For the first time in its history of making gaming GPUs, Nvidia hasn’t just dropped the ball, but embarrassingly fumbled it, watched it roll into a drain, and then trodden on a rake for an encore. If there’s one word you could use to describe Nvidia’s approach to gaming GPUs over the last 25 years, it’s “relentless,” but in 2025 you could replace that with “complacent.”
Until this year, Nvidia rarely rested on its laurels, and even when it made mistakes (I’m looking at you, GeForce FX and Fermi) they were at least made as a result of big architectural changes to GPUs, rather than complacency. But Nvidia’s latest gaming GPUs look like the result of a company that thinks it can coast along without shaking up its formula.
Of course, I understand why gaming isn’t a top priority for Nvidia right now. Jensen Huang is the businessman of the moment in his fancy leather jackets, the company is literally worth trillions, and the AI revolution is basically an industrial-scale money printer for Nvidia’s shareholders. The upshot of this laser focus on AI, though, is that Nvidia has taken its foot off the gas in the gaming world, not just in terms of the core tech for the RTX 5000 GPUs, but also in terms of quality control, manufacturing, and supply.
Supply of all Nvidia GeForce RTX 5000 GPUs is, for all intents and purposes, a work of fiction right now, particularly when it comes to RTX 5090 stock. My chats behind the scenes with (large) retailers tell a story of tiny deliveries of the new GPUs, often several weeks apart, with some of them still not turning up until the actual day of the launch. Demand is also high, of course, but supply is so far from where it should be that it’s laughable. Comparably, Radeon RX 9070 XT stock is larger by orders of magnitude, and is still selling out incredibly quickly.
Meanwhile, RTX 5090, 5080, and 5070 Ti cards are turning up with some of their ROPs not working, resulting in significant impacts on frame rates in some games, prompting Nvidia to request customers to send back faulty graphics cards for replacement. Then there are the horror stories of extremely high currents being pumped through the single 16-pin power connector on the RTX 5090, along with cables melting and power supply sockets being burned out. Knowledgeable PC overclocker and YouTuber Roman “De8rauer” Hartung has even described the RTX 5090’s power connection setup as “absolutely concerning,” after spotting one of the wires hitting 150°C.
But the biggest disappointment for me with this launch is that complacency – aside from the supply issues, the new GPUs aren’t terrible, they’re just tedious and unexciting. Nvidia has regularly put the competition to shame with its relentless march forward, such as introducing hardware transform and lighting with the GeForce 256, the first programmable shaders with GeForce 3, as well as ray tracing and DLSS with its RTX GPUs. Nvidia moved so quickly, and was so far ahead, that often the competition just didn’t have a chance to fight back.
Compare that to now, when the only big selling point of the new RTX 5000 series is multi frame gen, using AI to insert multiple frames between the ones genuinely rendered by the GPU to improve the frame rate. It works well if your starting frame rate is high enough, and I saw amazing results in my RTX 5090 review, but I found this tech made much less sense in my RTX 5070 review. What’s more, only a few games support it, with DLSS Override in the Nvidia App still frustratingly unsupported by most games.
Meanwhile, most of the GPUs themselves are barely any faster than their predecessors when it comes to rendering power, even in ray tracing, with the same amount of VRAM as their immediate predecessors. The only exception is the 32GB RTX 5090, and you have to pay an extra $400 over the RTX 4090’s MSRP for that bump in performance – that’s another disappointment after waiting two years for the next generation of GPUs, when you usually expect to get more performance for the same price.
What I find most baffling is that it’s Nvidia, not AMD or anyone else, that has been really pushing for advanced graphical features such as path tracing to be implemented into games, but it’s equipped its new GPUs with so little memory that they can’t actually do it. The RTX 5070 can’t even run Indiana Jones and the Great Circle at 1080p maxed out, and as I found in my RTX 5080 review, this $999 card can’t do it at 4K either. Multi frame gen can’t help if your GPU doesn’t have enough VRAM to properly render the frames in the first place.
As Intel found when it got used to the idea of churning out generations of near-identical quad-core chips over several years, if you rest on your laurels, the competition will come out fighting, and now Nvidia has a real battle on its hands. AMD has caught up with ray tracing now, and it has decent AI upscaling tech with FSR 4 as well. As I found in my AMD Radeon RX 9070 XT review, this new GPU is so much more powerful than the RTX 5070 that Nvidia should be ashamed.
Frustratingly, it almost feels pointless to criticize Nvidia for the GeForce RTX 5000 launch, as anything I say can be answered by pointing to Nvidia’s net worth right now. Nvidia didn’t need the RTX 5000 launch to go smoothly, it could have released a range of Nvidia-branded refried beans instead of gaming GPUs for this generation, and it would still be worth trillions. I’d like to advise sticking it to the man and voting with your wallet, but that’s hardly possible when there aren’t even any GPUs to buy – it looks as though Nvidia simply doesn’t care about PC gaming any more.
The tech world is fickle, though, and the gaming market is still huge. Nvidia bet the house on AI being enough to carry its otherwise underpowered GeForce RTX 5000 GPUs through this generation, with Jensen Huang even making the ludicrous claim that the RTX 5070 offers 4090 performance, a feat that it can pull off in relatively undemanding games that support multi frame gen, but not much else. Nvidia has the benefit of there being no competition at the high end to help it this time, as well as the huge market dominance of its brand, but it can’t rely on that forever.
Yes, there’s lots of scope for AI in neural rendering, and yes multi frame gen is a useful tool in the box, but it’s not enough to make a PC gamer choose an underpowered GPU over a more powerful one for the same price. Nvidia has taken PC gamers for granted, and the competition is now ferocious. If AMD carries on at this pace with its next-gen GPUs, and Nvidia is still napping, then RTX 6000 could be in a lot of trouble.
If you’re looking to buy a new GPU, check out our guide to the best graphics card, where we take you through all our current favorite options, if you can find stock.
You can also follow us on Google News for daily PC hardware news, reviews, and guides, or join our community Discord to stay in the know.
