News
South Korean media reports that the main suppliers of artificial intelligence (AI) chip packaging are concentrated in TSMC and ASE Technology Holding Co., which have been actively expanding production to meet the growing market demand. Despite efforts to develop technology and invest, South Korean companies like Samsung Electronics have not been able to narrow the gap with TSMC and ASE.
According to the Chosun Ilbo, industry insiders indicated that TSMC is expanding its advanced packaging (CoWoS) capacity by selecting a site in the southern region, while ASE also announced the construction of a second packaging and testing factory in California, USA, and plans to build another in Mexico. The rapid growth of the AI chip market highlights the increasing importance of semiconductor packaging and testing. As the benefits of semiconductor process miniaturization diminish and production costs rise, advanced packaging that can connect multiple components has become an ideal alternative solution. Some organizations predict that the semiconductor packaging market is expected to grow by more than 10% annually and expand to USD 90 billion by 2030.
Taiwanese companies like TSMC and ASE benefit a lot, almost monopolizing the contract manufacturing of AI chips for companies like NVIDIA and AMD. In terms of chip manufacturing, TSMC aims to double its CoWoS capacity from the previous year to meet increasing orders. TSMC recently announced plans to build two new advanced packaging factories in the southwest. The construction of the first factory was paused due to the discovery of ancient artifacts, but TSMC quickly sought a new site and announced an expansion of CoWoS facilities investment by 2025.
ASE, serving customers including Qualcomm, Intel and AMD, is also striving to increase equipment investment to meet rising orders. ASE, with the highest market share in the semiconductor packaging and testing field, is increasing its capacity and considering building a factory in Japan to match the growing demand. ASE’s CEO Wu Tianyu stated that they are looking for a location in Japan with a solid semiconductor ecosystem for the new factory.
Samsung has also announced packaging investment plans. The company intends to raise the investment in the new plant in Taylor, Texas, USA from USD 17 billion to more than USD 40 billion for the construction of an advanced packaging research and development center and facilities, in which it will allocate over KRW 2 trillion annually to expand advanced packaging production lines.
South Korean semiconductor back-end packaging and testing (OSAT) companies such as Hana Micron and Nepes are also striving for AI chip packaging orders based on technical development. Hana Micron, the leading OSAT company in South Korea, has announced its commitment to developing 2.5D AI semiconductor packaging. Nepes is developing Package on Package (PoP) technology, which integrates different semiconductors into one chip, with a target for commercial mass production in the second half of 2025.
Despite the efforts of South Korean companies, it is difficult to narrow the gap with Taiwanese companies in the short term. Taiwanese companies have actively developed advanced semiconductor packaging and commercializing CoWoS at a earlier time, while South Korean packaging companies lag in accumulated technologies. South Korean industry insiders point out that TSMC and ASE have been collaborating for over 30 years. Therefore, as TSMC secured a large number of AI chip orders, it would prove a boon to Taiwan’s packaging ecosystem. In contrast, South Korea’s packaging industry, which has long focused on the memory production market, still has a long way to go to expand its market and even compete with Taiwanese companies.
News
Unlike other major semiconductor manufacturers, including Intel and TSMC, memory giant Micron is not in a hurry to adopt EUV (extreme ultraviolet) lithography for its DRAM production. However, according to a latest report from Technews, in 2024, Micron plans to begin trial production using EUV on its 1γ (1-gamma) process technology at 10-nm level.
The report also notes that currently, all of the company’s mass-produced products are made using DUV (deep ultraviolet) lithography. However, after entering trial production in 2024 with EUV, Micron also anticipates that this process technology will enter large-scale production in 2025.
Another Korean memory giant, Samsung, announced in 2020 that it has successfully shipped one million of the industry’s first 10nm-class (D1x) DDR4 (Double Date Rate 4) DRAM modules based on EUV technology.
In 2021, SK hynix has started mass production of its 10-nm DRAM chips using EUV technology, and is said to invest USD 1.5 billion this year to acquire 8 advanced EUV lithography machines, according to an earlier report from Disc Manufacturer.
Previously, Micron CEO Sanjay Mehrotra stated during an earnings call that the trial production of 10-nm-class 1γ (1-gamma) process DRAM using EUV lithography is progressing well, and they are on track to achieve mass production by 2025 as planned. Currently, Micron is developing the 10-nanometer-class 1γ process DRAM manufacturing technology using EUV lithography at its Hiroshima plant in Japan, which is also the first site for the trial production of 1γ memory, according to Technews.
In order to meet the strong demand for high-performance memory chips driven by AI, Micron is reportedly building a pilot production line for HBM in the U.S. and is considering producing HBM in Malaysia for the first time.
Read more
(Photo credit: ASML)
News
The memory market is showing signs of recovery, with Japanese NAND giant Kioxia ending its production cut for NAND Flash at the end of June this year. The company’s Yokkaichi Plant and Kitakami Plant have resumed full production capacity. However, this move could influence the overall NAND market, impacting downstream memory companies like Taiwanese NAND manufacturers Phison, ADATA, Team Group and Apacer.
Citing industry sources, a report by the Economic Daily News states that with Kioxia’s capacity utilization returning to 100%, competitors like Samsung and SK Hynix may also increase their production to maintain market share. This potential surge in supply could lead to a market downturn.
Kioxia’s return to full capacity in June means that increased NAND chip production could enter the market as early as August or September, affecting the traditional peak season of downstream Taiwanese NAND controller chip manufacturers such as Phison, ADATA, and Team Group.
The report notes that Phison believes that current market conditions show strong demand for NAND chips used in SSDs. With NAND chip prices having returned to pre-pandemic levels, manufacturers are beginning to see normal profits. To compensate for losses in 2023, prices are expected to remain firm, maintaining a positive cycle.
In contrast, ADATA has a more conservative outlook on the market. The severe losses experienced by major global NAND chip suppliers in 2023 have led to an increase in production capacity and sales to recoup last year’s losses. This could result in another supply glut in the market.
TrendForce observes that restrained production increases in the first half of the year led to a rapid rebound in NAND Flash prices, helping manufacturers return to profitability. However, with significant production expansion planned for the second half of the year and retail market demand still weak, wafer spot prices are declining. The drop has been so significant that some wafer prices are now more than 20% below contract prices, making it difficult to sustain future contract price increases.
From a demand perspective, the third quarter will see continued investments in server infrastructure, particularly benefiting enterprise SSDs due to the expanding use of AI. However, consumer electronics demand remains sluggish, and with aggressive production increases by manufacturers in the latter half of the year, the sufficiency ratio of NAND Flash is expected to rise to 2.3% in the third quarter. The blended price increase for NAND Flash is projected to narrow to 5-10% quarter-on-quarter.
Read more
(Photo credit: Kioxia)
News
Thanks to the rebound in memory chip demand amid accelerated global AI development, Samsung Electronics reported its strongest sales and profit growth in years. According to its financial guidance announced on July 5th, the semiconductor giant projects its operating profit to grow more than 15-fold YoY to 10.4 trillion won (USD 7.5 billion) in its preliminary results for the April-June quarter, outstripping market expectations.
In addition, the company expects its sales to increase by approximately 23% to 74 trillion won. According to a report from Bloomberg, this marks the largest rise since the peak levels seen during the Covid-19 pandemic in 2021.
The forecast is way better than LSEG SmartEstimate’s earlier forecast, which expected Samsung Electronics’ operating profit for Q2 2024 to reach 8.8 trillion won (roughly USD 6.34 billion).
Samsung is scheduled to release final earnings, including divisional breakdowns, on July 31.
It is also worth noting that Samsung is releasing its results just days ahead of planned three-day walkouts by union organizers, starting from July 8th. According to Bloomberg, the move would involve over 28,000 members, including those at crucial chip plants, due to a wage dispute. The extent of participation in Monday’s walkout remains uncertain at this time.
Citing market sources, the report noted that Samsung’s Q2 financial results highlight the memory market’s robust recovery this year from a sharp decline post-Covid, driven by increased demand from data centers and AI development, which contributes to a turnaround in Samsung’s largest division, which had incurred losses the previous year.
According to TrendForce, Samsung’s global share of DRAM and NAND Flash output in 2023 was 46.8% and 32.4%, respectively. An earlier report by the Korea Economic Daily indicated that Samsung’s HBM production has been sold out in 2024.
According to the latest forecast by Trendforce, the HBM market is poised for robust growth, driven by significant pricing premiums and increased capacity needs for AI chips. HBM prices are expected to Increase by 5–10% in 2025.
Read more
(Photo credit: Samsung)
News
In order to address the growing demand for high-performance memory solutions fueled by the expansion of the artificial intelligence (AI) market, Samsung Electronics has formed a new “HBM Development Team” within its Device Solutions (DS) Division to enhance its competitive edge in high-bandwidth memory (HBM), according to the latest report from Business Korea. The new team will concentrate on advancing the progress on HBM3, HBM3e, and the next-generation HBM4 technologies, the report noted.
This initiative comes shortly after the Korean memory giant changed its semiconductor business leader in May. Citing industry sources, the report stated that Samsung’s DS Division carried out an organizational restructuring centered on the establishment of the HBM Development Team.
Also, the move attracts attention as on July 4th, a report from Korea media outlet Newdaily indicated that Samsung has finally obtained approval from NVIDIA for qualification of its 5th generation HBM, HBM3e, though the company denied the market rumor afterwards.
Samsung has a long history of dedicating to HBM development. Since 2015, it has maintained an HBM development organization within its Memory Business Division. Earlier this year, the tech heavyweight also created a task force (TF) to boost its HBM competitiveness, and the new team will unify and enhance these ongoing efforts, the report noted.
According to the report, Samsung reached a significant milestone in February by developing the industry’s first HBM3e 12-layer stack, which offers the industry’s largest capacity of 36 gigabytes (GB). Samples of the HBM3e 8-layer and 12-layer stacks have already been sent to NVIDIA for quality testing.
Regarding the latest development, TrendForce reports that Samsung is still collaborating with NVIDIA and other major customers on the qualifications for both 8-hi and 12-hi HBM3e products. Samsung anticipates that its HBM3e qualification will be partially completed by the end of 3Q24.
According to TrendForce’s latest analysis on the HBM market, HBM production will be prioritized due to its profitability and increasing demand. However, limited yields of around 50–60% and a wafer area 60% larger than DRAM products mean a higher proportion of wafer input is required. Based on the TSV capacity of each company, HBM is expected to account for 35% of advanced process wafer input by the end of this year, with the remaining wafer capacity used for LPDDR5(X) and DDR5 products.
Read more
(Photo credit: Samsung)