By now, Nvidia's new RTX 50 series GPUs might be all over your feeds. People are in awe of the massive performance leap Nvidia showcased and the price cuts that came out of nowhere. However, there have also been a lot of less-than-exuberant reactions to CEO Jensen Huang's CES keynote due to the fact that it didn't come across as fully transparent.

In many ways, the RTX 5000 announcement was very different from anything the company has done before. Perhaps it's because these new graphics cards are positioned differently from any of Team Green's previous offerings. And therein lies the source of this undercurrent of misleading claims and unfair comparisons by the company — something that I'll try to dissect in this article.

5 Nvidia is hiding the raw performance of the 50 series

No, the RTX 5070 isn't magically the new RTX 4090

Of all the things Nvidia claimed in its CES keynote, perhaps the one that caught the most attention was the new RTX 5070 (priced at $549) offering RTX 4090 levels of performance — a $1,599 GPU that almost always sold for much higher than that. Although Jensen clearly mentioned that this feat was only possible thanks to the full power of the Blackwell hardware and DLSS 4, especially Multi Frame Generation, with the latter being exclusive to the RTX 50 series, there's more to it than meets the eye.

Nvidia's own performance comparison between the RTX 50 and RTX 40 series GPUs conveniently omits any rasterization-only, non-upscaled scenarios. The games the company used to showcase the huge performance gains on the new generation of cards, such as Cyberpunk 2077, Alan Wake 2, A Plague Tale Requiem, and Far Cry 6, were all running with ray tracing and DLSS turned on.

Consumers will have to wait for third-party reviews to see the raw performance uplift of the Blackwell architecture compared to Ada Lovelace. I'm not too confident about those because the specs of the RTX 5070, seen head-to-head against those of the RTX 4090, make it abundantly clear that the former is a far cry from the RTX 40 series flagship. Nvidia hasn't suddenly put a massively improved GPU die inside the RTX 5070 — compared to the top-end RTX 5000 SKUs as well as the RTX 4090; it's a heavily cut-down product.

In all likelihood, we will see the RTX 5070 perform way worse than the RTX 4090, as a next-gen 70-class usually does. Similarly, without the AI-generated frames, thanks to Multi Frame Generation, none of the other SKUs that Nvidia announced will be as impressive as the company is making it seem in the charts it currently has on its website. I'm not saying AI frames are "fake" or anything of the sort, but Nvidia isn't being as transparent as some of us would like.

4 Nvidia isn't making the right comparisons

The charts on the website are a bit misleading

Even if you keep aside the omission of the rasterized performance of the RTX 50 series, Nvidia isn't really making fair comparisons with the RTX 40 series cards. For instance, the performance charts on the company's website pit the RTX 5080 against the RTX 4080, which might seem like the obvious thing to do at first glance. The RTX 4080 was launched at $1,200, whereas the new RTX 5080 offers double the performance at $999 — a sweet deal, right?

Not so fast, since the more accurate comparison should be the RTX 5080 vs. RTX 4080 Super. The refreshed Super card was essentially the same RTX 4080 but with a reduced price of $999. More importantly, however, the specs of the RTX 5080, when viewed against those of the RTX 4080 Super, are hardly impressive — a mere 5% increase in CUDA cores and ray tracing cores.

The same story plays out when you compare the RTX 5070 Ti to the RTX 4070 Ti Super instead of the RTX 4070 Ti. And going further down the stack, the RTX 5070 actually falls behind in specs to the RTX 4070 Super. Now, I know the new Blackwell architecture, coupled with the faster GDDR7 memory and slightly higher clock speeds, should also be taken into consideration, but these improvements can only do so much.

All the tall performance claims Nvidia has made are on the back of the AI-heavy DLSS 4 software improvements, but as I mentioned before, the raw performance of the latest cards is still shrouded in mystery.

3 The VRAM problem hasn't gone anywhere

Ask, and you shall never receive

Gamers and reviewers have been clamoring for Nvidia to put enough VRAM on its GPUs for ages, but things don't seem to be moving in the right direction. There have been multiple instances of Nvidia GPUs running out of VRAM in modern games, such as The Last of Us, Hogwarts Legacy, Resident Evil 4, and Dead Space. Nvidia, however, has been starving gaming GPUs of VRAM since the RTX 30 series.

Having less VRAM on a budget GPU is one thing, but an $800 RTX 4070 Ti outfitted with 12GB VRAM is borderline robbery. With the RTX 50 series, Nvidia is doing the same thing again, offering only 12GB of VRAM on the RTX 5070 and 16GB on the RTX 5070 Ti and RTX 5080. Only the flagship RTX 5090 has received a VRAM bump compared to its predecessor.

It seems Nvidia is planning to offer enough VRAM only on the RTX 50 Super refresh, which will be accompanied by higher prices, of course. This artificial differentiation forces people to spend more than they really should on a new gaming GPU and consistently reflects Nvidia's policies toward its desktop GPU vertical.

2 Prices are less than expected but still bad

It's too soon to be impressed by the price cuts

If you watched the RTX 5000 announcement or any of the videos about it later, you might have been pleasantly surprised by the prices of the four newly announced SKUs. While nearly everyone, myself included, was predicting a $1,200–$1,500 MSRP for the RTX 5080, Nvidia went with a "more affordable" $999. The RTX 5090 was bang-on at $2,000, but the RTX 5070 Ti and RTX 5070 at $749 and $549 were surprising, to say the least.

Some speculate the company simply leaked higher figures beforehand just to make the actual prices seem more digestible. Whether you believe that or not, the fact is that Nvidia has conditioned the average consumer to believe that a $999 80-class GPU should be considered "cheap." Furthermore, any price reduction on the RTX 50 series GPUs should be seen in light of the raw performance gains one is getting, which is still unclear.

Besides, the RTX 5080, even for a grand, doesn't offer more than 16GB of VRAM or even a big spec jump over the RTX 4080 Super. The massive performance gulf between the RTX 5080 and RTX 5090 will undoubtedly force some consumers to shell out the premium for the flagship card with double the CUDA cores, ray tracing cores, and VRAM.

If you're not too keen on spending $999 or even $549 on an RTX 5000 GPU, I'd suggest waiting for AMD's official RX 9000 announcement and deciding after analyzing what Team Red has cooked up this generation.

1 Nvidia finally makes it all about AI

The AI-pocalypse is here

For the first time, Nvidia made a GPU series announcement all about the wondrous AI capabilities it has been working on. The company proudly displayed the AI TOPS numbers next to the prices of each of the four new RTX 5000 SKUs in its keynote, conveying to the average consumer that AI-powered performance will take center stage from this point forward.

As expected, CEO Jensen Huang's keynote was all about the company's latest AI developments, but even the short section on the gaming GPUs was completely dominated by the AI-driven Multi Frame Generation, neural shaders, and TOPS numbers. Nvidia is fully embracing the fact that any gen-on-gen improvements to its graphics cards will be born out of AI advancements rather than brute-force hardware leaps.

Perhaps that was the natural progression we were always heading toward, as Moore's Law became less and less a "law," but not everyone would be as impressed by this new normal as Nvidia would like them to be. Using AI to boost gaming performance isn't bad at all, but gamers have the right to be miffed about getting sub-30 FPS performance at 4K on $2,000 GPUs, which only, when upscaled and bolstered by AI-generated frames, can cross the 60, 100, or 200 FPS thresholds.

Let the hype settle down

As always, you should wait for third-party reviews and benchmarks to make your opinions and decisions about the RTX 50 series graphics cards. The review embargo should be lifted in a couple of weeks, and you'll be able to decide whether the raw performance of these new GPUs is worth your money or not. Regardless, "AI-powered frames" are the direction in which Nvidia is heading. Whether that's good or bad for the industry is something only time will tell.

AMD's RX 9000 cards were absent from CES, but the company is expected to announce them soon at a separate event. And if the price and performance leaks regarding the RX 9070 and RX 9070 XT turn out to be true, Nvidia could have a real fight on its hands, at least in the mid-range segment. Waiting and watching, hence, is the recommended mood of the hour.