AMD is in talks with Samsung Electronics to potentially manufacture part of its upcoming chips using the South Korean company’s 2-nanometer process technology. This information, reported by Korean outlet EDaily and echoed by various industry publications, comes at a time when the surge in demand for CPUs, GPUs, and AI accelerators is stressing the manufacturing capacity of leading foundries.
Neither AMD nor Samsung has officially confirmed this cooperation, so it should be regarded as a negotiation in progress rather than a finalized contract. Nonetheless, it reflects a growing trend: major chip designers can no longer rely solely on a single advanced manufacturing partner if they want to sustain the rapid growth driven by AI.
Sources in South Korea indicate that discussions accelerated after AMD CEO Lisa Su’s visit to Korea in March, during which she toured Samsung’s facilities in Pyeongtaek. That same month, AMD and Samsung signed a memorandum to strengthen collaboration on advanced memory for AI, including HBM4 for future Instinct GPUs and DRAM solutions for the sixth-generation EPYC processors, codenamed Venice.
AI is also increasing pressure on CPUs
Over the past two years, the conversation about AI infrastructure has been dominated by GPUs. NVIDIA has set the pace with AMD expanding its Instinct family, and hyperscalers competing to secure accelerators, HBM memory, high-speed networks, and power capacity. But as AI inference and agent-based AI become more prominent, CPUs are regaining importance.
An AI system running autonomous agents does much more than process isolated requests. It manages workflows, calls external tools, transfers data, handles services, controls queues, interacts with databases, orchestrates accelerators, and sustains long-running processes. GPUs handle intensive calculations, but CPUs serve as the intelligent core organizing much of the platform.
AMD’s recent results highlight this shift. The company reported $10.253 billion in revenue for Q1 2026, a 38% year-over-year increase. Its data center division reached $5.8 billion, up 57%, driven by strong demand for EPYC processors and Instinct GPUs. CEO Lisa Su stated that AI inference and autonomous AI are fueling demand for high-performance CPUs and accelerators.
This growth necessitates greater manufacturing capacity. AMD’s recent success heavily relies on TSMC for both CPUs and advanced GPUs. Notably, the EPYC Venice chip was the first HPC product manufactured on TSMC’s 2 nm N2 node in April 2025. This milestone strengthened AMD’s partnership with TSMC but also underscored how future products depend on highly sought-after node technologies.
Samsung aims to re-establish itself as a serious alternative in foundry
Samsung Foundry has been working for years to elevate its status as a competitor to TSMC in advanced semiconductor manufacturing. The company adopted Gate-All-Around transistors with 3 nm technology but faced concerns about performance, customer adoption, and process efficiency. The move to 2 nm presents an opportunity to reposition itself, provided Samsung improves yields, design ecosystem, capacity, and customer confidence.
Samsung promotes its SF2 process as an evolution of its GAA technology, offering improvements in performance, power efficiency, and chip size over previous generations. It also works on compatibility between SF2 and SF3 processes, along with developing high-density libraries, IP, and solutions tailored for high-performance computing markets—where fast I/O, advanced memory, and packaging are as crucial as the node itself.
The potential entry of AMD as a customer would carry significant symbolic value. Samsung already supplies chips for Tesla, including AI5 and AI6, and the Taylor, Texas facility is preparing to produce AI chips with 2 nm processes. Gaining a share of AMD’s future production would secure a high-profile client in data centers and enhance Samsung’s position as an alternative to TSMC for AI workloads.
The AMD-Samsung relationship is not starting from scratch. An agreement announced in March covered HBM4 memory for AMD’s Instinct MI455X GPU and advanced DRAM for EPYC Venice and AI platforms like AMD Helios. Reuters also reported that both companies are exploring foundry collaborations, adding context to the recent reports from Korea.
Diversifying doesn’t mean abandoning TSMC
Any potential 2 nm production partnership with Samsung should not be seen as AMD breaking ties with TSMC. Rather, it would be a strategic move to secure dual supply sources. TSMC remains the leader in advanced nodes and HPC packaging, as evidenced by AMD’s recent use of TSMC’s N2 for Venice. However, TSMC’s capacity is under immense pressure from Apple, NVIDIA, AMD, Qualcomm, MediaTek, Tesla, and major AI clients.
Having a second source provides strategic advantages—not only in pricing but also in scheduling. If some CPUs or AI chips can be manufactured at Samsung, AMD can better respond to demand spikes, negotiate more flexibly, and mitigate risks associated with supply concentration. Scaling production across two foundries is complex; it requires adapting designs, libraries, validation, packaging, and performance tuning. Still, the benefits can outweigh the challenges if the product volume justifies it.
Samsung’s primary challenge will be demonstrating it can produce high-performance chips meeting AMD’s quality standards. In server CPUs, margins are tight, and customers expect consistent performance, efficiency, reliability, compatibility, and availability over years. Yield issues or power consumption problems could impact costs, timelines, and confidence.
For AMD, this presents an opportunity to broaden its supply chain amid increasing demand from AI-driven data center applications. Their portfolio now extends beyond EPYC, encompassing Instinct accelerators, networking, memory, and key cloud partnerships. As autonomous AI becomes more prevalent, server CPU demand may grow alongside GPUs. Meeting this demand will require more wafers, HBM, advanced packaging, and industrial flexibility.
On a geopolitical level, a partnership with Samsung might diversify supply sources geographically—producing chips in South Korea or at the Taylor plant—reducing reliance on Taiwan. For clients in the US and Europe concerned about supply resilience, this could be strategically advantageous.
Of course, the key details are still missing: official confirmation, production volumes, specific products, and schedules. Until AMD and Samsung clarify whether an agreement exists, which chips are involved, and where they will be manufactured, this remains a market signal. Yet, it’s a strong indication that the booming AI market is prompting leading semiconductor designers to seek expanded advanced manufacturing options, and Samsung aims to re-enter the top tier of global foundries.
Frequently Asked Questions
Has AMD confirmed it will manufacture 2 nm chips at Samsung?
No. Currently, it is in advanced negotiations as reported by Korean and industry sources. No official contract has been announced by AMD or Samsung regarding 2 nm manufacturing.
Why would AMD look to Samsung when it already works with TSMC?
To diversify manufacturing capacity, reduce dependence on a single foundry, and better meet the high demand for CPUs and AI chips.
What role does EPYC Venice play in this story?
Venice is AMD’s upcoming server CPU platform, which was the first HPC product announced to be built on TSMC’s N2 node, exemplifying AMD’s push for advanced process technology.
What benefits would Samsung gain from an agreement with AMD?
Samsung Foundry would boost its credibility in advanced nodes, secure a major data center client, and strengthen its competitiveness against TSMC in AI and HPC markets.