The king is dead. Long live the king. Since it launched in October 2022, Nvidia’s GeForce RTX 4090 has been the flapship GPU to beat. It’s a monstrously powerful card and one that Team Green’s rivals AMD have yet been able to rival. It truly is in a league of its own. And yet now it looks like it’s about to get bodied by an even faster graphics card.
Enter the long-rumored RTX 5090. Sources are beginning to whisper about a release date for Nvidia’s top-of-the-stack terror, with the 5090 supposedly penciled in for “late Q1 2025 or early Q2 2025 launch” (thanks, Notebookcheck). YouTuber “Moore’s Law is Dead” has also leaked the specs of the laptop version of the RTX 5090 (also just a rumor at this point), and while it obviously can’t match the desktop version, it ain’t exactly a potato.
With the help of TechPowerUp and trusted X leaker “kopite7kimi” and MLiD, I’ve put together a table comparing the rumored specs of the RTX 5090, RTX 5090 Laptop Edition, the RTX 4090 and the RTX 4090 Laptop Edition. Spoiler: my fellow 4090 owners might want to look between their fingers.
RTX 5090 vs RTX 5090 laptop vs RTX 4090 vs RTX 4090 laptop
| Header Cell - Column 0 | Nvidia RTX 5090 (rumored specs) | Nvidia RTX 5090 Laptop (rumored specs) | Nvidia RTX 4090 specs | Nvidia RTX 4090 Laptop specs |
|---|---|---|---|---|
| GPU processor | GB202 | GB203 | AD102 | AD103 |
| Processing cores | 21,760 | TBA | 16,384 | 9,728 |
| Bus width | 512 bit | 256 bit | 384 bit | 256 bit |
| Memory size | 32GB | 24GB | 24GB | 16GB |
| Memory type | GDDR7 | GDDR7 | GDDR6X | GDDR6 |
| TFLOPS | 109.7 | TBA | 82.58 | 32.98 |
| TDP (Total Power Draw) | 600W | TBA | 500W | 120 |
As a fan of the best PC games, I should be drooling right around now. 21,760 CUDA cores! A 512 bit bus! 32GB of RAM! 109.7 TLOPs! Some of those rumored RTX 5090 specs are downright scary. And from what we now know about the 5090 laptop GPU, it looks set to eviscerate the 4090 laptop version when it comes to memory performance.
Hell, the mobile 5090 even outdoes my desktop 4090 thanks to that GDDR7 RAM. Just remember to keep your pockets stuffed full of salt, as Moore’s leaked specs obviously haven’t been confirmed by Nvidia.
For context, an Xbox Series X has 12 TLOPs. Oh, and on that topic, I realized my 4090 is actually longer than Microsoft’s console. Yes, I was actually sad enough to break out a tape measure. My colossal INNO3D card measures 13-5 inches long, while the Series X is 11.85 inches. Now just imagine how large an RTX 5090 could be. I had to buy a new full-tower PC case to fit my 4090 in it because it was soooo much bigger than the RTX 3090 I’d previously owned.
I imagine there are a whole lot of hardcore PC players out there worrying if the 5090 will fit into their current tower case — I doubt a mid-tower is going to get the job done. I’d also be shocked if the first batch of RTX 5090 laptops didn't bring the chonk and some serious heat to boot.
The rumored jump in memory between the two RTX 5090 models compared to their RTX 4090 counterparts looks significant, to put it lightly. The RTX 4090’s 24GB of memory already feels overkill when it comes to gaming (12GB is normally enough to play most modern titles at high quality levels). So the 32GB the RTX 5090 is allegedly packing is practically unhinged.
And 24GB of on-board graphics memory in a laptop? Yowsa. Prosumer video editors who have to render colossal 4K video files are going to have a field day with both versions of the 5090.
How much will the RTX 5090 and RTX 5090 laptops cost?
As, no doubt, will scalpers. The RTX 4090 officially released for $1,599, yet it was difficult to buy at that price ‘cause scalpers gonna scalp. While I’d love to see Nvidia sell the RTX 5090 for that same price, I’d be shocked if it happens. My take? I’d be surprised if it costs less than $1,999.
As for how much a RTX 5090 laptop will set you back? Considering the likes of the Alienware m18 in its 4090 config costs $3,799, I’m going to go out on a limb and say “A LOT”.
With the increasing advances in the AI-driven DLSS 3.5 (like frame generation), my desire to buy a RTX 5090 or a laptop with its mobile GPU is actually pretty low. I’m hardly going to be thirsting after a RTX 5090 laptop when I bought the best gaming laptop less than six months ago.
I adore the Asus ROG Zephyrus G14 (2024), but I can’t deny even the “The MacBook Pro of gaming laptops” — cheers, for that one, Tony Polanco — would get bodied by those rumored 5090 mobile specs. After all, the G14’s laptop RTX 4070 only has 8GB of GDDR6 and 128 bit bus. Looking back at those numbers above, my beloved Zephyrus would get smoked.
I’m also perfectly happy with my RTX 4090, which granted, is hardly a “standard” GPU. When I can run the vast majority of my Steam library at 4K/120 fps (admittedly with the help of Nvidia supersampling), I’m not exactly gasping at the chance to throw Team Green 2000 Big Boys.
Both the desktop and laptop versions of the RTX 5090 will be incredible cards, I’m in little doubt of that. Yet with AI increasingly making pure hardware grunt less important when it comes to achieving high frame rates in the best Steam games, this is one PC player who’ll just have to “struggle” on with his RTX 4090 and the laptop RTX 4070 in his awesome Zephyrus G14.
Now if you’ll excuse me, I have to go and play the world’s smallest violin.
More from Tom's Guide
- I’m seriously thinking about selling the best GPU on the planet
- I’m super glad Nvidia isn’t releasing a 4090 Super — here’s why
- I own an RTX 4090 gaming PC, yet I’d rather play my Switch OLED
| Price | Row 0 - Cell 1 | Row 0 - Cell 2 |
| Row 1 - Cell 0 | Row 1 - Cell 1 | Row 1 - Cell 2 |
| Row 2 - Cell 0 | Row 2 - Cell 1 | Row 2 - Cell 2 |
