NVIDIA Reportedly The First & Only Customer For TSMC’s Bleeding-Edge A16 Process Node, Utilizing For Next-Gen GPUs

NVIDIA is reportedly the only customer in line for TSMC's next-gen A16 process technology, eyeing it for future GPUs such as Feynman.

NVIDIA Feynman GPUs To Leverage TSMC's A16 (1.6nm) Process Technology

In a report by DigiTimes, it is once again claimed that NVIDIA is the first and only customer in line for TSMC's next-gen A16 process technology. This process tech will lay the foundation of NVIDIA's next-generation GPUs, such as Feynman, which will succeed the upcoming Rubin 2026 & Rubin Ultra 2027 GPU lineups.

Given the growing demand for chips within the AI sector, NVIDIA is ramping up its Blackwell Ultra chips while also prompting TSMC to speed up its P3 plant, which will play a key role in mass-producing Rubin chips. These chips are said to leverage TSMC's 3nm process technology, a step up from the 4NP tech in use by Blackwell and Blackwell Ultra chips. Last month, it was estimated that TSMC's 3nm production can hit up to 160,000 wafer units by the end of this year.

If these estimations are accurate, then it goes in line with what NVIDIA's CEO, Jensen Huang, stated during GTC US 2025. He said that next-generation Vera Rubin Superchips are all set for production in 2026, and given the speedy production, they can be sent out as early as Q3 2026.

It is understood that NVIDIA is currently the only customer for the A16 process, and the Kaohsiung P3 plant will begin mass production in 2027 in response to NVIDIA's product roadmap. Based on this, it is estimated that after Apple enters the 2-nanometer generation, the next generation process will skip the A16 and go directly to the A14.

Supply chain sources indicate that the expansion of 3nm capacity is in response to large orders from NVIDIA for products entering the 3nm generation.

via DigiTimes

But coming back to TSMC's A16 process technology, the future Feynman GPUs are the most likely candidate from NVIDIA to leverage this node.

In terms of improvement, TSMC's A16 node offers 8-10% speed improvement, 15-20% power reduction, and 7-10% higher chip density versus the N2P node. So NVIDIA, coming from N3P (Rubin) to A16 (Feynman), will see a major upgrade from the process tech alone.

A16 also utilises Nanosheet with SPR (Super Power Rail), offering improved backside power delivery, and curated for AI/HPC markets. The process is all set for production by the second half of 2026.

Being the first customer for A16, TSMC may not only give NVIDIA early access to its A16 node, but early supply can also be dedicated to NVIDIA. NVIDIA's and TSMC's relationship has only grown thanks to the AI craze. The companies have celebrated the first-ever Blackwell wafer produced on US soil recently, & are deep in talks to further heighten their relationship to tackle supply constraints within the AI segment.

So far, NVIDIA has had a clean sweep of the AI segment thanks to its future thinking and planning, but more and more competition is lining up against it from the likes of AMD, Microsoft, Google, and others. It will be interesting to see how the segment develops within this decade, as it will give a clear indication of whether AI will be a short-lived scenario or something that has long-term implications on the tech industry.

NVIDIA Data Center / AI GPU Roadmap

GPU Codename	Feynman	Rubin (Ultra)	Rubin	Blackwell (Ultra)	Blackwell	Hopper	Ampere	Volta	Pascal
GPU Family	GF200?	GR300?	GR200?	GB300	GB200/GB100	GH200/GH100	GA100	GV100	GP100
GPU SKU	F200?	R300?	R200?	B300	B100/B200	H100/H200	A100	V100	P100
Process Tech	TSMC A16?	TSMC N2P?	TSMC N3P?	TSMC 4NP	TSMC 4NP	TSMC 5nm	TSMC 7nm	TSMC 12nm	TSMC 16nm
CPU	Rosa	Vera	Vera	Grace	Grace	Grace	N/A	N/A	N/A
Memory	HBM4e/HBM5?	HBM4	HBM4	HBM3e	HBM3e	HBM2e/HBM3/HBM3e	HBM2e	HBM2	HBM2
Launch	2028	2027	2026	2025	2024	2022-2024	2020-2022	2018	2016

Follow Wccftech on Google to get more of our news coverage in your feeds.