Nvidia management: How can we not just double the need for AI chips, but make the demand exponentially greater? Hmmm ... here's an idea ...