AMD's next-generation RDNA 4 graphics won't compete with Nvidia at the high end, the latest rumor from Twitter contributor Kepler: mass hysteria among PC enthusiasts, cats and dogs living together, and basically doomsday coming. On the surface, it sounds like very bad news. But is it really?

The Twitter account Kepler is one of the more reliable sources of leaked information on graphics technology. But what does it really mean if it is true?" roughly speaking, there are two interpretations, one quite positive and the other quite devastating.

Interpretation one is that high-end GPUs are irrelevant to 99% of gamers and that AMD's focus on affordable graphics cards is actually a good thing; offer 75% of the performance of Nvidia's high-end GPUs for half the price and all is well, AMD.

These days, maybe it should be half the performance for a quarter of the price. At any rate, the basic idea makes sense, and AMD has a track record in this regard: it was only two generations ago that AMD didn't care about high-end and limited its RDNA 1 family to the Radeon RX 5700 XT, priced between the Nvidia RTX 2060 and 2070 cards That is.

AMD did much the same with the Radeon RX 400 and 500 series of GCN-based Polaris chips; the RX 480 launched in summer 2016 for $239, just after Nvidia launched its powerful $659 GeForce GTX 1080. The same was true for the RX 580 and later the RX 590. They did not take the fight to Nvidia at the high end.

Going back even further, something similar happened in the days of the Radeon HD 4870 in 2008, when it had just become an ATI. The card cost $299, while Nvidia's GeForce GTX 280 was $650.

Given all this, AMD's lack of commitment to the high end is nothing new, and therefore nothing to panic about; if AMD can offer a great graphics card for roughly $250 to $500, who cares what Nvidia does for $1,500 or more? Who cares what Nvidia does for over $1,500 if AMD can offer a good graphics card for around $250 to $500?

On the other hand, there is a sense that AMD wants to compete at the top table of graphics cards but keeps failing. There were rumors of a faster RDNA 1 card (tentatively called the 5900 XT), but it never appeared.

The 6900 XT, on the other hand, came out in 2019 and was quite competitive at the top end. The new RX 7900 XT and RX 7900 XTX were clearly attempts to compete with Nvidia's RTX 4090, but fell a bit short of their goals.

In general, there is no denying that AMD's GPU efforts have been a bit hit or miss. For example, where is AMD now in gaming laptops?

Therefore, catastrophists will see the RDNA 4's rumored inability to compete at the high end as the next step in AMD's exit from the desktop gaming GPU market as well. Instead, AMD will focus on console graphics and APUs, leaving the desktop gaming graphics market to Nvidia.

Well, Nvidia and maybe Intel. However, Intel's Arc GPUs have hardly dominated the market. And if AMD cannot compete with Nvidia, what chance does Intel have?

Ultimately, we don't know how this will play out; it's hard to even know what AMD intends to do. If there is no high-end RDNA 4 GPU, is it because AMD tried and failed like RDNA 1, or is this a deliberate strategy from the start?

One realistic possibility is that Nvidia is not really making money on a GPU like the RTX 4090, even at a price of $1,600; Nvidia is selling the AD102 GPU used in the RTX 4090 for $6 ,800 for a professional graphics card and $8,000 for a server card. Of course, these cards have twice as much VRAM, 48 GB, but that alone does not add $5,000 to the cost.

In short, the AD102 is a huge chip and very expensive to manufacture. So the RTX 4090 is GeForce's loss leader even at $1,600, and Nvidia may lose money on every RTX 4090 in exchange for the increased brand power of being able to claim the top spot in gaming performance.

And maybe that's a game AMD doesn't want to get involved in. After all, it is really pointless to come in second and lose money on every card.