During GDC 2026, Nvidia presented new progress in its Neural Texture Compression (NTC) technology, showing that the future of gaming depends not only on raw power but also on better optimization, where AI can play an important role.
Following the recent announcement of DLSS 5, the company explained in more detail how neural networks can change the way VRAM is managed, making it possible to reduce video memory usage.
Nvidia Highlights Neural Texture Compression
The most interesting part of the presentation was the Neural Texture Compression demo. In a test scene called “Tuscan Wheels,” Nvidia managed to reduce VRAM usage from 6.5 GB, using traditional BCN compression, to just 970 MB with NTC. This means an 85% reduction in memory use.
What is even more impressive is that this reduction does not come with a loss in visual quality. Nvidia said that, under the same memory usage, NTC keeps details much sharper than conventional methods.
By compressing assets more efficiently, this technology could also lead to smaller game installs, lighter patches, and lower download bandwidth requirements. It also allows GPUs with less memory to handle high-resolution textures that were previously impossible.
Nvidia also introduced the idea of “Neural Materials.” Instead of processing multiple heavy texture channels, the system uses a compact representation that a small neural network decodes in real time.
In testing, this reduced a 19-channel setup to just eight channels, while also improving rendering times by 1.4 to 7.7 times at 1080p resolution.
With these developments, Nvidia continues to focus on using AI to improve both game performance and image quality, something many players are hoping to see more of in the future.
