← Back to timeline
Keynote2026-07-20

The speaker states how much faster an AI surrogate model predicts results for geometries it has not previously seen compared with running the simulation.

Loading moment…

Neil Ashton · NVIDIA · NVIDIA Keynote Live at SIGGRAPH 2026

It is more than 10,000 times faster to do the prediction of an unseen geometry that the model has not seen or a boundary condition, and achieve accuracy within one or two percent. It almost seems like magic, but it but it is true.

Transcribed from the source video. Automatic captions, corrected for known mis-transcriptions.

Key terms

boundary condition
The specified conditions at the edges of a simulated domain, such as inlet speed or temperature, that define a particular case.
unseen geometry
A shape or design the model was not trained on, used to test whether it generalises.

Educational summary of a public statement. Not investment advice.