NVIDIA 7nm (Ampere) GPUs Landing Within 9 Months, Analyst Reports


NVIDIA has been awfully quiet this past year, only rolling out its premium RTX series and cruising along in the wake of its powerful 16nm-based Pascal and Turing GPUs, but it looks like the time has finally come for the company to jump ship to 7nm. According to an analyst from Susquehanna Financial Group, the company is planning to shift to 7nm within 9 months of today. This means that you can expect a 7nm graphics card launch by 2H 2020 - a timeline that has already been leaked at least 4 times before.

NVIDIA 7nm graphics cards (Ampere?) landing within 9 months - claims analyst from Susquehanna Financial Group

The GPU industry has, as a result of NVIDIA's apathy, been fairly slow not only in terms of the usual perf per $ growth, but as a function of the absolute performance growth as well. With NVIDIA's competitive incentive broken (we have AMD to thank for that, unfortunately), the company also ran afoul in its usually stellar earnings records. In fact, while NVIDIA remains on the 16nm process, AMD became the process leader by moving on to TSMC's 7nm process. Here is what the analyst had to say:

On Tuesday, Susquehanna Financial Group analyst Christopher Rolland reiterated his Positive rating for Nvidia shares, citing the strong sales for the Nintendo videogame console. Nvidia makes chips for the gaming device.

“We think Nvidia faces the most reasonable Street expectations in quite some time, with potential tailwinds from improving DC [data center], Switch [Nintendo console], and high GPU [graphics processing unit] attach rates (laptops/desktops), all in front of a litany of upcoming 7nm [nanometer] launches over the next nine months,” he wrote. Source.

NVIDIA it seems has finally decided to play catch up and will be shifting to its own 7nm by 2020. According to what we have heard so far, this will be Samsung's 7nm EUV process and should offer a significant step up in performance from previous generations (even TSMC's non-EUV based 7nm process). 9 months amounts to roughly 3 quarters, and with a launch in 2020, you will first start to feel the impact in the third-quarter earnings (exactly a year from now). In other words, NVIDIA is slowly but surely working its way back up to getting Jensen his coveted record quarters.

What we know so far

We have previously heard of NVIDIA's Ampere GPUs when they passed their EEC certification, but  nothing more came up since then. Now, however, we have a tentative timeline: they will be launching in 2020. It is highly likely that NVIDIA will continue with their RTX philosophy and take that to the next level with Ampere. Right now, the Turing GPU is capable of raytracing at 1080p 30 fps for light to moderate path ray tracing workloads. The Ampere GPU will be able to go further.

The fact that it is based on Samsung's 7nm EUV process means we are looking at a performance advantage as well as a power efficiency advantage. Not only that, but believe it or not, 7nm EUV is actually supposed to be easier to fab than standard UV multi-patterning efforts. Think of EUV as sort of a reset of the difficulty curve as the company moves to a new light source. This will, however, require extensive re-tooling, but the economies of scale will almost certainly prove to be worth it. At a bare minimum you are looking at a 50% increase all things considered and watt for watt.

Here's the thing right, NVIDIA is one of the biggest customers of TSMC and has been their loyal patron since pretty much the dawn of modern gaming tech. If they are actually planning to shift to Samsung's 7nm technology, then that will have repercussions not only for the company, but for TSMC as well. There are two possible things that are going on here, either Samsung is offering them a better deal financially, or NVIDIA has reason to believe Samsung's 7nm tech is better. We can guess one reason for why this might be the case. Right now, TSMC's 7nm process is not based on EUV, but they do have an EUV node planned. The process that NVIDIA is planning to shift to, at Samsung's, is EUV.

That would imply that they have reason to believe that Samsung's EUV process is better positioned to help them achieve their goals than the TSMC-based one. Another potential reason is that TSMC cannot offer them a large amount of volume and will never prioritize them over the likes of Apple Inc. At the same time, AMD is using up a lot of their capacity and things are getting too cramped in there for NVIDIA's liking. Samsung's foundries on the other hand have ample capacity and considering the giant that Samsung Electronics is, can simply throw money at yield problems to make them disappear.

Everything considered, Samsung is the logical partner for a company as ambitious as NVIDIA. With a big question mark on Intel foundries' capabilities right now and TSMC hogged down, Samsung remains the only leading edge foundry for NVIDIA to tap into. GlobalFoundries dropped out of the leading edge race earlier this year - not that they would have been considered to begin with.