Samsung Electronics is poised to initiate production of its advanced high-bandwidth memory (HBM) chips, specifically the HBM4 generation, starting next month, with plans to supply these critical components to Nvidia. This development marks a significant step in Samsung's efforts to strengthen its position in the competitive AI hardware market, as revealed by a source familiar with the matter to Reuters on Monday.
Intensifying Rivalry in the Memory Chip Sector
The move comes as Samsung strives to catch up with its domestic rival, SK Hynix, which has established itself as a primary supplier of sophisticated memory chips essential for Nvidia's AI accelerators. Earlier last year, Samsung faced challenges with supply delays that adversely impacted its earnings and share prices, prompting a strategic push to regain market leadership.
Market Reactions and Financial Implications
In response to this news, Samsung's shares experienced a notable increase of 2.2% during morning trading, while SK Hynix shares declined by 2.9%. This market movement underscores the high stakes involved in the HBM chip segment, where both companies are vying for dominance. The South Korean firm has also seen a significant rise in contract prices for its 32 gigabyte DDR5 memory chip modules, which surged to $239 in November from $149 in September, reflecting strong demand and supply dynamics.
Details on Production and Supply Plans
While the source declined to specify the exact volume of chips Samsung intends to supply to Nvidia, the Korea Economic Daily reported on Monday that Samsung has successfully passed HBM4 qualification tests for both Nvidia and AMD. According to chip industry sources cited by the newspaper, Samsung is set to begin shipping these chips to Nvidia next month. A Samsung spokesperson has refrained from commenting on the matter, and Nvidia was not immediately available for response.
SK Hynix's Competitive Moves
On the other side, SK Hynix announced in October that it has finalized HBM supply discussions with major customers for the upcoming year. The company plans to commence the deployment of silicon wafers next month into its new fabrication facility, M15X, located in Cheongju, South Korea, to produce HBM chips. An executive from SK Hynix mentioned this to Reuters earlier this month, though it remains unclear whether HBM4 will be included in the initial production phase.
Upcoming Earnings and Industry Outlook
Both Samsung and SK Hynix are scheduled to announce their fourth-quarter earnings on Thursday, where they are anticipated to provide further details regarding HBM4 orders. This disclosure is expected to shed light on the competitive landscape and future strategies in the high-stakes memory chip industry.
Nvidia's Role and Future Developments
Nvidia's CEO, Jensen Huang, highlighted earlier this month that the company's next-generation Vera Rubin platform is in "full production." This platform is designed to be paired with HBM4 chips, with a launch planned for later this year, indicating a robust demand for advanced memory solutions in AI applications.
The race to supply HBM4 chips underscores the critical role of memory technology in powering AI innovations, with Samsung and SK Hynix at the forefront of this technological advancement. As production ramps up, the industry will closely monitor how these developments influence market shares, pricing, and the broader AI ecosystem.
