Why the RTX 50 Series Is the Most Disappointing NVIDIA GPU Generation Ever

NVIDIA’s latest GPU architecture, Blackwell, is impressive when it comes to the company’s data center product stack. But over on the gaming side of the street, things aren’t as rosy. In fact, RTX 50 GPUs are so underwhelming that I consider RTX 50 the most disappointing NVIDIA GPU generation ever.

Grossly Disappointing Performance Compared to the 40 Series

NVIDIA has always delivered the goods regarding performance uplift over the previous GPU generation. For decades, each new GTX, and later RTX, graphics card lineup would blow the previous generation out of the water.

We had some proper duds that barely moved the gaming performance needle, like the GeForce 9000 series. But that was a simple refresh of the Tesla architecture that debuted on the GTX 8000 series. Aside from the GTX 9800 GTX and the 9800 GTX+, GeForce 9000 GPUs were pretty much the same in terms of gaming performance compared to their GTX 8000 counterparts.

Even the universally underwhelming generations, like the RTX 20, delivered where it mattered. With the RTX 20, we got DLSS, and while the RTX 2070 didn’t surpass the previous generation’s halo product, the legendary GTX 1080 Ti, the RTX 2080 easily hopped over the previous flagship in gaming performance.

Then came the RTX 40 series, and something was up. While the RTX 4090 was a beast, other family members offered somewhat disappointing performance uplifts reminiscent of the RTX 20 series while rocking massive price bumps compared to their RTX 30 equivalents.

If the RTX 40 series was a disappointment, the RTX 50 lineup is a flat-out disaster regarding performance improvements over the previous generation. The only bright spot is the RTX 5090, with the rest of the stack offering barely any uplift over their predecessors.

The RTX 5080 is only about 12-15% faster than the RTX 4080 and miles behind the RTX 4090, something unheard of in the past when the XX80-series GPU would always beat the previous generation flagship. The RTX 5070 Ti surpasses the RTX 4070 Ti by less than 20%.

The RTX 5070, which does come with a pretty solid 23-25% performance boost over its predecessor—but is only about 6% faster than the RTX 4070 Super—is still seen as a disappointment thanks to NVIDIA’s misleading marketing claiming it’s as fast as the RTX 5090, the same 12GB memory buffer found the RTX 4070, and poor availability, which leads me to my next point.

Availability Problems Persist

Not being able to find new gaming GPUs at launch is completely normal. Even a decade ago, you’d have a hard time picking up a GPU a few weeks after it debuted. But about a month and a half after it landed, the RTX 5090 is still in dismally short supply. If you somehow find it, be ready to spend the equivalent of a decent used car.

The RTX 5080, which launched the same day as the RTX 5090, is intermittently in stock, both on Newegg and Best Buy. But you can only get expensive OC models, which command prices of at least a couple of hundred dollars over the card’s $1,000 MSRP.

The RTX 5070 Ti launched just a week after the two flagship options but it, too, is rare to find in the wild. And if you do manage to spot one, you’ll, again, have to pay a chunky premium over its $750 MSRP. Lastly, the RTX 5070 was only released a few weeks ago. So, while it’s normal not to find it in stock, I reckon it, too, will continue to be hard to find in the upcoming weeks.

Sales data confirms that RTX 50 cards are hard to get by. According to NVIDIA, via The Verge, the RTX 50 series shipped twice as many units compared to the RTX 40 series during the first five weeks of availability, but there’s a catch.

While we saw launches of four different RTX 50 SKUs over the last five weeks, the only RTX 40 GPU available during the same period back in 2022 was the halo model, the RTX 4090.

In a nutshell, NVIDIA claims that it took four RTX 50 models to double the shipping numbers of the RTX 4090, which didn’t really fly off the shelves due to its $1,600 MSRP. At least you could get the RTX 4090 at or near MSRP for a while. Good luck managing to snag any of the RTX 50 cards at MSRP.

Bogus MSRP

It looks like GPU MSRP is worthless these days. None of the MSRP RTX 50 models are currently in stock, and it looks like they won’t reappear on shelves in the future.

If you go to ASUS’s online store, you will see that you cannot even find an RTX 5080 model at its $1,000 MSRP. The most affordable option, the RTX 5080 Prime, Prime being ASUS’s “MSRP” series of cards, costs precisely $1,264.99 at the time of writing. MSI doesn’t even sell its MSRP Shadow RTX 5080 model anymore. The only Shadow model you can find is the pricier OC model that retails for $1,249.99.

The situation with RTX 5090 cards is even worse, with the most affordable models costing hundreds of dollars over the card’s $2,000 asking price (if you can find one).

The RTX 5070 Ti and RTX 5070 can occasionally be found at MSRP if you’re quick enough to order one before bots and scalpers get them, or if you’re lucky to have a Micro Center near you. But considering what happened to the MSRP RTX 5090 and RTX 5080 cards, I wouldn’t be surprised if we soon see MSRP RTX 5070 Ti and RTX 5070 listings disappearing for good.

Lastly, it would be best if I didn’t even talk about the situation in Europe, Australia, and other markets, where MSRPs are even higher, and the MSRP cards are nowhere to be found as well.

Budget and Mid-Range GPUs Stuck With 8GB and 12GB of Memory

While NVIDIA did equip the RTX 5070 Ti with 16GB of memory, the RTX 5070 is stuck with a 12GB memory buffer that slowly becomes too little for gaming at 1440p and high details. Turn on ray tracing in modern AAA games, and you’ll nearly max out the memory usage. It looks like the RTX 5070 will have a similar fate to the RTX 3070, whose 8GB of memory was too little for gaming with high or ultra visuals shortly after the card came out.

The situation with the RTX 5060 looks even more dire. Kopite7kimi, a well-known leaker with a near-flawless track record, recently shared the RTX 5060 specs. It looks like the card will have only 8GB of memory, which is unacceptable in 2025 unless we’re talking about a sub-$200 GPU, which the RTX 5060 certainly won’t be. Remember, 8GB of memory debuted on the GTX 1070, which launched almost nine years ago!

My first-hand experience is that 8GB of memory isn’t enough even for 1080p gaming with max texture quality, even without ray tracing. I had to drop textures to medium in Horizon Forbidden West when streaming the game to my ASUS ROG Ally to avoid the massive performance drop once the memory buffer fills up after about 20 minutes of playing the game with textures set to high. I’ve had similar experiences in Stalker 2 and Forza Horizon 5, so if RTX 5060 debuts with 8GB of memory, it will be DOA.

On the other hand, an unconfirmed retail listing leak claims the card will come with 12GB of memory. I hope that Kopite7kimi is wrong and that the RTX 5060 will indeed pack 12GB of memory. Otherwise, anyone who ends up buying it will be in for a nasty surprise.

Missing Render Output Units and Melting Power Cables

The last two nails in the RTX 50 coffin are the missing ROPs (render output units) and the melting power cable debacle.

Some RTX 5090, RTX 5080, and RTX 5070 Ti cards are defective in that they lack a certain number of ROPs, making the affected cards up to 10% slower in games, according to Gamers Nexus.

This is a grave oversight by NVIDIA; the company somehow failed to notice the issue during the QA process. The number of affected cards is small, but that doesn’t reduce the gravity of this fiasco. To add insult to injury, while NVIDIA claims no mobile RTX 50 GPUs are affected, JarrodTech’s (one of the best laptop reviewers out there) sources claim that NVIDIA asked them to double-check their laptops.

Luckily, it looks like the RTX 50 mobile GPUs do not suffer from missing ROPs. Still, the fact that NVIDIA asked manufacturers to double-check the laptops doesn’t paint a pretty picture regarding its QA process.

The second flop includes melting power cables on some RTX 5090 and RTX 5080 graphics cards. Now, the 12VHPWR power connector NVIDIA used on RTX 40 GPUs was already pretty darn bad, but the 12V-2×6 connector that replaced it is even worse.

Power cables will continue to melt unless NVIDIA replaces the 12V-2×6 connector or redesigns RTX 50 circuit boards to include a better fuse protection solution. At the moment, everything points out that NVIDIA will do neither.

The sole silver lining might be that RTX 5090 cards have been nowhere to be found for weeks now. There’s a chance the company is trying to solve the issue under the radar, and that the future RTX 5090 cards won’t come with shoddy power connectors.

As for myself, I’ve gotten an AMD RX 9070 XT instead of an RTX 5070 Ti. The AMD card offers more value along with an excellent upscaling solution in the form of FSR 4, which finally matches DLSS in picture quality.

If NVIDIA delivers more bang for the buck compared to AMD with its next generation of gaming GPUs (I seriously doubt this), I’ll switch back because I hold no love for any brand. But the fact of the matter is that the RTX 50 series is a massive disappointment, the biggest from NVIDIA to date, as far as I’m concerned.