AMD Chief Executive Says She Is Exploring a Foundry and Memory Path With Samsung, Spotlighting Supply Tightness in AI Chips

Advanced Micro Devices is again putting its manufacturing strategy in focus as demand for AI-related semiconductors continues to run ahead of supply. In recent market coverage, AMD Chief Executive Lisa Su was described as exploring a foundry partnership and a memory-related deal involving a major Korean technology company, with the goal of expanding the throughput of chips used in AI infrastructure.

The discussion centers on foundry capacity, meaning the external manufacturing operation that produces chips on silicon wafers using specialized process technology. AMD designs the chips, but leading-edge manufacturing is often constrained by wafer capacity, process development, and packaging. If AMD can secure more reliable access to additional capacity, it can reduce the risk that product launches and AI customer demand get delayed by production limits.

The same coverage also referenced memory, another key component in AI systems. In practice, AI accelerators depend not only on compute chips, but also on high-bandwidth memory and the overall supply chain that feeds data-hungry workloads. Any coordinated effort around memory, even at the planning stage, would be aimed at strengthening the system-level availability of AI servers and other end products.

For AMD, the timing matters because the company’s AI business is closely tied to how quickly it can deliver production units into data centers. When foundry schedules and memory supply are tight, customers may face longer wait times for systems and upgrades. That can affect revenue timing and the pace at which customers can deploy AI infrastructure. Market participants typically watch whether chip designers can convert strong product demand into actual shipped quantities.

Although the coverage frames the efforts as an active exploration led by Su, it does not spell out the specific structure of any arrangement. It does not disclose the names of particular Samsung business units, the stage of negotiations, whether the parties have signed any binding contract, or what process node or product families would be targeted. It also does not describe expected timelines for ramping capacity or delivering memory components tied to AMD platforms.

From a stock perspective, the market often treats new manufacturing or supply-chain partnerships as potential levers for future growth. If AMD can expand supply through additional foundry or memory pathways, the company may be better positioned to meet customer demand without relying solely on existing channels. That said, expectations can shift quickly if the exploration does not progress to signed agreements or if the operational ramp takes longer than anticipated.

Sector context is important here. The AI chip cycle has created a broad industry scramble for advanced manufacturing resources, including wafer fabrication and advanced packaging. At the same time, memory makers and suppliers have faced their own capacity constraints, especially for components optimized for AI acceleration. In that environment, arrangements that help align compute and memory availability can become as important as the chip design itself.

What remains uncertain is whether AMD will disclose concrete terms, such as projected incremental volumes, specific products, or the financial impact of any eventual foundry and memory collaboration. Until there are confirmed announcements, investors will likely focus on AMD’s subsequent guidance, supply and shipment commentary, and any official communications that clarify the scope and timing of negotiations.