Nvidia's next-gen GPUs to have a node advantage over AMD's RDNA3 GPUs

Next-gen GPU rumors are coming thick and fast. The latest is that Nvidia's upcoming RTX 40 series of graphics cards will be built with TSMC's 4nm process node, giving the series a slight advantage over the 5nm process that AMD is expected to use for its next-gen RDNA3 GPUs.

Moore's Law Is Dead (via wccftech) believes that Ada Lovelace will be built using TSMC's 4nm node. Talking about nodes can be confusing. It was thought that Nvidia would use a node referred to as 4N, which is a customized 5nm class process, however MLID specifically states '4nm'. He knows what he's talking about.

Nvidia's workstation 'Hopper' architecture is a 4nm product, but that's a very high margin product, and Nvidia can easily recoup costs from sales to secure bleeding edge wafer starts. We know Nvidia is paying a lot of money for access to TSMC's production, but is the asking price for 4nm too high for a lower margin GeForce card?

Confused? So are we. Whether Ada Lovelace ends up being a 5nm or 4nm or 4N or whatever, the vast majority of consumers won't pay any attention. As long as their chosen card performs as they expect, the node in use won't matter.

Whether this means Nvidia really will have a single nanometre advantage over AMD's competing products remains to be seen, but in the end, it might not make all that much difference. Rumours point towards Ada Lovelace being very power hungry, meaning Nvidia isn't using the slight power headroom advantage it might have enjoyed as it seeks maximum performance.

There's also the possibility that AMD's chiplets design could utilize different nodes, making node comparisons a largely meaningless endeavor. Its cards are also expected to take a step up in power consumption.

All will become clear in time. RTX-40 series cards are expected to launch sometime in the second half of 2022.