If you’ve been a PC gamer for at least a few years, you probably know how exciting it used to be to upgrade your GPU. If you jumped from a GTX 1080 to an RTX 3080, for example, the leap was massive. But the story has been shifting and is very different in 2025. Performance gains for the RTX 5000 series over the previous generation just haven't been very impressive when multi frame generation is out of the picture. And, the story might be the same for the upcoming RTX 6000 series cards.

Processing node advancements are getting limited

The major performance driver can’t do much more

Only a few years ago, significant performance jumps were driven by advancement of processing nodes. For example, Nvidia went from TSMC’s 16nm in Pascal to the refined 12nm variant for Turing, and then to Samsung’s 8nm for Ampere. Now, though, we’re at a point where Moore’s Law can’t hold any more as transistors can only shrink so much and GPUs are already incredibly dense. Nvidia used a refined 5nm node for the RTX 5000 series cards (and AMD used a 4nm process for the RX 9000 series cards)

This is not to say that there isn’t any room left. For example, Nvidia will likely shift to a more efficient (possible a 3nm) node for RTX 6000 cards, but you won’t see a massive boost in performance simply due to the shift in processing nodes. Since the current 5nm node is already an enhanced version, the difference in density won't be as big as we're used to seeing. Also, high-end graphics cards are already pushing 350W to 450W of power consumption, and cooling designs have essentially plateaued. Any raw performance improvement will only cost more power. Thus, you can expect much of the benefit of newer nodes to go into maintaining similar performance at lower power, not blowing past it.

Software features are the main highlight

Native rasterization performance is taking a backseat

Another shift is that the “benefits” of new GPUs are increasingly tied to software more than hardware. Nvidia’s performance figures for the RTX 5000 cards had DLSS 4 in the mix, which the previous generation cards do not support. DLSS 4 enables multi frame generation, using motion vectors and interpolation to insert additional frames between “real” frames. The 4X mode can theoretically quadruple FPS by generating three such frames for every traditionally generated frame. This is what enables the dramatic performance improvements over the previous generation cards; remove MFG and the performance gains are quite mediocre.

The funny thing is that MFG is not some magical performance boost that will, say, make an RTX 5050 run Cyberpunk 2077 at 150+ FPS. You’ll generally want a base FPS of 100 FPS or more to use MFG (or DLSS 3's frame generation) for a "decent" experience. The lower your base FPS is, the higher the chances of artifacting and latency issues. So, MFG is really only helpful if you’ve got a high refresh rate monitor but you simply can’t hit that FPS natively with your graphics card.

Anyhow, the point is that a significant portion of that “performance uplift" comes from software and the value of buying a new GPU becomes less clear. When the tech isn’t available on older cards (which is the case for DLSS 3 and 4), it feels anti-consumerist; users that are a single generation behind can’t utilize that shiny new feature.

Forget the two-year upgrade cycle

It's not worth it anymore

The bottom-line is: if you're looking to upgrade from one generation to the next on the same price segment, you won't find yourself being impressed.

The scenarios where the upgrade might be worth it are more extraordinary now, but they still exist. If you’re holding on to a GTX 1080 Ti or a similar, old card, you will experience a giant leap in virtually every dimension: performance, features, and efficiency. The typical generational move, especially within the same tier, now delivers incremental improvements that are not justified for a large portion of PC gamers.

If you are already running a last-generation card, jumping tiers (like moving from an RTX 4070 to an RTX 5080) is when you’ll see a decent performance improvement that’s not powered by frame generation. But again, this means you’ll be paying much more for performance that you should’ve gotten at less, which would’ve been the case in the past.

Lastly, and this goes without saying: if you’re okay using MFG for a serious performance boost, then upgrading to the latest generation cards isn’t a bad option. Note though that if you plan on running games at high refresh rates, you’ll still want a card (and PC) that can output around 100 FPS before you slap on DLSS 3 or MFG.

GPU upgrades just aren’t that exciting anymore

The Nvidia GPU market has seemingly entered a new era and picking a new graphics card is a bit harder, unless you’re upgrading from a fairly old generation. Performance gains by generation are stagnating as transistor density is beginning to hit its limits, and uplift is only down to architectural improvements, upscaling, and frame generation. So, simply upgrading to the next generation may not lead to the performance boost you were looking for. If you're on Ampere or an older platform, upgrading to the latest generation can certainly be worth it. If you’re rocking the latest generation (or RTX 4000), you may have to wait a little longer than usual before the next worthy player comes through.