Article count:25239 Read by:103424336

Account Entry

HBM, urgent expansion of production

Latest update time:2024-09-13
    Reads:

????If you hope to meet more often, please mark the star ?????? and add it to your collection~

Source: Content from China Business Times, thank you.

Due to the strong demand for high-bandwidth memory (HBM), SK Hynix is ​​rumored to convert the DRAM production line at the M10F semiconductor plant in Icheon, South Korea, to produce the fifth-generation HBM3E. After the M10F production line is started, SK Hynix's monthly HBM production capacity is expected to increase to 150,000 pieces.
According to foreign media reports, SK Hynix is ​​converting its M10F production line for DRAM to produce HBM. The M10F production line will be used to produce HBM3E. The goal is to complete infrastructure construction before the beginning of 2025 and introduce equipment starting in the first quarter of 2025.
Industry insiders estimate that the M10F production line can produce about 10,000 HBMs per month, which is smaller than the existing HBM production line. Currently, SK Hynix's monthly HBM production capacity is about 140,000 pieces. After the M10F production line is officially put into production, SK Hynix's monthly HBM production capacity will increase to 150,000 pieces.
AI has become an arms race in the technology industry, and HBM is one of the most important components. Every time the AI ​​chipset processes a task, the GPU will transfer data to the memory, which requires the high-speed data transmission capability of HBM.
It is understood that the seven major technology giants in the United States, including Apple, Microsoft, Google, Amazon, Nvidia, Facebook and Tesla, have all requested SK Hynix to supply HBM, causing SK Hynix to actively expand its HBM production capacity.
Considering factors such as the mass production time of M10F, SK Hynix will place orders for equipment from the fourth quarter of 2024, and will introduce packaging equipment, testing equipment, etc. for memory stacking.

SK Hynix's competitors, such as Samsung Electronics and Micron, are also expanding HBM production. In addition, Chinese companies such as Changxin Memory have announced their entry into the HBM market. In response to this situation, SK Hynix has formulated a strategy of "focusing on mass production of the most advanced HBM". It will focus on developing and mass producing the most advanced HBM and gradually phase out old products.

Behind the fierce HBM competition between Samsung and SK Hynix

Samsung Electronics and SK Hynix are competing fiercely in the high-bandwidth memory (HBM) market, while Nvidia, a leader in artificial intelligence (AI) semiconductors, has become the focus. Recently, Nvidia has frequently interacted with the two Korean manufacturers, which is seen as a move to stimulate price and technology competition.

Analysts pointed out that although Nvidia has been in contact with HBM suppliers recently, there has been no clear information about actual orders, which has caused the industry to suspect that Nvidia intends to stimulate competition between the two HBM manufacturers.
Previously, Nvidia CEO Huang Renxun mentioned testing Samsung's 12-layer HBM3E products, but there is no clear news yet. However, in April 2024, Huang Renxun suddenly met with SK Group Chairman Cui Taiyuan in Silicon Valley, which was interpreted as an attempt to push Samsung to speed up production and lower prices.
Industry analysts point out that HBM3 prices have risen more than fivefold since 2023, and the upcoming HBM3E products may be even more expensive. Therefore, for Nvidia, this may lead to excessive R&D and production costs, which is one of the reasons why Nvdia stimulates competition between the two Korean chip manufacturers.
According to South Korea's Economic Daily, Samsung recently mobilized "400 experts" to form a 100-person special team to improve the yield of 12-layer HBM3E and strive to pass Nvidia's quality certification test within May. In addition, Samsung has arranged more than 300 people to participate in the development of HBM4, and plans to complete the research of HBM4 by the end of 2024 and officially supply it to Nvidia in 2025.
Previously, SK Hynix announced that it plans to provide 12-layer HBM3E samples in May 2024 and plans to achieve mass production in the third quarter of 2024. SK Hynix also revealed a roadmap for mass production of 12-layer HBM4 in 2025 and supply of 16-layer products in 2026.
Since Samsung announced the development of 12-layer HBM3E ahead of its competitors in February 2024, there have been constant rumors of early mass production. The industry generally believes that the mass production time is the second quarter of 2024, earlier than the third quarter (July to September) announced by SK Hynix. The investment of 400 experts this time highlights Samsung's determination to win the title of "the first mass-produced 12-layer HBM3E."
Samsung used to lag behind SK Hynix in the HBM business as Samsung disbanded its HBM business unit in 2019, leaving SK Hynix as the only stable producer of HBM3 when major tech companies began investing in generative AI development in 2023.
Currently, SK Hynix has a market share of over 90% in the high-yield HBM3 market and has a deep cooperative relationship with Nvidia, while Samsung mainly supplies the lower-priced HBM2 and HBM2E. Therefore, Samsung attaches great importance to leading in the mass production of 12-layer HBM3E in order to seize the market and further improve profitability.
It is reported that Nvidia's new generation of AI accelerators B200 and GB200 upgraded versions may be equipped with more than 8 12-layer HBM3E chips. Samsung believes that if it can compete with SK Hynix on the same terms, it is very likely to reverse the situation. Nvidia's intention to increase HBM suppliers is also beneficial to Samsung.
Some analysts also pointed out that Samsung's 8-layer HBM3E mass production time is about a month later than SK Hynix, and the current output is also relatively low. In 2024 when the demand for 12-layer HBM3E is not strong, stable yield rather than early mass production is the key to ensuring profitability and technological competitiveness.

END


????Semiconductor boutique public account recommendation????

▲Click on the business card above to follow

Focus on more original content in the semiconductor field


▲Click on the business card above to follow

Focus on the trends and developments of the global semiconductor industry

*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.



Today is the 3884th content shared by "Semiconductor Industry Observer" for you, welcome to follow.


Recommended Reading


"The first vertical media in semiconductor industry"

Real-time professional original depth

Public account ID: icbank


If you like our content, please click "Reading" to share it with your friends.

 
EEWorld WeChat Subscription

 
EEWorld WeChat Service Number

 
AutoDevelopers

About Us Customer Service Contact Information Datasheet Sitemap LatestNews

Room 1530, Zhongguancun MOOC Times Building,Block B, 18 Zhongguancun Street, Haidian District,Beijing, China Tel:(010)82350740 Postcode:100190

Copyright © 2005-2024 EEWORLD.com.cn, Inc. All rights reserved 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号