The AI Tech Crunch: Are We Looking At A "Dark Age" For Gaming Hardware?

Memory giant Micron has announced that the days of its Crucial memory brand are coming to an end, the firm shifting resources away from consumers and towards business customers in a clear shift towards maximising profits from the AI boom. Losing a major supplier of memory to the PC market clearly isn't good news for the industry - less competition, higher prices - and the all-consuming nature of the AI gold rush has profound implications across all areas of gaming hardware: less access to the cutting-edge process nodes needed for more performant silicon and a potential dearth of resources for feeding the PC market. Technological innovation in the hardware space could be stifled, tomorrow's PC being little better than the hardware of today.

Let's consider the here and now. In terms of affordability of PC parts, prices have risen for SSDs - though there are still relative bargains to be found. Current generation graphics cards do still seem to be available at reasonable prices, close to their original MSRPs - if not lower in some cases. CPUs are fine too. However, RAM is the pain point. Two 16GB DDR5 modules for 32GB total start at £300 on Amazon UK and around $330 in the US. More premium, lower latency kits cost much more.

On the face of it, that doesn't sound too bad. However, buying a new CPU may involve moving to another motherboard, which in turn may need new RAM. The nature of the PC is that certain upgrade choices start a cascade of new component requirements, so it's little surprise to hear reports (unconfirmed, to be clear) that motherboard sales are down 50 percent. But the point is that one major pricing upset within the ecosystem has dire implications for more widescale upgrades. Shifting to pre-built systems is a potential option - but the blast radius caused by the sudden explosion in RAM pricing will get there too, eventually. By how much will prices increase there? Of course, right now we don't know.

Looking to the medium term, let's consider the prospects for other PC components beyond memory and where the continued, rapacious demand from the AI market will cause issues. For graphics, I don't expect to see moves from Nvidia to refresh the RTX 50-series line any time soon: the leaked Super refresh specs were all about memory upgrades, which now do not seem viable. There were hopes for AMD's RDNA 5 to arrive at some point in 2026, but I don't see why that would happen: RDNA 4 is a solid product.

When we would actually expect to see next-gen Nvidia Rubin or AMD RDNA 5 is surely going to be 2027 at this point - and if RTX 50-series Super is delayed (Nvidia will argue that it hasn't announced anything, therefore there is no delay) and arrives in, say, Q3 2026, we would expect that to push back Rubin until much later into 2027. Graphics innovation drives the PC gaming market and updates to existing lines could be a long way off, let alone true next-gen upgrades. How long really depends on what 3nm inventory AMD and Nvidia have secured from chip manufacturer TSMC and if memory is available at a reasonable price to make a sizeable roll-out possible.

There is more cause for optimism in the CPU space. In November 2025, AMD confirmed that Zen 6 is coming this year, based on the TSMC 2nm node - where we can safely assume that orders have been made and wafers secured. Intel also has exciting products in the form of the mobile-orientated Panther Lake, side-stepping TSMC by using its own foundry. Nova Lake for the desktop space sounds remarkable, with the top-end SKU rumoured to be hitting up to 52 cores - though we'll need to persist through a refresh of Arrow Lake first.

Those already on a DDR5-based platform can upgrade and keep their existing memory with these new products. Those moving on from a DDR4 system will face the same problems we are starting to see now with memory pricing and availability. There may also be question marks over motherboard pricing: if the market size shrinks, margins may be increased.

Looking to buy a console instead to avoid the drama and confusion? There may be difficulties there too. The same GDDR6 memory found in graphics cards is used in Xbox Series and PlayStation 5 hardware. The AI market craves the LPDDR5X RAM modules used in Switch 2. We've seen price rises in the console space - despite Sony and Microsoft hardware that's now over five years old - and we may well see them again in 2026.

With demand for memory and silicon wafers so intense, questions need to be raised about the viability of actual next generation consoles. Will sky-high memory prices have an impact on console pricing? Will the 3nm process node earmarked for the next-gen AMD console processors used in the upcoming Xbox and PlayStation get vacuumed up by the all-consuming AI beast? The case could be made for delaying new hardware.

Pushing new graphical frontiers to the next level tends to define what are generally considered to be "gaming generations" so the idea that new consoles and graphics cards may be pushed back, alongside pricing concerns for the kit already available makes me wonder - are we looking at some kind of period of stagnation for gaming technology?

Some might say we are already there. The traditional idea of huge generational leaps in the PC space is clearly diminishing. In 2020, GeForce RTX 30-series was a big leap at genuinely impressive prices. RTX 40-series was too a couple of years down the road, but you had to pay extra for that performance boost - and to be clear, the market did, based on the Steam Hardware Survey. RTX 50-series saw far less meaningful performance increases, with only the monstrous RTX 5090 outpacing its predecessor to a significant degree.

I would expect Nvidia Rubin and AMD RDNA 5 to see another impressive performance uplift owing to the 3nm process node, but it may be the last for some time - and you may need to accept higher prices, RTX 40-series style. But the concept of what a leap in performance actually is - or more specifically, where it will come from - will change. If hardware can only go so far, the focus will shift to software. And it is already happening.

Nvidia will direct your attention to actual path-traced video games where the vast majority of the pixels you see are generated by AI. It will continue to push the concept of multi frame generation as an actual performance increase, much to the chagrin of the tech press and enthusiasts. AMD? Well, clearly Mark Cerny and Sony are deeply involved in both the hardware and software side of RDNA 5 (or whatever it will be called) and the effort there is similar to Nvidia: a more holistic approach to rendering with more silicon dedicated to ray tracing acceleration and - yes - machine learning-based functionality. If you've not been paying attention to our Mark Cerny interviews, check out the design of the PS5 Pro itself and Sony's work on ML-based super resolution.

On the PC side, AMD's FSR Redstone will be revealed soon, highlighting the firm's commitment to machine learning features. Nvidia will continue to deploy its vast engineering resources towards improving DLSS and ML frame-gen. And the objective here is clear: if users want new experience and higher frame-rates, silicon alone isn't going to deliver. It'll be a combination of hardware and software.

So if the AI crunch is going to delay game-changing new hardware for a year or two, the focus will inevitably shift to software - whether that's working more intensely on the current generation consoles to extract more from them (did you see that Witcher 4 demo running on PS5?) or on doubling and tripling down on technologies like DLSS and FSR in the PC space. If GPU vendors can't refresh their hardware, they'll push features more heavily instead, adding new technologies and refining existing ones until they stop being value-added extras and instead become a must-have part of the package. DLSS super resolution made that transition and proves it can be done and I'm confident that at some point frame generation will do the same.

In the short term though, the AI crunch is going to be tricky. Just how tricky, we don't know. Whether it's the fear of missing out or the greed of the retail market, DDR5 memory prices have skyrocketed. Micron's withdrawal from the consumer space is the strongest signal yet that businesses may forsake PC in favour of a more lucrative market - which could be viewed as a betrayal bearing in mind the historical association between the platform and the Crucial brand. More may follow. The knock-on effects to the console and graphics card markets are also yet to be determined - but we have reasonable pricing now and some great products, so as we've stated recently, there are good reasons to buy now. It doesn't seem likely that prices will reduce in the short term, let's put it that way.

What we've seen so far in terms of memory pricing on retail sites looks distressing - but it remains to be seen how this will impact the broader market. One might imagine that companies like Dell and Lenovo still have a great degree of negotiating power when it comes to adding DDR5 and LPDDR5X to their prebuilt PCs and notebooks. Similarly, the impact on GDDR6 and GDDR7 pricing for graphics cards remains to be seen - rumoured price increases look "fine" but what if retail prices end up disproportionately marked up again? We have many questions.

In a few weeks' time, the industry congregates in Las Vegas, Nevada for the Consumer Electronics Show. Perhaps with such a high concentration of brands, executives and journalists in one place, we might start to get some firm answers.