Nvidia GeForce RTX 50 Series graphics cards loom on the horizon, with both RTX 5090 and 5080 landing on shelves come January 30. There’s an extra wait for RTX 5070 and 5070 Ti due February, but first, here’s a closer look at the fastest flagship Team Green has ever offered to tide you over.
With such an unassuming, unbranded outer box, you’d be forgiven for thinking this was anything but Nvidia’s top-of-the-line GPU. It’s a decidedly clever decision, though, masking what lies beneath while it’s in transit, both protecting the cargo from bumps along the way and removing extra temptation for it to unceremoniously go missing.
Sliding it open rolls out the red carpet, with GeForce RTX 5090 sat snug in a luxurious cream-coloured cardboard box. I’m all for eye-catching shapes in my packaging, particularly as they look fantastic on the shelf, but it goes a long way to making the experience feel as grand as the graphics card itself. The locks are intuitive enough with arrows indicating to pull in the opposite direction, freeing the pixel pusher from its confines.
Nvidia’s new Founders Edition design is both familiar and refreshing. It retains the same gunmetal grey and black colour scheme as RTX 30 and 40 Series including the striking X pattern. Look a little closer, and you can just about see LED strips poking through, converging in the centre. Alongside the returning light-up GeForce RTX logo on the side, this is sure to give it a bit of flair.
The biggest difference comes from swapping the renowned flowthrough cooler seen on the past two generations and reverting back to a front-loaded dual fan config. Dubbed the ‘Double Flow Through’ design, each fan packs seven thick blades to push air across a 3D vapor chamber and thicker vents to chill a shrunken PCB. Nvidia says this is its most compact board to date, keeping the chassis far smaller than its predecessors.
It’s this cooling magic that bucks the trend of ever-growing graphics cards, swapping out imposing behemoths for a sleeker dual-slot flagship that’s the same 137 x 304 x 40mm as RTX 5080. It’s easier to install without needing a sag bracket owing to its lighter approach and should squeeze into rigs that other 90-class cards won’t go near. Win-win.
I’m not the biggest fan of branding, as I feel design ethos should speak to the product’s identity more than a big logo. Whereas previous models adorned huge block letters distinctly visible in vertical setups, Nvidia strikes a fantastic compromise by moving the smaller RTX 5090 etching to the rear.
The result is a remarkably professional-looking graphics card no matter the orientation. Frankly, it’s a shame there’s no bleached version for white setups.
While good things do indeed come in small packages, it’s still one hungry beast under the hood, carrying a demanding 575W TDP. You might get away with a slimmer power supply, but Nvidia recommends nothing short of 1,000W to keep everything running smoothly, and aftermarket cards might raise this bar further with their own cooling solutions and lighting kits.
A single 16-pin power connector is more than enough to saturate it, this time angled 45° to make it less of a pain to access. Some readers have asked why not put it at the end of the chassis to avoid wires creeping over like some Lovecraftian hand, but this all boils down to compatibility and Nvidia doesn’t want to make the card longer than it needs to be.
One thing to keep in mind is that the included 600W cable doesn’t leave a lot of headroom for overclocking, but there is a 4x 8-pin splitter in the box just in case you need one. We’ll see more on performance when the benchmarks start pouring in.
Visual connectivity finally catches up to its rivals from AMD and Intel, as Nvidia kits RTX 5090 out with three DisplayPort 2.1b with UHBR20 support and one HDMI 2.1b. Both are capable of 4K resolutions up to 480Hz, but DisplayPort 2.1b goes a step further. Should you ever find the use-case, 8K/165Hz with DSC (Display Stream Compression) is on the cards.
Of course, you’ll almost certainly need the helping hand of DLSS 4 and its neural rendering capabilities to get anywhere near this level of detail. Still, don’t count out its rasterised strength just yet, as the Blackwell flagship boasts a boost clock of 2.41GHz, a whopping 21,760 CUDA cores, and swaps in 32GB of GDDR7 for 1.79TB/s of memory bandwidth across a 512-bit bus.
GeForce RTX 5090 is one to watch come January 30, as Nvidia sets the bar for third-party vendors once again. There are few designs that compare to Founders Edition’s aesthetic and I’m still wide-eyed that the chip slips into something so slim. We’ll soon find out whether performance is just as astounding come review time.
