NVIDIA to Become TSMC’s First A16 (1.6nm) Customer Amidst AMD’s Drive for Advanced AI GPUs

Recent reports indicate that NVIDIA will become the first customer to utilize TSMC’s advanced A16 node. This decision represents a significant pivot for NVIDIA, which has historically leaned on older process technologies from TSMC.

NVIDIA’s Strategic Shift: Embracing TSMC’s A16 Node in a Competitive AI Landscape

NVIDIA is poised to enhance the performance capabilities of its upcoming AI architectures by leveraging TSMC’s state-of-the-art manufacturing processes. Previously, the company was not among the initial adopters of TSMC’s latest nodes, a privilege typically reserved for industry leaders like Apple, MediaTek, and Qualcomm. However, as indicated in a report by Ctee, NVIDIA’s approach is about to change dramatically, with plans to be the first firm to implement the A16 process. This new node is expected to deliver substantial performance increases while adhering to the principles of Moore’s Law.

Historically, NVIDIA had not been the pioneer in utilizing TSMC’s latest technological advancements since the 110nm node era. Instead, the company has prioritized architectural developments to boost performance. The introduction of the A16 node is being heralded as a revolutionary step for TSMC, thanks largely to innovations like Gate-All-Around FETs (GAAFET) and Super Power Rail (SPR) backside power delivery. Among these, the Significant Feature Reductions (SFR) are particularly noteworthy, promising to further elevate computing efficiency. However, further discussion on this topic will follow at a later date.

TSMC A16™ slide detailing nanosheet technology with power improvements, aimed at datacenter AI, targeting production in 2H26. — Image Credits: TSMC

We anticipate that NVIDIA could begin utilizing the A16 node in late 2027 or early 2028, contingent on TSMC’s timeline for high-volume manufacturing, projected for late 2026. NVIDIA’s roadmap suggests that this technology might debut with either the Rubin Ultra or Feynman GPU series, with a stronger likelihood for Feynman to unveil more innovations as well. This development is promising news for TSMC, as the adoption of its high-end manufacturing nodes by leading AI hardware producers will significantly bolster revenue from this generation of processes.

Source & Images