Nvidia could trim RTX 5090 and 5060 for laptops when it comes to VRAM – but the RTX 5080 GPU could be very different
Nvidia’s Blackwell laptop graphics cards will arrive in early 2025 (hopefully) and we’ve just discovered a leak about the amount of video memory (VRAM) they should work with.
This one is from Moore’s Law is Dead (MLID), a regular leaker on YouTube (and in our experience one of the more reliable sources for GPU and CPU leaks).
MLID has the alleged full details of the VRAM and they work as follows. The flagship RTX 5090 will likely keep exactly the same configuration as the RTX 4090, which some people will no doubt find disappointing. This means it works with 16 GB of VRAM and a 256-bit memory bus.
Where the better news theoretically comes in is that the next model isn’t available, presumably the RTX 5080 – or maybe we’ll call it the second-tier Blackwell GPU (you’ll see why in a moment) – with the same 16GB of VRAM equipment (and 256-bit bus). That’s a solid increase over the current RTX 4080 laptop GPU which has 12GB.
However, there is a small catch here. If you remember, the RTX 4080 mobile GPU used a smaller chip than the flagship – the AD104 chip, instead of the AD103 in the RTX 4090. But with the RTX 5080, a recent rumor (again from MLID) claims that it will have GB203 , exactly the same as the flagship 5090, meaning it’s potentially going to be a lot punchier (with that VRAM upgrade on top too).
What this also means is that Nvidia may want to reflect that and call it the RTX 5080 Ti, instead of just the regular 5080.
And that could have a knock-on effect on the rest of the range, as if that’s the plan then the third-tier GPU (RTX 4070 equivalent) could be the RTX 5080. But if the second tier Blackwell GPU is the RTX 5080, then the third tier will obviously be the RTX 5070.
You get the idea either way, but whatever it’s called, this third-tier GPU will likely have 12GB of VRAM, up from the 8GB seen on the RTX 4070.
Finally, the fourth-tier GPU – most likely the RTX 5060, or perhaps RTX 5070 if Nvidia follows the 5090/5080 Ti/5080 naming scheme – will leave 8GB of VRAM. (Hear a chorus of boos and hisses).
Analysis: good, bad and 8 GB ugly
So there’s good news and bad news – but add spice, and lots of it, and remember that this is just grapevine talk.
On the plus side, we could get a powerful RTX 5080 compared to the RTX 4080 – so much so that Nvidia could call it an RTX 5080 Ti. Either way, the top options could be beefed up, and it’s good to see the third-tier GPU eventually getting an increase of 4GB of video RAM (upped from 8GB to 12GB).
The not-so-great chatter here, of course, is not increasing the VRAM of the flagship mobile GPU for Blackwell, and even worse, maintaining an 8GB configuration with that fourth-tier GPU (most likely RTX 5060 – or as mentioned would this might even include the RTX 5070).
For gaming in the future – remember, these are the next generation of GPUs that will rule for a few years, until 2027 (at least) – those two VRAM gears are far from great, and the choice of 8GB lower seems particularly shaky.
Assuming Nvidia goes this route, of course, and we should underline that this may not happen at all – so it’s too early to worry too much about these prospects.
However, as for the possible level of accuracy here, MLID says this information comes from some of its best sources. We’re talking about people who have come up with early photos of GPUs before – in short, well-connected Nvidia leakers. Given that MLID places a lot of weight on these predictions.
As always, let’s see if other leaks line up with MLID’s predictions. But if this is true, the top and bottom of the Blackwell spectrum of GPUs for gaming laptops are in danger of disappointing. In theory, these mobile graphics cards should hit the market quite early in 2025, most likely with a March release, as previously reported.