Nvidia’s lineup of GeForce RTX 50-Series Graphics Cards expanded this week with the arrival of the 5070 Ti, which joins the higher-end 5080 and 5090 models that launched late last month. Just like those two GPUs, the 5070 Ti is hard to find in stock at Amazon, Best Buy, Newegg, and other retailers. We’d recommend checking the links below, as restocks tend to be sporadic and rather unpredictable. The standard 5070 GPU is expected to release before the end of February, too.
Nvidia RTX 50-Series Restock Tracker
The flagship card, the RTX 5090, is priced at $2,000, and Nvidia claims it’s twice as powerful as the RTX 4090 but requires a 1,000-watt power supply–a huge leap from the RTX 4090’s 450W baseline power draw. Despite its power demands, the RTX 5090 is more compact than its predecessor, fitting into smaller PC builds.
The slimmer size and higher performance are true of all the cards in the lineup, with Nvidia claiming the $1,000 RTX 5080 to be twice as fast as the RTX 4080. The budget-priced $550 RTX 5070 offers performance comparable to the RTX 4090, at nearly a third of the price. The 5070 Ti, meanwhile, sits between the 5070 and 5080 in terms of price ($750) and performance.
Rather than relying on pure processing power to achieve these benchmarks, the new 50-Series cards leverage Nvidia’s latest machine learning technology, including RTX Neural Shaders and the latest DLSS 4 upscaling. Not only does DLSS 4 allow for upscaling lower-resolution gameplay to 1440p or 4K resolutions, it also includes Multi Frame Generation, which Nvidia says is capable of generating up to three additional frames for every one frame of rendered gameplay. It’s essentially doing for gaming what frame smoothing does for movies and TVs and creating the appearance of higher, smoother frame rates without the hardware actually rendering them all.
So, while the 50-Series GPUs’ actual hardware performance uplift is technically lower in terms of raw processing power without the aid of AI technology–the 5070 has less processing power than the 4090, for instance–it also means the GPUs are smaller and more affordable compared to their previous-gen counterparts since you don’t need as much horsepower to achieve high-end results.
Of course, AI upscaling and frame generation have tradeoffs compared to rendering everything with hardware. These machine learning technologies can potentially match hardware rendering the quantity of performance, but the issue is the quality of that performance. Upscalers like DLSS, AMD’s FSR, and Sony’s PSSR on PS5 Pro can be inconsistent (or even incorrect, in extreme cases), while frame generation can introduce noticeable input delay–something that playing at high frame rates is intended to alleviate, not exacerbate.
To be fair, these are issues Nvidia–and game developers–are aware of and actively working to address. Besides, plenty of games already utilize the features to great results, and the 40-series cards used DLSS to achieve big performance gains over the 30-series–unfortunately, it was also incredibly expensive. If the next generation of the technology works as intended, the 5090 could enable another massive performance boost, while the mid-range RTX 50 GPUs could offer a much more affordable option for achieving 4K resolution gameplay and high FPS.
RTX 50-Series laptop GPUs are on the way as well, with the first models expected to launch in March.