Meet the New King of Frames: Unboxing the GeForce RTX 5090 Founders Edition

PCMag editors select and review products independently. If you buy through affiliate links, we may earn commissions, which help support our testing.

It's been two years since Nvidia last dropped a flagship GeForce graphics card, and its "xx90"-class models grab all the eyeballs when they debut. Today, the highly anticipated GeForce RTX 5090 made a grand entrance at PC Labs. Naturally, we couldn't resist tearing into the packaging the moment it landed. (Note: The RTX 5090 and its 5080 sibling, the first two RTX 50-series cards to roll out, won’t hit store shelves until January 30.)

Ready to geek out? Let's dive into the RTX 5090's sleek design, and check out Nvidia's fresh take on packaging—yes, the box. For once, the "box" in an unboxing may be almost as interesting as the contents.

Team Green's Newly Green Packaging

The RTX 5090 is packaged with eco-conscious materials, incorporating them into the packaging design in an attractive way. Mind you, the outer box is nondescript brown cardboard...

Foxconn manufacturing labels and other internal identifiers pepper the box.

To get inside, you need to grab one of two pull tabs and peel a security strip off three sides of the box...

The long edge reveals a tagline, “Inspired by Gamers. Enhanced by AI. Built by Nvidia” beneath the tear strip.

That slogan is also a hint for where to lift. The box hinges open from the front, like so...

On opening it up, you can see an interior carrier for the card itself, plus some additional pointers. (Note the arrows.)

To get the actual card carrier out, you flip open the two box "wings" on either side to give you the space to grab the narrow ends.

Once you work free the card carrier, you'll see an accessory box is also included, in a niche below the carrier.

We'll take that out and put it aside for now...

Let's look at the card carrier. The box itself is mostly made of layered corrugated cardboard with hardboard at top and bottom...

The lid's top hardboard layer has GEFORCE RTX 5090 cut into it. Nvidia is keen to point out that no ink is involved in this portion of the boxing.

In theory, you could use the box as a display piece once you install your GPU.

Enough of the Box...How About That 5090 Card?

To get inside the carrier, you have to slide off two hardboard "brackets" to free the two halves. It's a clever design, and notably, no plastic is involved.

Lifting off the lid, there's the card of the hour.

The card's heft is profound, with a cool, all-metal-body design that's machined to the usual elite quality standards of the Founders Edition cards. The chassis of the GeForce RTX 5090 Founders Edition measures exactly 12 inches (304mm) long, 5.3 inches (137mm) high, and has a two-slot (40mm) width. Note: This is the Nvidia Founders Edition of the 5090, and these tend to be trimmer than most. Other RTX 5090 models will differ in size, and many, as we saw at CES 2025, will be much larger.

The Founders Edition design makes it compatible with many small-form-factor (SFF) PC cases, appealing to users with compact builds. That's in keeping with Nvidia's SFF-Ready initiative announced last year at Computex.

The card’s cooling solution features what Nvidia calls a "Double Flow Through" design, with two fans pushing air through the GPU. It's similar to what we've seen on earlier Founders Edition cards. But the actual PCB arrangement and layout under the metal shell has been revamped, according to Nvidia.

The multi-part PCB layout separates key components onto different physical boards within the card. A central PCB houses the GPU, memory, and power subsystem, while the PCI Express connector (a PCIe 5.0 connection, Nvidia notes) and the I/O-port block are on separate boards connected by a flexible high-speed signaling board and interior cables. This modular approach aims to enhance airflow and thermal efficiency, as well as (theoretically) make warranty repairs easier.

Nvidia describes the internal cooling as a "unified 3D vapor chamber," with the chamber design allowing heated vapor to travel directly into the card's interior heat pipes for dissipation. Heat is then directed upward and outward via outlet slits on the card edges to reduce the recirculation of heated air. The combination of straight and curved fins in the heatsink is designed to ease the passage of this air.

As for the I/O panel, you can see the usual array of three DisplayPort and one HDMI. In this case, they are DisplayPort 2.1b (supporting UHBR20) and HDMI 2.1b. These ports support the latest high-resolution, high-refresh-rate displays.

Now, here's an interesting detail: The card's power connector is nestled in a niche at a rough 45-degree angle on the top edge.

We like this; it doesn't stick straight up and isn't precarious- and fragile-feeling like in earlier implementations on Founders Edition cards. (Hopefully, the cable will get some bracing from the card body in this arrangement.) It also means no more 12VHPWR cables mashed up against the case side panel or glass, which is both unappealing to look at and distressing from the point of view of bending the cable.

A New Approach to Power

Speaking of which, what's in the accessory box?

A small getting-started pamphlet, and a converter cable. The adapter goes from the 12VHPWR connector on the card to four 8-pin PCI Express power connectors from your PSU...

Those four connectors reflect the high power demands of this GPU. (Nvidia suggests a 1,000-watt power supply; you'll need something high-end, indeed, to even gain you four 8-pin leads in the first place.) We have to say: At least at first impression, this adapter is a vast improvement over the stock ones supplied with the RTX 40-series and earlier Founders Edition cards. The cables are lighter, and they feel very soft and flexible, which ought to make routing and easing them around curves much less fraught.

The design also means putting a lot less stress on the 12VHPWR connector at the card end. (We know how badly bending those cables turned out in some earlier implementations.)

A Quick Spec Recap, With Closer Looks Coming

As reported earlier, the RTX 5090 is built on Nvidia’s "Blackwell" architecture, which proposes a host of advancements in performance, AI processing abilities, and efficiency. It features 92 billion transistors and 21,760 CUDA cores, complemented by fifth-generation Tensor cores and fourth-generation Ray Tracing (RT) cores. These specifications indicate a significant step forward in computational capability.

The GPU includes 32GB of GDDR7 memory, rated by Nvidia for 1.8TB per second of bandwidth—a vast increase over its RTX 4090 predecessor. Indeed, between the boosted specs and the veritable bushel of AI-related capabilities fleshed out by the company a couple of weeks ago during CES 2025, the GeForce RTX 5090 Founders Edition looks to be a bundle of innovation.

Support for DLSS 4 and all the rest suggest a game-changing card. But its real-world performance and value will only reveal themselves on the test bench. (We can only hope the engineering matches up to the thoughtful packaging and design.) Stay tuned for performance benchmarks and comparisons that will shed light on how the $1,999 RTX 5090 fares in the real world.