What Nvidia’s AI success means for the future of its GPUs

Things could be changing massively

The RTX 4090 with the Nvidia logo next to it

You can trust PC GuideOur team of experts use a combination of independent consumer research, in-depth testing where appropriate – which will be flagged as such, and market analysis when recommending products, software and services. Find out how we test here.

Last Updated on

Nvidia is celebrating record profits as it becomes the fourth biggest company in the world, even surpassing Amazon and Google and it’s all thanks to AI. Team Green has made massive advancements outside the gaming sector thanks to generative AI, so much so that its own GPU division now accounts for a fraction of its total revenue according to Founder, President, and CEO Jensen Huang.

In a recent earnings call Nvidia announced record sales to become the fourth most valuable company in the world, but that’s not due to its best graphics cards. The key findings of the Q4 FY24 report is that revenue is up 126% to a staggering $60.9bn, of this amount, just 17% ($10.4bn) is attributed to the RTX 40 series, which includes the new launches of the RTX 4070 Super, RTX 4070 Ti Super, and the RTX 4080 Super. It also factors in Chat with RTX LLM and microservices such as the Nvidia Avatar Cloud Engine and DGX Cloud. The efforts in professional visualization are also significant as well, up a total of 105% from last year, too.

Then there’s the fact that the revenue from Nvidia’s gaming division only grew up 15% from full year revenue. In contrast, the company’s Data Center revenue full-year revenue rose by 217%, up a full 409% from the previous year. It’s clear from the data that Team Green’s seeing far stronger successes both financially and growth wise from investing in generative AI compared to making graphics cards, and this could have a serious impact on the future of its GPU division.

Gigabyte RTX 4080 Super Windforce V2

Gigabyte Geforce RTX 4080 Super Windforce V2.

GPU

AD103

CUDA Cores

10,240

VRAM

16GB GDDR6X

Bandwidth

736.3 GB/s

Memory bus width

256-bit

Base clock speed

2295 MHz

Boost clock speed

2550 MHz

Nvidia’s biggest strength isn’t its graphics cards anymore

Nvidia is arguably still best known for its GeForce graphics cards which have been a force in the computing scene since 1999 with the release of the GeForce 256. For almost 25 years, it has been a market leader in consumer hardware, and kept up momentum with the GTX and RTX series, as we’re now deep into the third-generation of latter hardware with the RTX 40 series which launched at the end of 2022. Well, the time’s coming around again, and the end of the year could see the RTX 50 series drop, or will it?

That time frame would typically be in line with when we expect the RTX 50 series, based on the historical evidence. That’s because the RTX 30 series released in September 2020 as did the RTX 20 series. That positions the latest card to release either at the tail end of Q3 or throughout Q4, as the RTX 40 series did, but does Nvidia have any incentive to release another GPU generation? From the rumors circulating, it appears as though it’s not so cut and dry. We’re going over these rumors for informational purposes.

Prominent hardware leaker channel Moore’s Law is Dead alleges that Nvidia is ultimately waiting on AMD to act before deciding whether or not to release the RTX 50 series in 2024 or to delay until 2025. An insider source claims that the Blackwell GPUs could “be ready to launch in Q4 2024 if we want it to”. This could come down to the fact that 2024 started with the launch of the RTX 40 Super series, which effectively replaced both the mid-range and high-end Ada cards, but Team Red could pre-emptive strike.

That’s because it’s been rumored that AMD RDNA 4 will be firmly in the mid-range according to prominent hardware leaker Kepler via Twitter. The post alleges that “Navi 4 lineup will not have any high-end GPUs” and to “think of it like RDNA1 or Polaris generation”. Videocardz, which often hosts prominent rumors and leaks asked for clarification, to which Kepler responded that they “heard from 3 sources”. Reportedly, then, Nvidia will hold fire unless AMD comes out of the gate with hardware that competes.

It’s an interesting play from Team Green if there’s any truth to the rumors given the most recent GPU generation. AMD’s most powerful graphics card is the RX 7900 XTX which is roughly about as powerful as the RTX 4080 Super, and lags behind the significantly more expensive RTX 4090. The two video card generations utilize different approaches with hardware, too, with the former still running GDDR6 and the latter leading with faster GDDR6X memory. Different capabilities here, but there’s sure to be some envy from the former.

ASUS TUF Radeon RX 7900 XTX OC

GPU

Navi 31 die

Stream processors

6,144

VRAM

24GB GDDR6

Bandwidth

Base clock speed

1,929 MHz

Boost clock speed (OC Mode)

2,615 MHz

Nvidia holds all the cards unless AMD strikes back

What this all means is that Nvidia doesn’t necessarily need to come out of the gate with a leading GPU generation unless it’s threatened by AMD’s latest. Team Red tends to lead in terms of value with generally cheaper cards that lag behind in direct comparison. Based on this information, we can infer that the RTX 50 series could be incredibly powerful but even more expensive, aimed purely at enthusiasts, and likely to go even further in on generative AI rather than applying purely to a gaming market.

Then there’s the fact that Nvidia’s gaming division accounts for less than 20% of its yearly revenue right now. If you look at the raw numbers of the fiscal performance, there’s little incentive for Team Green to go all in when it’s the “World Leader in Artificial Intelligence Computing” according to its website. The GPU market could become far less competitive and more expensive should Nvidia decide it’s not worth trying to compete, leaving only AMD and Intel to fill the gap.

For right now the company’s focus appears to be on creating large language models, partnering with cloud service providers and building up the Nvidia AI enterprise. Nvidia CEO Jensen Huang knows the parameters his company is working in, and if this is leading the success then there’s no reason to stop and refocus. There are vast software libraries managing huge AI workloads that can do incredible things like speech recognition and natural language processing. As for how that fits into gaming GPUs, it’s anyone’s guess.

With that said, Nvidia GPUs are high-performance and use a combination and use machine learning for their deep learning super sampling, with an AI model that’s its own platform altogether. Native performance of the chips isn’t really as it was anymore. This tech boasts more more complexity with Ada architecture’s reliance on AI software than previous generations. Nvidia’s AI platform will grow, and the future of machine learning looks to happen around the cards, instead of the video cards leading the change. Our recommendation is to wait and see how things play out.

Is RTX 50 series coming?

Nvidia has yet to announce the RTX 50 series, however, the hardware has been rumored to be released at the end of 2024.

Aleksha McLoughlin is Hardware and News Editor for PC Guide and she oversees buying guides, reviews, news, and features on site. She was previously Hardware and Affiliates Editor at VideoGamer.