Wow, the graphics card market is in complete chaos!

On one hand, the veteran RTX 3060 has been forcibly "resurrected from the dead," while on the other, the entire RTX 50 series is lining up for price hikes—some even getting downgraded instead. What a bizarre plot...

First, let's talk about this "classic revival." The RTX 3060, launched in 2021 and retired honorably in 2024, was unexpectedly dug up from its grave by Nvidia in 2026.

According to the latest news, RTX 3060 models from various brands will be restocked between March 10th and 20th. However, the revived versions are mostly the "Lite Edition" with 8GB VRAM, where the memory bus width has been cut from 192-bit to 128-bit—a direct performance downgrade.

What’s even more absurd is that NVIDIA is only supplying the GPU cores this time, leaving AIB partners to negotiate with memory suppliers themselves. Talk about passing the buck flawlessly.

The price is rumored to be similar to its original launch, but given current material costs, manufacturers might be selling at a loss just to make noise. So, don’t expect high availability.

Here's something even more exciting: due to the soaring cost of VRAM, the RTX 50 series graphics cards have seen a "universal price hike."

Taking ASUS models circulating online as an example, almost the entire lineup has increased in price, with even relics like the GT 710 rising by 40 to 55 yuan.

Breaking it down: the flagship RTX 5090 D v2 jumped by 500 yuan, most RTX 5070 Ti models rose by 200-300, the lower-end RTX 5080 variants increased by 100-200, and the RTX 5060 saw a uniform bump of 50 yuan.

The only exception is the RTX 5060 Ti 8GB, which remained unchanged—but not necessarily out of goodwill. It's likely because NVIDIA is aggressively pushing this version, as the 16GB variant is too costly.

Interestingly, at least for now, AMD's graphics card prices haven't budged at all. So, it seems the blame can't be entirely pinned on VRAM.

Amidst the wave of price hikes, the entry-level RTX 5050 has undergone a "reverse upgrade."

It will shift from the current 128-bit 8GB GDDR6 to a 96-bit 9GB GDDR7 configuration. Although the bus width has been reduced, the higher frequency of GDDR7 may actually result in slightly improved bandwidth.

So why doesn’t NVIDIA just go with a 128-bit 12GB setup? Simple—they don’t want to steal the RTX 5060’s thunder.

Speaking of the RTX 5060, things get even messier. Due to an oversupply of NVIDIA’s GB205 chips (used for the RTX 5060 Ti/5070, etc.) and a shortage of the more entry-level GB206 chips, the RTX 5060 might be forced to use a "double-cut" GB205 chip instead.

While the specs remain unchanged, the whole situation reeks of inventory clearance.

In short, the current GPU market is like this: if you buy a new card, you’ll have to pay more; if you opt for a rehashed older model, it’ll still be a cut-down version. With all these moves, AMD might be the happiest right now—after all, when everyone else is raising prices, not doing so becomes the best value proposition.