Focus on Next‑Generation AI Memory

Under the agreement announced Wednesday, Samsung will commit to supplying its latest sixth-generation High‑Bandwidth Memory (HBM4) to AMD for its upcoming Instinct MI455X AI accelerators. These high‑performance memory chips are designed to enable faster AI training and inference workloads, providing critical bandwidth and energy efficiency that modern AI applications demand.

In addition to HBM4, Samsung will also provide optimized DDR5 memory tailored for AMD’s sixth‑generation EPYC server processors, codenamed “Venice,” strengthening the memory foundation for next‑gen CPU and accelerator combinations.

Samsung has already been a key supplier of advanced memory to AMD, including HBM3E chips used in earlier AMD Instinct accelerators. The new MoU expands this role and strategically positions Samsung as a primary memory partner in AMD’s AI ecosystem.

Exploring Foundry Collaboration

Beyond memory supply, the partners also agreed to explore opportunities for a foundry partnership under which Samsung could manufacture future AMD chips. Although formal terms and timelines were not disclosed, the discussions signal AMD’s interest in diversifying its manufacturing base beyond existing foundry relationships a potentially significant shift in the global semiconductor landscape.

Samsung’s advanced foundry capabilities, particularly its progress in cutting‑edge process nodes and integrated packaging technologies, have made it a compelling candidate for broader collaboration. Industry analysts note that Samsung’s foundry business has been positioning itself to compete more aggressively with established leaders like TSMC as AI‑driven custom silicon demands grow.

Strategic Context in a Competitive Market

The announcement coincides with the annual Nvidia GTC developer conference, where Nvidia also highlighted its own foundry collaboration with Samsung and showcased Samsung‑made HBM4 memory. This underscores how key partnerships with memory and foundry suppliers have become strategic assets in the AI arms race.

The HBM market itself is intensely competitive. While Samsung holds a significant share, it trails rivals like SK Hynix in overall market dominance. Long‑term agreements like this one with AMD aim to reinforce Samsung’s position as demand for high‑performance memory continues to rise across cloud data centers and AI infrastructure.

Industry and Market Impact

The collaboration comes amid broader moves by AMD to secure chip supply commitments; recent reports indicate multi‑billion‑dollar deals with major cloud players such as Meta and OpenAI, which will further drive demand for AMD’s AI accelerators and memory components.

Shares of both companies reacted positively in extended trading, with investors signaling confidence in longer‑term strategic positioning, although market responses varied across sectors. Observers say the deal could reshape supply dynamics for AI hardware by strengthening ecosystem relationships that span silicon design, memory technology, and manufacturing.

Voices from the Leadership

In a joint statement, Samsung’s Vice Chairman & CEO Young Hyun Jun emphasized the importance of deepening collaboration across memory, logic and advanced packaging to support AMD’s AI roadmap. AMD Chair and CEO Dr. Lisa Su highlighted the significance of integrated supply and innovation across compute stacks from silicon to rack‑scale systems to accelerate AI capabilities with real‑world impact.

________________________________________________________________________________________________________________

Among the significant announcements, Nvidia revealed plans to sell hundreds of thousands of its AI chips in Saudi Arabia, starting with an initial batch of 18,000 “Blackwell” chips for Humain, an AI startup launched by the Saudi Arabian sovereign wealth fund. Additionally, chipmaker Advanced Micro Devices (AMD) revealed a $10 billion partnership with Humain. Qualcomm also entered into an agreement with Humain, signing a memorandum of understanding to develop and build a central processing unit (CPU) for data centers. Qualcomm had acquired server CPU manufacturer Nuvia in 2021 but has not yet released any products.

Trump’s Gulf tour began on Tuesday, marked by a strategic economic agreement with Saudi Arabia, as the country extended a warm welcome. The purpose of Trump’s visit is to secure investments totaling trillions of dollars. According to the White House, the U.S. and Saudi Arabia will share technology investments. Saudi firm DataVolt is expected to invest $20 billion in AI data centers and energy infrastructure in the U.S. Meanwhile, U.S. tech giants like Alphabet (Google), Oracle, Salesforce, AMD, and Uber are set to invest $80 billion into advanced technologies across both nations.

Trump’s next stop will be the UAE on Thursday. Reports suggest that the Trump administration is close to finalizing a deal for the UAE to purchase large quantities of Nvidia’s AI chips.

Saudi Arabia is working to diversify its economy and reduce reliance on oil revenue, aiming to become a global leader in AI technology. To this end, Saudi Crown Prince Mohammed bin Salman launched Humain on Monday, which will focus on AI development in the kingdom.

Nvidia and Humain will collaborate to establish Saudi Arabia as a major player in AI and cloud computing, with plans to build AI factories equipped with Nvidia’s advanced GPUs over the next five years, with a total capacity of up to 500 megawatts. AMD’s partnership with Humain also includes a five-year plan to deploy AI infrastructure and create a next-generation cloud computing platform.

The Humain-AMD agreement, besides hardware, will help develop a platform with global significance, enhancing performance, openness, and reach. Part of the strategy is to avoid dependency on a single AI hardware vendor.

Humain, which will be chaired by Crown Prince Mohammed bin Salman, operates under the Public Investment Fund and will focus on AI services, including data centers, cloud capabilities, and advanced AI models. As Humain CEO Tareq Amin highlighted, the initiative is a critical step in establishing AI infrastructure within Saudi Arabia.