Fluid Intelligence: A Forward Look on AI Foundation Models in Computational Fluid Dynamics
Ashton et al.
What it does
Breaks industrial CFD down into its components and proposes a scaling law that counts the cost of generating data as well as training.
Why it matters
Shows when data generation rather than training dominates compute, argues that high-fidelity transient data is the route to a CFD foundation model, and gives public estimates of what one would cost.