Does the long term and increased usage of CFD imply that we no longer need wind tunnels to investigate flow fundamentals, but just to check more complex geometry? If computers become significantly faster, will CFD be used more, and wind tunnels less?