During GTC 2026, NVIDIA didn’t just talk about DLSS or traditional visual improvements. The company introduced a much deeper and potentially disruptive approach: integrating neural networks directly into game rendering with Neural Texture Compression. The goal is clear, drastically reduce the weight of textures and accelerate material calculations, two of the most costly elements in modern graphics engines.
Neural Texture Compression : Textures reduced by over 6 thanks to AI
Where traditional methods like BCn store compressed complete textures, NVIDIA proposes a completely different approach: storing a compact representation and letting a small AI reconstruct the texture in real-time on the GPU.
The gain is spectacular. In a presented scene, a classic version occupied around 6.5 GB of VRAM. With neural compression, this same scene dropped to only 970 MB. This represents a reduction of about 85%, a huge number in a context where video memory has become a limiting factor in many recent games.
Neural Texture Compression : NVIDIA aims to revolutionize gaming from within: AI lightened textures and materials up to 7.7x faster
Beyond memory space, NVIDIA also claims to improve visual quality. With equivalent memory budget, AI-compressed textures show fewer artifacts and retain more details than traditional methods. For gamers, this could mean lighter games, faster installations, and better graphics quality without a spike in hardware requirements.
Neural materials for accelerating rendering
The other presented innovation concerns materials, a key element of rendering that determines how light interacts with surfaces. Today, these materials rely on complex and costly computational models, often composed of multiple layers and textures.
NVIDIA proposes an alternative with Neural Materials. The principle is similar: replace these complex structures with a compact neural representation capable of reproducing the material’s visual behavior with much less data.
In the demonstration, a traditional material using 19 channels is reduced to only 8 channels in its neural version. This simplification significantly lightens the computational load while maintaining faithful rendering.
An AI integrated directly into the graphics pipeline
One of the most important points of this approach is its direct integration into the rendering pipeline. Unlike technologies like DLSS, which intervene at the end of the chain on a 2D image, these solutions act upstream, at the heart of image generation.
The networks used are deliberately simple, of the MLP type, in order to be executed millions of times per image without skyrocketing costs. They are designed to work directly on the GPU, exploiting Tensor Cores, which allows maintaining high real-time performance.
Towards less demanding games… or even more ambitious ones
With these technologies, NVIDIA is not simply looking to improve existing aspects but to redefine a portion of modern graphics rendering. By reducing memory consumption and computational costs, developers could either lighten hardware requirements or push the level of detail and complexity of games even further.
One major unknown remains: how they will be used. As usual, these gains could be used to optimize performance or increase visual ambitions, further pressuring hardware. One thing is certain, with Neural Texture Compression and neural materials, NVIDIA is already paving the way for a new generation of graphics engines.
