When the news of AMD's delayed RX 9000 launch first came in, I was confident it would hurt the company's chances of making a dent against Nvidia's RTX 50 series. Even AMD fans had had enough of the protracted tease, and wanted some confirmation around the price and performance of the long-awaited RX 9070 XT and RX 9070. So, when the company finally announced the cards on March 6th, almost no one expected the event to go as well as it did.

Not only did AMD surprise everyone with the pricing, but it also came dangerously close to making the RTX 5070 and RTX 5070 Ti irrelevant, even if they were selling at MSRP. Team Red ensured the launch was all about raw performance instead of tall claims like a certain 70-class card beating a previous-gen 90-class card. Nvidia had already set a low bar, but AMD's RX 9000 series didn't just raise it; it smashed right through it to a near-flawless launch.

5 It offered significantly more value

Better than anyone hoped for

AMD made it clear months ago that it wasn't planning to compete with Nvidia in the high-end segment, focusing instead on mid-range cards, i.e., the ones most people buy. Still, speculation was rife that AMD would drop the ball again when it came to the final pricing of the RX 9070 XT and RX 9070. Many reports claimed the XT variant would be priced at over $700, and the non-XT would clock in at around $600-$650.

AMD surprised everyone with MSRPs of $599 and $549 for the RX 9070 XT and RX 9070, respectively. AMD's revamped naming scheme for its latest cards had already made it clear that these GPUs will be going against the RTX 5070 Ti and RTX 5070, respectively, but the company's launch event hammered the point home. By offering virtually the same raw performance as Nvidia's RTX 5070 Ti at a significantly lower price, AMD's RX 9070 XT instantly became the value champion of this generation. The RX 9070, while not as impressive, still beat the RTX 5070 in terms of cost per frame.

Of course, in the current market, none of the MSRPs actually exist, but when prices eventually stabilize, AMD is likely to retain its value crown. AMD is also trying its best to keep the current prices down by restocking the new GPUs aggressively, so we can hope more gamers can grab the RDNA 4 GPUs at reasonable prices.

4 The focus was on rasterization

Fake frames be gone

If you've seen or read anything about Nvidia's Blackwell launch, you know the company wanted everyone to focus on the incredible FPS gains enabled by its Multi Frame Generation (MFG) feature exclusive to the RTX 50 series cards. Team Green conveniently sidestepped any discussion on whether the AI-generated frames would be equivalent to those rendered by the game engine (which they obviously aren't). The input latency and artifacts inherent to frame generation are only worsened in the 3x and 4x modes of MFG. Plus, MFG works best only when it makes the least sense to use it.

AMD clearly came prepared with its RX 9000 launch, since the entire presentation was mostly centered around rasterization or native performance. The company ensured people knew it was actively avoiding discussing AI-generated or "fake" frames. Whether it was comparing its new cards against the RDNA 3 GPUs or Nvidia's Blackwell offerings, it barely spent any time on FSR-enhanced graphs.

AMD's launch was clearly the one with the better optics — prices were better than expected, dependency on AI performance was extremely low, and the performance comparisons with its previous-gen cards were transparent and effective.

3 AMD didn't skimp on VRAM

Kind of a pattern, but welcome, nonetheless

Whatever Nvidia's achievements with Blackwell, the company refused to outfit its cards with enough VRAM. Except for the $2,000 RTX 5090, none of the other SKUs got sufficient VRAM, in my opinion. 16GB on the RTX 5080 and RTX 5070 Ti, and a measly 12GB on the RTX 5070 are embarrassing by themselves, but AMD piled up on top with the RX 9070 XT and RX 9070. Both the cards featured a 16GB framebuffer, which isn't massive, but it's far better than anything on the Nvidia GPUs, especially at the price points in question.

The limitations of 12GB of VRAM are already apparent on the RTX 5070, as it struggles to maintain a decent framerate in some demanding titles. The RX 9070, on the other hand, offers 50% more VRAM for the same MSRP. It might lag behind the RTX 5070 in most ray-traced games, but overall, it's still going to age better than the Nvidia offering. For some reason, Nvidia refuses to acknowledge the impact VRAM has on high-resolution gaming, and its mid-range cards have consequently been taking the brunt of it.

2 Ray tracing and FSR 4 beat expectations

AMD is no longer the underdog

We knew AMD was cooking up something powerful with RDNA 4, thanks to early confirmations of dedicated ray tracing hardware, and a heavy focus on AI and ML in FSR 4. However, when the company finally unveiled its ray tracing benchmarks and FSR performance, everyone was left pleasantly surprised. While Team Red hasn't yet caught up with Nvidia in RT performance, it's closer than ever in many titles, and it has shown a tremendous uplift compared to RDNA 3. Ray tracing was one of the weakest areas of AMD cards. The trend has certainly been reversed.

Another weakness of AMD GPUs was the performance of FSR against DLSS. FSR 3.1 looks downright horrible when compared head-to-head against FSR 4. The textures and sharpness during movement, and the handling of objects like fences, water, and grass have improved drastically, so much so that FSR 4 now stands firmly between DLSS 3 and DLSS 4 in terms of overall quality and efficiency. Visual artifacts, excessive blur, and problems with fine details — everything has been improved to the level that FSR has gone from barely usable (in many scenarios) to competitive with DLSS.

Nvidia would not have expected AMD to make such a huge leap in ray tracing and FSR, and considering the RTX 50 series didn't do much in terms of delivering strong generational gains, the competition between the two companies is more exciting than ever.

1 No blatant lies like Nvidia

"RTX 5070 = RTX 4090"

Of all the disappointing elements of Nvidia's RTX 50 series launch, perhaps the most brazen was the claim that the RTX 5070 would offer RTX 4090 levels of performance. Nvidia hoped most people would look at the RTX 5070's 4x MFG numbers alongside the RTX 4090's 2x frame generation numbers, and reach the same conclusion. While many consumers might still agree with Nvidia, the truth is that even with the same FPS numbers, the "performance" is hardly the same between the RTX 5070 and RTX 4090.

This is because the technically weaker card, the RTX 5070, needs to rely on 4x frame generation to come close to offering RTX 4090's FPS numbers (with 2x frame generation). This means the Blackwell card showcases significantly worse latency and visual artifacts vs. those on the RTX 4090. The actual experience of playing the same game at the same framerate on these two very different GPUs couldn't be more different.

AMD, thankfully, didn't make any false claims in its presentation. It might have oversold some of the benchmarks, but that's about it. Team Red focused on native performance, visited AI enhancements only briefly, and put all the cards on the table. The launch was as transparent and to the point as one could have hoped, throwing into question Nvidia's intentions when showcasing its RTX 50 series cards.

AMD came off as a stand-up guy

That's essentially what it is. While Nvidia went hard with MFG, boasting huge AI-driven gains over the RTX 40 series, AMD kept things simple, focusing on rasterization performance and steering clear of any unfounded claims. The availability, and hence, pricing of both the RTX 50 and RX 9000 series is out of whack right now, but as time goes by, AMD's launch will be seen as transparent and sincere by more and more consumers.