The RTX 50-series was a disaster at launch. Offering only minor performance gains over the previous generation, and combined with the extreme supply shortage at launch and the fact that most of the GPUs in the lineup were plagued by severe technical issues, the Blackwell GPUs ended up being a disappointment and Nvidia’s worst launch to date.
Now, nearly 16 months after the release of the RTX 50-series, the situation has only worsened. The ongoing memory shortage and PC hardware price inflation have deepened Nvidia’s woes in the consumer GPU space. If you’re planning on buying any GPU from the RTX 50-series, I strongly urge you to reconsider.
The Blackwell lineup is faulty
Factory fumble
Credit: Gavin Phillips / MakeUseOf
One of the biggest reasons why the RTX 50 GPUs are so controversial is that they were faulty out of the box. There have been several reports, including this detailed analysis by Tom’s Hardware, of cards like RTX 5070, RTX 5080, and RTX 5090 being shipped with fewer ROPs (Render Output Units), which are physical components on a GPU responsible for the final image output. This manufacturing flaw reduced performance on affected Blackwell GPUs by around 4%.
Not to mention that the beefy 12V-2×6 connector also started melting and, oftentimes, fused to the GPU itself. Some cases were more severe than others, in which the GPU itself caught literal fire. Unfortunately, the list doesn’t end here, and users reported driver issues, PCIe 5.0 stability issues, and more, but most of the software-side errors are now fixed.
Nvidia’s performance claims for the Blackwell GPUs fell short
Sorry, but I’m not sold on fake frames
Credit: Make Use Of
The RTX 50-series wasn’t what most had expected in terms of a generational leap. Nvidia’s main selling point for these cards was multi-frame generation with DLSS 4, which introduces frame multipliers by 3-4X. This isn’t pure performance, and I am definitely not a fan of fake frames generated via AI that introduce other problems like ghosting, input latency, and artifacts.
To show what I mean, Nvidia CEO Jensen Huang claimed that the RTX 5070 would beat the RTX 4090, but that turned out to be a hoax. This was true only with frame generation; otherwise, the RTX 4090 is vastly superior, being better at rasterized workloads without frame generation, has 12GB more VRAM, and nearly thrice the CUDA cores of the RTX 5070. Here’s a video from Digital Foundry comparing both cards in depth.
Besides any pure performance gain, the next iteration of Nvidia’s upscaling technology, the DLSS 5.0, exclusive to the RTX 50-series cards, is also launching later this year. The initial reception hasn’t been great, and the technology almost feels like an AI slop generator, completely altering the visual and artistic direction intended by the game developers. That’s another reason not to buy the Blackwell cards.
GPU prices have only gone higher since
The RTX 5090 is currently going for nearly twice its MSRP
PC hardware prices have soared beyond comprehension, and in times like these, buying used components seems like the right choice. The RTX 50-series cards were already quite expensive at launch, and the flagship RTX 5090 launched at a whopping $2,000. That said, I have rarely seen the Blackwell cards at their MSRP, first due to the supply shortage at launch, and now, the memory fiasco.
GDDR7 memory is one of the most sought-after DRAMs, which is why GPU prices have also seen a significant increase in recent times. The RTX 5090 is going for nearly $3500 at the moment, and with the current momentum, I foresee it touching the $5000 mark in the coming weeks with another wave of shortages as well. With such high prices, scammers are also listing fake RTX 5090s, so be wary of anything that looks too good to be true, because most of the time, it is.
There may not be a Super refresh release this time
Nvidia isn’t getting a second chance to amend its mistakes
Image by Jayric Maning –no attributions requiredCredit: Jayric Maning / MakeUseOf
It’s pretty clear that Nvidia has begun prioritizing enterprises over consumers, and amidst the AI-led memory shortage, a new report from The Information claims that Nvidia has postponed any plans for an RTX 50-series Super refresh this year. Even if Nvidia were to release the RTX 50 Super lineup, it won’t happen until 2027. It makes sense, though, prices for GPUs are already high, and with the current RTX 50-series already selling at such high rates, the Super refresh would be a mistake and would undoubtedly launch at absurd prices.
Generally, with the Super refresh, Nvidia gets the chance to iron out the kinks with the base series, but unfortunately, that won’t be happening this time around. No Super refresh also means that prices for the base RTX 50-series won’t go down either.
AMD just won the GPU war
They outdid themselves with the RX 9000 GPUs
Amir Bohlooli / MUOCredit: Amir Bohlooli / MUO
AMD had been out of commission until the RDNA 4 GPUs were released, marking its comeback in the graphics card space. Now, AMD didn’t target the higher-end market with the RX 9000 lineup; instead, it focused on the lower- to mid-end market and won over the masses, including me. Hence, there are only three desktop RDNA 4 GPUs: RX 9060 XT, RX 9070, and RX 9070 XT.
Specifically, the RX 9070 XT was a massive success, and the single model alone outsold all past Radeon GPUs by 10 times. If you’re looking for 1440p gaming, I highly recommend the RX 9070 XT as the go-to option, as it features 16GB of VRAM, 4096 cores, and 128 ROPs. The RX 9070 XT still goes for around its original MSRP of $650–700 and is almost on par with the RTX 5070 Ti, which is selling for nearly double the price at $1100.
For those with a lower budget, the RX 9060 XT, specifically the 16GB model, is a great option — delivering just 5-10% lower performance than the $450 RTX while costing $100 less. That said, I don’t advise buying the 8GB option since it is no longer viable for modern AAA games.
Waiting for better GPUs isn’t exactly smart
At the current pace, newer hardware is evidently getting delayed unless manufacturers don’t want to launch at skyrocketing prices. I expect the next generation of RTX GPUs to also be delayed, and even if the AI-led memory debacle were to end today, the repercussions would still be felt for a little longer. While I recommend skipping the RTX 50-series cards, you can opt for the RDNA 4 GPUs. However, if you’re looking for something more powerful, the RDNA 4 lineup doesn’t include higher-end models — in which case, you can go for the RTX 4090, the 2nd-fastest GPU right now. As I mentioned earlier, Nvidia’s benchmarks claim Blackwell cards are better than the RTX 4090, but that’s based on multi-frame generation. In terms of raw performance, it outperforms everything except the RTX 5090.
Brand
AMD
Interface
PCIe 5.0 x16
Memory
16GB GDDR6
Power
304 W
AMD’s RX 9070 XT is a high-end GPU from the RDNA 4 family with 16GB of VRAM, 128 ROPs, 4096 stream processors, and 64 ray tracing cores, making it a capable GPU ideal for 1440p gaming. The RX 9070 XT is considered the best GPU due to its price-to-performance ratio and is generally also superior to its competitor, the RTX 5070 TI, for the same reason.
