AMD Shifts 4nm Chip Manufacturing from Samsung to TSMC´s Arizona Facility

AMD pivots its 4 nm chip orders from Samsung to TSMC´s Arizona plant, reinforcing TSMC´s manufacturing dominance amid ongoing semiconductor industry shifts driven by Artificial Intelligence and performance needs.

AMD has reportedly decided to redirect its 4 nm chip orders away from Samsung and toward TSMC´s state-of-the-art manufacturing facility in Arizona, United States. This is a strategic reversal from a previous move in May 2023 when AMD announced plans to shift certain 4 nm CPU production from TSMC to Samsung. The partnership with Samsung had covered the company´s SF4X process technology for EPYC server processors, Ryzen APUs, and Radeon graphics cards, forming a crucial aspect of AMD´s multi-vendor supply approach. Industry sources now point to AMD´s concerns regarding Samsung´s process stability and consistency as the primary reasons for the realignment, reportedly resulting in the cancellation of plans for mass-producing graphics chips using Samsung´s 4 nm technology.

This strategic supply shift further cements TSMC´s leading position within the advanced chip fabrication sector at a time when much of the semiconductor industry, including Artificial Intelligence computing needs, is heavily reliant on a handful of specialized foundries. While this enhances consistency and supply assurances for AMD, there are underlying geopolitical risks associated with concentrating production in a single vendor, especially in the current global and technology-political climate. AMD is also expanding its commitment to TSMC, confirming that its next-generation ´Venice´ processors, leveraging TSMC´s 2 nm process technology, have successfully completed testing in the Arizona facility and remain on track for a planned 2026 release.

Despite the setback, Samsung Electronics is making notable progress. Recent reports indicate its 2 nm (SF2) process is achieving initial production yields above 30%, exceeding industry expectations for a new node introduction. Samsung aims to stabilize yields in the latter half of 2025 and commence mass production of the Exynos 2600 mobile chip, marking a significant milestone for its foundry ambitions. Additionally, Samsung appears close to finalizing a manufacturing agreement with Qualcomm to produce the upcoming Snapdragon 8 Elite 2 processor on its advanced 2 nm platform, demonstrating the company´s ongoing efforts to remain a key player in the complex, competitive global foundry landscape.

73

Impact Score

Small language models vs large language models in enterprise artificial intelligence

The article argues that most business workflows benefit more from small, domain-specific language models than from massive general-purpose systems, especially in closed, well-defined environments. It explains how right-sized models cut cost and latency while improving reliability, and how enterprises can combine small and large models in tiered architectures.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.