AMD Forges Dual Alliances in Strategic Push Against AI Market Leader

In a concentrated display of strategic maneuvering, Advanced Micro Devices (AMD) announced two significant partnerships during CEO Dr. Lisa Su’s recent visit to South Korea. The agreements, signed with Samsung Electronics and NAVER Cloud within a 24-hour period, represent a direct challenge to Nvidia’s established supremacy in the artificial intelligence semiconductor sector.

Securing Next-Generation Memory with Samsung

A cornerstone of AMD’s strategy involves securing a reliable supply of advanced memory, a critical bottleneck in AI infrastructure. The company solidified this through a memorandum of understanding with Samsung Electronics. Under this agreement, Samsung is designated as the primary supplier of next-generation HBM4 memory for AMD’s upcoming Instinct MI455X AI accelerators. This memory will be built on a 10-nanometer DRAM process, capable of data transfer speeds reaching up to 13 gigabits per second.

Furthermore, Samsung will provide optimized DDR5 memory for AMD’s sixth-generation EPYC server processors. The collaboration extends a nearly two-decade-long relationship between the two firms; Samsung previously supplied HBM3E chips for AMD’s MI350X and MI355X accelerators. Notably, the new agreement also opens discussions for a potential foundry partnership, where Samsung could manufacture future AMD chips on a contract basis.

This deal provides AMD with a secured pipeline for high-performance memory in a competitive market. For context, Samsung currently holds approximately 22% of the global HBM market, trailing leader SK Hynix, which commands a 57% share.

Should investors sell immediately? Or is it worth buying AMD?

Gaining Foothold with NAVER Cloud

Concurrently, AMD entered a strategic collaboration with NAVER Cloud, the cloud computing arm of the South Korean internet giant. The central objective is to migrate NAVER’s large language model, “HyperCLOVA X,” to run on AMD’s GPU hardware. AMD will supply Instinct MI455X accelerators for this initiative, and the two companies will jointly work on optimizing software stacks using AMD’s ROCm open software platform.

Industry observers view this partnership as a potential first step toward displacing Nvidia GPUs within NAVER’s operations, though the scale of any such transition remains uncertain. The alliance also includes plans for joint research initiatives and providing AI computing capacity for academic projects.

Strategic Timing and Broader Ambitions

The timing of Dr. Su’s visit carried symbolic weight, occurring during the same week as Nvidia’s annual GTC developer conference, where CEO Jensen Huang also announced a Korean partnership and praised HBM4 technology. Market analysts interpret AMD’s moves as an attempt to forge a cohesive ecosystem—combining AMD GPUs, Samsung memory, and NAVER’s AI services—through direct negotiations with the CEOs of each company.

AMD’s data center business provides a strong tailwind for this strategy, having recently reported a 39% year-over-year revenue increase. This segment now constitutes over half of AMD’s total revenue. Company leadership is targeting an annual growth rate exceeding 35% for the next three to five years. By securing Samsung as a key HBM4 supplier and establishing NAVER as its first sovereign AI partner in Korea, AMD has made its ambition to offer a complete AI infrastructure alternative to Nvidia more tangible.