ceases MLC NAND production, LGD will only have Kioxia as an existing supplier. LG Display is therefore seeking more suppliers.

Regarding the reason behind Samsung Electronics' plan to exit the MLC NAND market:

MLC NAND accounts for less than 1% of Samsung's overall revenue, and the global NAND market has shifted towards TLC and QLC technologies, with TLC accounting for up to 62% of the market share. At the same time, Samsung is promoting the mass production of new NAND products such as 176-layer, 238-layer, and 286-layer, and the demand for old production line upgrades and technology migration further compresses MLC capacity.

Samsung is shifting MLC-related capacity to focus on the automotive electronics field, recognizing that automotive electronics' requirements for storage chip reliability fall between consumer-grade and industrial-grade, with MLC still offering a competitive cost-performance ratio. Through capacity transfer, Samsung can tap new profit growth points in emerging fields and optimize its business structure.

Against this backdrop, the aforementioned domestic storage chip companies are expected to welcome greater development opportunities.

02

These Titans, Splitting Businesses

Western Digital Exits SSDs, Focusing on HDDs

Western Digital has also made bold strategic adjustments.

In March of this year, Western Digital announced that it had officially spun off its NAND flash business and would no longer produce NAND and SSDs.

Moving forward, Western Digital will concentrate on its mechanical hard drive market, with the original NAND and SSD businesses taken over by SanDisk.

This decision is driven by a combination of factors:

First, with the consumer-grade SSD market becoming increasingly competitive, technology leaders such as Samsung and Kioxia dominate, leading to continued declines in profitability due to product homogeneity and price wars. Western Digital's decision is to concentrate resources on "tracks with more growth potential".

Second, AI is driving an explosion in demand for data centers and cloud computing, making enterprise-class high-capacity HDDs (such as products above 20TB) the "new blue ocean" in the storage market, necessitating adjustments in Western Digital's development direction.

Third, by spinning off the SSD business to its subsidiary SanDisk, it achieves "specialization" - SanDisk focuses on flash memory technology, while Western Digital delves into HDD innovation. This may help both companies better leverage their respective strengths.

Western Digital's business adjustment differs from the aforementioned companies' decisions. The company divested part of its business and transferred it to SanDisk, meaning that the relevant market share did not become vacant. The above-mentioned companies have released some market space, allowing other companies to expand their businesses.

It is evident that the storage market is bustling, with many of the aforementioned companies belonging to the storage chip sector. Among them, SK Hynix's adjustments do not stop here.

SK Hynix Cuts CIS Chip Business

Previously, SK Hynix had two major product lines: memory semiconductors, primarily represented by DRAM and NAND Flash; and system semiconductors, represented by CIS.

Although CIS is one of SK Hynix's two major product lines alongside memory semiconductors, its profitability is far inferior.

In 2023, SK Hynix's share in the CIS market was only 4%, with revenue of approximately $870 million. In contrast, the mobile CIS market is dominated by Sony, Samsung, and OmniVision, with Sony holding a market share of up to 55%, followed by Samsung with 25%, and OmniVision ranking third with a market share of 13%. In the automotive CIS market, Onsemi, OmniVision, and ST rank in the top three. Additionally, the continued downturn in the smartphone market has severely impacted the profitability of the CIS business.

Due to changes in the market environment and increasing competitive pressure, SK Hynix's CIS business has poor profitability. The company has gradually reduced CIS manufacturing wafer starts and cut R&D expenditure.

In 2024, SK Hynix reduced CIS R&D investment, with expected monthly capacity below 7,000 12-inch wafers, more than halving from 2023.

In March of this year, SK Hynix announced the closure of its CIS department.

The reason behind this is that originally, SK Hynix developed the CIS business because it could utilize equipment and technology phased out from memory production, with low marginal costs, and also serve as a foothold to delve into non-memory markets. However, under market impacts, Samsung and SK Hynix have limited competitiveness in the high-end CIS market, with stagnant or even declining revenue. Shifting to HBM has become a natural choice for SK Hynix. To concentrate resources and consolidate its leading position in the AI chip field, SK Hynix has cut CIS R&D investment, reduced capacity, and transferred relevant personnel to the HBM department.

Sony Plans to Split Its Semiconductor Business

In the CIS field, another leading company reforming its business is Sony. It is reported that Sony Group is considering splitting its semiconductor business, aiming to achieve the split and independent listing as soon as 2025. Sony's initial plan is to distribute most of its chip business shares to shareholders, retaining only a small portion for itself.

However, Sony has not yet officially responded, with only insiders revealing that internal discussions are indeed underway, and considering the instability of the international market, such as US tariff policies, plan details may still be adjusted.

It is reported that Sony possesses the world's top CMOS image sensor technology. High-end sensors used in cameras by mobile phone giants such as Apple and Xiaomi mostly come from Sony. However, this technology, which once brought Sony a 25% profit margin, has encountered growth bottlenecks in recent years. According to Sony's third-quarter financial report for fiscal year 2024, operating profit in the Imaging & Sensing Products segment has slipped to just over 10%, making it the only segment among Sony's six major business segments to experience negative growth.

There are multiple factors contributing to this change. The current global smartphone market is generally weak, directly impacting demand for camera sensors. Coupled with US tariffs on China, supply chain costs continue to rise. Faced with this situation, separating the chip division can allow it to finance and expand more flexibly.

Intel's Frequent Moves

In April of this year, Intel announced the sale of a 51% stake in its Altera business to private equity giant Silver Lake Capital, with Intel retaining the remaining 49%.

Altera is a company focused on System-on-a-Programmable-Chip (SOPC), with early core products being FPGA (Field-Programmable Gate Array) and CPLD (Complex Programmable Logic Device), serving the communications, industrial, data center, and other fields. In 2015, Intel acquired Altera for $16.7 billion, operating it as the Programmable Solutions Group (PSG).

Intel's initial acquisition of Altera aimed to leverage its technical strength in the programmable chip field to expand its business in non-traditional CPU markets, in response to market changes and competitive pressures. However, in recent years, Intel has faced numerous challenges, and Altera has failed to help Intel achieve significant breakthroughs in new fields as expected. In January 2024, Altera was spun off from Intel as an independent entity and relaunched its original brand in March 2024.

Intel's split of the FPGA business is also a re-examination of the market and its own business.

This transaction establishes Altera's operational independence and makes it the largest pure FPGA semiconductor solutions company. The outside world believes that this shows Intel is beginning to spin off non-core businesses and assets to focus on core businesses. The transaction will help boost Intel's cash levels, as it actively cuts costs to strengthen its balance sheet.

In May, industry sources reported that Intel is considering spinning off its Network and Edge business unit (NEX), which primarily serves the telecommunications equipment and edge computing fields. Starting from Q1 2025, Intel will no longer separately disclose NEX financial data, having incorporated it into the Data Center and PC business, sending a clear signal of spin-off.

Samsung Electronics Plans to Spin Off Its Foundry Business

This is not the first time news has emerged about Samsung Electronics splitting its foundry business.

Samsung's foundry uses a 3-nanometer cutting-edge process to manufacture semiconductors and plans to achieve mass production of a 2-nanometer process within the year. Although Samsung's foundry scale is smaller than that of TSMC, the world's largest foundry, as the second-largest player in the global foundry industry, Samsung's foundry still has considerable competitive advantages. However, it has struggled to secure orders.

While Samsung's foundry technological capabilities and production volume issues contribute to this problem, industry analysts primarily attribute it to conflicts of interest. Since the DS division also has System LSI, which is responsible for semiconductor design, large technology companies specializing in design, such as Apple, NVIDIA, and Qualcomm, fear that if they outsource work to Samsung's foundry, their design technology may leak to System LSI. Therefore, separating the foundry is seen as a potential solution to the order shortage. Separating the foundry may also provide an opportunity to escape trillions of Korean won in deficits.

It is reported that due to continuous technological setbacks and deteriorating profitability, the System LSI business unit has been undergoing a comprehensive review by Samsung's Global Research Management Diagnosis Office since the beginning of this year. According to industry insiders, the management diagnosis is nearing completion, and the fate of this business unit is expected to be decided soon. Once the organizational restructuring of the DS division (including decisions regarding the System LSI business unit) is completed, the direction for the separation of the foundry may become clearer.

03

What Opportunities Do "Unshackled" Chip Titan Businesses Bring?

Faced with a fiercely competitive market, semiconductor companies choose to "unshackle" some business segments, not only to unload some of the asset burdens of development but also potentially rejuvenate the newly separated business segments with new development vitality.

For domestic semiconductors, the strategic retrenchment of giants is both an opportunity and a touchstone.

Amidst intensified market competition and mounting cost control pressures, the strategic adjustments in the global semiconductor industry are essentially unavoidable. As DDR3 memory prices continue to plummet and the iteration of MLC NAND technology slows, companies must confront the stark reality that profit margins in traditional businesses have been squeezed to a critical juncture. The spin-offs and market exits of giants like Samsung and SK Hynix are not tragic tales of self-sacrifice but rather astute decisions based on technological trends and cost structures, reallocating resources towards high-value-added areas like HBM and CMOS to fortify their positions for the forthcoming technological revolution.

For China's semiconductor industry, the market gap created by the retreat of these giants presents both an opportunity and a challenge. In mature sectors such as DDR3 and MLC NAND, domestic manufacturers are entering a crucial window period of growth. However, the linchpin of future progress hinges on steering clear of the pitfalls of low-price competition and forging differentiated advantages through technological advancements.