Nvidia's obsession with DLSS 5 is finally giving AMD the opening it needs

Published Aug 4, 2026, 12:00 PM EDT Abhinav pivoted from a career in banking to pursue his first love in writing. Even while working full-time, he continued contributing as an editor-at-large, a role he has held for more than 7 years. A lifelong tech enthusiast who has built three gaming and productivity powerhouse PCs since 2018, his passion for technology keeps him closely following the semiconductor industry, from NVIDIA and AMD to ARM. His MSc dissertation explored how artificial intelligence will reshape the future of work, reflecting his curiosity about the wider social impact of emerging technologies. If you've followed the discourse around GPUs since GTC 2026, you probably already know the popular opinion. DLSS 5 and its neural rendering landed really badly, gamers and developers revolted, and it seems like Nvidia's software moat, which has been the biggest reason to buy GeForce over Radeon, is finally showing signs of leak. It would seem as though Radeon finally has an opening when the next generation of GPUs hit the market, but the reason for it isn't going to be just software, as one would imagine. Consumer market sentiment shifts around software now and then, but the key decisive factor between whether Nvidia or AMD leads the race is economics, and the economics of graphics memory seem to have collapsed the entire battlefield into one segment where Radeon has always fought best. The DLSS 5 backlash is as real as it gets Upscalers should run games fast, not run different ones altogether Ever since the RTX 20-series was released, Nvidia's software stack has been one of the most compelling arguments for choosing GeForce over Radeon, and the value proposition was simple enough for the average gamer. DLSS made games run faster and look better, which is what anyone assembling a build for gaming would ever want. The Radeon lineup offered comparable rasterization capabilities and comparable VRAM configurations for a slightly lower price, and yet, the software suite around RTX cards always seemed to tip the purchase decisions in Nvidia's favor. Around the Turing era, the unique selling point for Nvidia was ray-tracing capabilities. Around the Ampere era, it was DLSS 2.0, and by the time Ada Lovelace came around, the argument shifted to DLSS 3 and Frame Generation. These have all been examples of features that consumers have felt compelled to pay an Nvidia premium for over the past six years in the industry. In Blackwell and the rumored upcoming Rubin era, Team Green has attempted to extend the very same logic to its absolute limit with DLSS 5 and neural rendering, and so far, it hasn't been received well. TechPowerUp recently polled its readers on neural rendering, and nearly 20,000 votes came back with 58% saying AI should not alter games at all. The disapproval rates were similarly high in PC Guide's poll, with the majority voting against the future Nvidia has planned for rendering. What appears as an obvious outlier in the pattern that Nvidia has repeated for close to half a decade now, is that never in the history of its unveiling has a new rendering technique met with such strong public disapproval. The results of both polls came a month before Nvidia's SIGGRAPH 2026 presentation, which, with the introduction of masking and Model sliders, arguably felt like timely damage control. Clearly, it seems, that a majority of gamers are unwilling to pay the Nvidia tax for the next generation if the headline product of the generation is neural rendering, and so far, there seems to be no evidence of another major feature with similar disruptive potential in the works. The DRAM inflation pushed the high-end out of reach for many And that leaves two contestable segments left Now, there's one thing to consider before I make my argument. AMD buys memory from the same Samsung, SK Hynix, and Micron oligopoly that Nvidia does, so that means it buys memory at the same inflated prices. That's exactly why it has raised its own GPU and memory prices this year as Nvidia has. Therefore, the VRAM pricing is just as detrimental for Team Red as it is for Team Green, in effect being an equalizer. Where VRAM pricing hits the hardest, is undoubtedly, in the flagship segment that belongs to the RTX 5090-class GPUs. This also happens to be the one tier where Radeon does not compete. AMD hasn't introduced an RTX 5090 competitor this generation, and if the rising memory costs continue to push the flagship costs into unviable territories, it isn't losing a customer it was competing for. As the cost of building a high-end GPU rises, the amount consumers are willing to spend on it doesn't necessarily rise linearly, and it's more likely that a consumer budgeting for an $800 GPU will start looking at a $600 segment instead. In this case, one could reasonably expect more buyers to tend towards the upper-mid-range market, and that's precisely the segment where AMD has historically shipped its greatest products. One needs to look no further than the case of the RX 9070 XT. There's only one thing AMD needs to do And that's something it has been doing all along In the current generation of Radeon graphics, AMD didn't field a product against the RTX 5090, and instead concentrated its strongest GPU around the upper mid-range instead. This also happens to be a segment where another $100 becomes a more significant buying decision for the person making the purchase, and that's something that has been true long before memory economics began distorting buyer decisions. The RX 9070 XT demonstrated why the strategy worked before, and I would argue it serves to demonstrate in the current economic landscape that the PC hardware industry finds itself in. The 9070 XT wasn't the fastest card of the generation, and neither was it the most generous when it came to VRAM allocation. It offered enough performance, enough VRAM, and a price that was undercut to be compelling enough to make the closest GeForce competitor difficult to justify. Radeon has already caught up in enough of the areas that matter to the average consumer for the price to become the key decisive factor. AV1 encoding with B-frames, day-zero support for local AI models, and FSR 4.1 are all capabilities the Radeon GPU already has today, before its next generation even arrives. In that way, all that's left for AMD is to build a better mid-range GPU to disrupt the market. The next GPU war is coming to AMD's turf, and could be fought on AMD's terms Nvidia seems to have handed AMD a rare opportunity to deliver what the average buyer wants from a next-gen GPU, and sure enough, even the market seems to be moving closer to where Radeon has always been the strongest. If that is indeed what happens in the next 12–18 months, AMD only needs to win where most consumers can afford to buy.