Samsung Vice Chairman Jun Young-hyun held a private meeting with NVIDIA CEO Jensen Huang. Their negotiations have expanded from HBM4 to HBM5, foundry services, and joint ASIC development. Samsung is now poised to secure the next-generation Groq LP40 chip order, which was widely expected to go to Taiwan Semiconductor. Although a long-term agreement has not yet been finalized, this deep strategic alignment—spanning multiple core segments of the AI supply chain—is quietly reshaping the competitive landscape in high-end semiconductors.
$Samsung Electronics (005930.KR)$and$NVIDIA (NVDA.US)$is pushing bilateral cooperation to a new depth, with negotiations expanding from current product lines to next-generation high-bandwidth memory and chip foundry services, marking a further strategic alignment between these two tech giants in the AI infrastructure supply chain.
According to TrendForce News on Tuesday, Samsung Electronics Vice Chairman Jun Young-hyun met with NVIDIA CEO Jensen Huang on June 8 to discuss potential cooperation on HBM and foundry services.
Jun Young-hyun stated that the immediate priority is ensuring stable supply of HBM4 and SOCAMM this year, while both parties also held discussions on long-term collaboration starting next year, covering HBM4E, foundry services, and HBM5.
This meeting followed Jensen Huang’s high-profile visit to Korea and a series of meetings with SK hynix, drawing significant attention to whether Samsung can solidify its position within NVIDIA’s supply chain. Although Jun Young-hyun did not confirm whether the two sides would sign a long-term memory supply agreement, he told Newspim that Samsung would spare no effort to support NVIDIA’s success as a key partner.
Foundry collaboration extends to next-generation Groq chips
At the foundry business level,$Samsung Electronics (005930.KR)$and$NVIDIA (NVDA.US)$the scope of collaboration is expanding. Samsung is negotiating with NVIDIA to produce next-generation chips using advanced process nodes, including the Drive AGX Thor autonomous driving chip and Groq's Language Processing Unit (LPU).
Jun Young-hyun noted that Samsung is currently manufacturing NVIDIA’s autonomous driving chips and Groq chips using 4nm and 8nm process technologies, with collaboration extending to the next-generation Groq chip. Samsung is already producing the third-generation Groq LPU (LP30) using its 4nm process, and Jun’s remarks suggest Samsung is also capable of manufacturing the next-generation LP40—despite industry expectations that Taiwan Semiconductor would secure the order due to its advanced packaging advantages.
During a Q&A session at NVIDIA’s Korea AI Ecosystem Reception, CEO Jensen Huang also commented on the partnership. According to Dealsite, Huang stated that NVIDIA and Samsung have a long-standing collaboration in ASICs and are currently co-developing new ASIC products, with a similarly extensive history of cooperation in memory technologies.
Memory supply covers the full Vera Rubin platform product lineup
In the memory business,$Samsung Electronics (005930.KR)$has already provided multiple products for$NVIDIA (NVDA.US)$Samsung is supplying sixth-generation HBM4 memory for the Vera Rubin platform, offering a data transfer rate of 11.7 Gbps; it is also providing SOCAMM2 modules based on LPDDR5X for the Vera CPU, along with a PCIe Gen6-based PM1763 storage solution.
Regarding next-generation products, Samsung’s HBM4E combines DRAM core chips with its in-house developed 4nm foundry base die, achieving a data transfer rate of 14 Gbps and reaching up to 16 Gbps in testing. If this specification can be successfully mass-produced, it will enable Samsung to compete more directly with SK hynix in the high-end HBM market.
Long-term agreements remain unresolved, and the cooperation framework is still taking shape.
Although the scope of collaboration between the two parties continues to expand, a critical long-term supply agreement has yet to be finalized. When asked whether a long-term memory supply agreement would be signed, Jun Young-hyun declined to confirm explicitly, using relatively cautious language.
In the broader context, this round of talks covered multiple dimensions—including HBM4, HBM4E, HBM5, foundry processes, and joint ASIC development—indicating that the partnership has evolved from single-product procurement toward deeper technological collaboration. For investors, key indicators of substantive progress in this cooperation will be whether Samsung can successfully secure LP40 orders on the foundry side and further increase its share in the HBM supply chain.
Want to select stocks or get a stock diagnosis? Curious about the opportunities and risks in your portfolio? For all your investment questions,Just ask Futubull AI!
Editor/melody