AMD’s efforts to secure high‑bandwidth memory and shape AI data center standards via Korean partners

AMD is intensifying its strategic efforts to secure critical high-bandwidth memory (HBM) supplies and shape the future of AI data center infrastructure through deepening collaborations with South Korean partners, particularly Samsung Electronics. These moves come amid unprecedented demand surges and supply constraints in the semiconductor industry driven by the explosive growth of AI workloads, which are reshaping memory priorities and fabrication strategies globally.

---

Lisa Su’s High-Level Engagements in South Korea: Securing AI Memory Supply and Roadmap Alignment

In early 2026, AMD CEO Lisa Su undertook a series of pivotal visits to South Korea, engaging directly with Samsung Electronics Chairman Jay Y. Lee and top executives to negotiate and solidify partnerships centered on AI-specific high-bandwidth memory (HBM). These discussions are crucial as Samsung and SK Hynix dominate the HBM market, which is now the focal point of intense industry demand due to AI accelerator expansion.

Key highlights from these engagements include:

• Guaranteeing stable and enhanced HBM supply for AMD’s upcoming AI hardware, including the highly anticipated MI450 GPU accelerator, which requires large volumes of cutting-edge memory to meet performance targets.
• Coordinating memory roadmap alignments tailored for AI workloads, with Samsung committing to collaborate on optimizing memory architectures to boost efficiency and throughput.
• Responding to market signals like NVIDIA CEO Jensen Huang’s acknowledgment of “unprecedented” memory shortages that are rivaling even advanced logic fabrication as a bottleneck in AI chip deployment.
• Seeking preferential supply agreements and capacity guarantees to shield AMD from volatile pricing and allocation pressures, ensuring AMD’s AI product timelines and cost structures remain viable amid heightened competition.

---

Leadership in Open Optical Interconnect Standards: Shaping Interoperable AI Data Centers

Beyond securing memory supply, AMD is taking a proactive role in defining open standards for AI data center infrastructure. This initiative aims to enable scalable, interoperable, and high-performance AI clusters by harmonizing compute, memory, and interconnect technologies.

AMD’s leadership includes:

• Spearheading the Optical Connectivity Initiative (OCI), an industry consortium promoting open optical interconnect standards designed to reduce vendor lock-in and improve supply chain resilience.
• Collaborating closely with South Korean semiconductor leaders and ecosystem players—including memory vendors like Samsung and interconnect specialists such as Broadcom—to align hardware components across GPUs, CPUs, and memory modules with open optical interfaces.
• Driving a unified framework that addresses the growing complexity of AI data center deployments, facilitating seamless data flow between AMD’s MI series accelerators and other infrastructure components.
• Supporting faster innovation cycles and operational efficiency by promoting interoperable hardware stacks that can be adopted broadly across hyperscalers and AI infrastructure providers.

These efforts reflect AMD’s ambition to embed its technology deeply within evolving AI ecosystems, ensuring its products remain integral to next-generation data centers.

---

Industry Dynamics: The ‘Great Wafer Cannibalization’ and Supply Chain Pressures

AMD’s strategic moves must be understood within the broader context of industry-wide shifts caused by AI-driven wafer reallocation, often dubbed the “Great Wafer Cannibalization.” As AI demand reshapes chip fabrication priorities, foundries and memory manufacturers are reallocating wafer capacity from traditional consumer and enterprise products toward AI-specific silicon and memory.

This trend has several implications:

• The South Korean memory duopoly—Samsung and SK Hynix—faces escalating pressure to prioritize HBM production for AI accelerators, tightening supply and driving up prices.
• Wafer cannibalization intensifies competition for limited fabrication resources, increasing the risk of supply chain disruptions for vendors without secured agreements.
• AMD’s efforts to secure preferential memory supply and influence open interconnect standards are designed to mitigate risks posed by this volatile environment.
• Hyperscalers and AI hardware vendors share these concerns, underscoring the critical importance of stable memory and interconnect availability to meet aggressive AI deployment schedules.

---

Strategic Significance and Market Impact

AMD’s engagements in South Korea and leadership in open data center standards underscore several strategic imperatives:

• Mitigating supply chain risks by locking in critical HBM supply and collaborating on next-generation memory architectures tailored for AI performance.
• Shaping AI data center infrastructure standards to foster a more resilient and interoperable ecosystem that reduces fragmentation and vendor lock-in.
• Positioning AMD competitively against rivals such as NVIDIA, which faces similar memory shortages, by securing preferential access and influencing ecosystem direction.
• Enhancing global supply chain resilience amid geopolitical uncertainties and regulatory challenges that could disrupt semiconductor sourcing.

---

Summary and Outlook

AMD’s intensified collaboration with South Korean semiconductor leaders, highlighted by CEO Lisa Su’s high-profile engagements with Samsung, represents a critical strategic push to secure high-bandwidth memory supplies essential for AI hardware growth. At the same time, AMD’s leadership in the Optical Connectivity Initiative (OCI) and broader industry standardization efforts aims to shape the future of AI data center infrastructure, promoting interoperable, scalable, and resilient ecosystems.

Together, these initiatives position AMD to navigate the increasingly complex AI semiconductor landscape by combining supply chain security with collaborative innovation, ensuring the company maintains momentum in the fiercely competitive AI accelerator market. Looking ahead, AMD’s ability to sustain these partnerships and influence standards will be pivotal as AI continues to drive transformative demand shifts across the semiconductor industry.