No new graphics cards until 2028? Just as well
According to its manufacturing partners, AMD isn’t planning to launch RDNA 5 cards until late 2027. Realistically speaking, it’s more likely to be 2028. In fact, Nvidia’s on a similar timeline with its Rubin architecture. That may sound like doom at gloom, but in reality, it’s pretty much a win-win.
Tom’s Hardware recently reported what AIB partners were whispering about behind the scenes at Computex 2026. Namely, that new AMD graphics cards aren’t anticipated to be released until the second half of 2027 at the earliest. Some manufacturers are even betting on early 2028. It’s the same story for Nvidia and its Rubin architecture for consumer graphics cards. The latest generation’s Super refresh might be scrapped entirely, but that isn’t set in stone yet. News like this used to get me all worked up. These days, it just elicits a shrug.
Software upgrades are replacing hardware ones
When was the last time I noticed my graphics card had got noticeably faster? Not after buying a new one, but after a driver update. Technologies such as AI-powered frame generation, ray reconstruction and neural upscaling deliver genuine improvements in performance – on existing hardware.
Take DLSS 4.0, which came out at the beginning of 2025, for example. It’s taken upscaling to a level where it’s become virtually indistinguishable from native rendering. With DLSS 4.5, even performance mode is so good that it practically makes native 4K resolution unnecessary. AMD’s FSR 4.1 followed suit with fewer artefacts and better detail. Rather than a stopgap measure, upscaling has become a genuine alternative.
Even more impressive progress has been made when it comes to frame generation. Once considered a workaround for underpowered hardware, it now delivers frame rates that hardware alone could never achieve. The RTX 50 series introduced 4x multi-frame generation, with DLSS 4.5 boosting that to up to 6x. While it might not be suitable for competitive gaming, it’s a game-changer in single-player titles.
The biggest performance leap we’ve seen in recent years, however, didn’t emerge from the factory floor. Instead, it came from the software lab. AMD and Nvidia are even rolling out these technologies for older GPUs, demonstrating that the future is in software.
The latest hardware more than cuts it
What’s something your current card can’t do that you really need? With an RTX 5070 or RX 9070 XT, all the latest games run without a hitch. Even older graphics cards deliver decent enough performance, albeit maybe with a slightly lower frame rate. Since the RTX 40 and RDNA 3 generations, GPUs have reached a level that new models can’t really beat in a noticeable way. In games, you see the same textures.
What’s more, console release cycles are likely to get longer as a knock-on effect of the AI boom and resulting memory crisis. Even things that initially seem unrelated to graphics cards can impact them. Games are released for both PC and consoles, with the latter holding back progress in terms of graphical quality. If software is developed for the same hardware over an extended period, progress made on graphics slows. Older generations of consoles will also continue to receive new games for longer. This is because PlayStations and other consoles are getting increasingly expensive, so many people are waiting before switching to a new one. There’s a good chance your current GPU is more powerful than a PS5. With this in mind, your graphics card’s performance might stay good enough for a while yet.
A new generation of GPUs used to mean twice the performance, but those days are over. Physical limitations brought on by heat, power consumption and wafer costs are slowing down development. Nvidia and AMD are squeezing every last bit of performance out of their chips, but the gains are getting smaller. RDNA 5 and Rubin will deliver high-performance cards, but they won’t be revolutionary – at least not in terms of hardware.
A year without a GPU release will be good for your wallet
AMD, Nvidia and TSMC are competing for manufacturing capacity. The gaming market is losing this battle against the more lucrative data centre business. An H100 or MI300X is a thousand times more profitable than an RTX 5080. So rather than being a disaster, a GPU release break is a logical consequence of those developments. Plus, a longer product cycle might be beneficial in developing the next generation of GPUs. I’d rather have a product that’s fit for everyday use at launch than one that’s half-finished.
The industry having other priorities impacts prices too. And, as we all know, GPU prices are steep. The pragmatic way of looking at the lack of new flagship GPUs? You’ll get to keep your money. There’ll be no hype to tempt you into overspending. No first edition with teething problems. No feeling that your hardware is outdated. Instead, it’ll be a quiet year that’ll go easy on your wallet.
No cause for FOMO
FOMO is the GPU industry’s most powerful marketing tool. Every generation is touted as a quantum leap forward, despite often being just a small step. Previously exclusive features are now gradually becoming available to older GPU generations as well. If you already have a solid graphics card, you won’t miss out on anything capable of fundamentally changing your gaming experience until 2028. No games will overwhelm your GPU, no feature will be off-limits – and it’s all thanks to software.
So, don’t panic. Enjoy the games you have, and let the industry get on with developing RDNA 5 and Rubin.
From big data to big brother, Cyborgs to Sci-Fi. All aspects of technology and society fascinate me.
This is a subjective opinion of the editorial team. It doesn't necessarily reflect the position of the company.
Show all
