Impressive results: Nvidia presents a possible solution to the VRAM crisis

High-resolution textures, larger open worlds and more advanced lighting systems are steadily driving up the VRAM demands of modern games. At the same time, memory capacity on many mainstream graphics cards has stagnated for years – or at least has not kept pace with rising game requirements. One example is Nvidia’s mid-range GeForce RTX 5060, which comes with 8 GB of VRAM and currently sells for around $350 on Amazon. Given the ongoing memory crunch, that is not great news for gamers. With Neural Texture Compression and Neural Materials, Nvidia has now presented a possible solution to the problem at GTC 2026.

With NTC, textures are no longer stored in the traditional way, but instead compressed into a compact neural representation. A small neural network running on the GPU reconstructs the required image data in real time. In a demo featuring a Tuscan villa, memory usage dropped from 6.5 GB to just 970 MB while image quality remained nearly identical, according to Nvidia. Those are impressive results that could genuinely help ease the VRAM problem. On top of that, games could become smaller, updates could require less data and downloads could become more efficient overall.

Neural Materials works in a similar way, but focuses on material properties such as multi-layered surfaces with different lighting responses. Nvidia says the technology could deliver speedups of up to 7.7x at 1080p. However, this approach appears to be less mature so far.

The gaming community has reacted to Nvidia’s presentation with mixed feelings. Unlike more controversial AI techniques such as DLSS 5, the new technology is largely seen in the YouTube comments as a meaningful step forward. Some, however, view NTC as little more than a stopgap for a problem Nvidia created itself and say the company should simply equip its graphics cards with more VRAM instead.