10x the digital world at least. Actually, that might be low if we consider the number of GPUs being installed each year, the improvement in algorithms, along with the increase of data especially synthetic.
All of this will keep multiplying at a rate that far exceeds Moore's Law. This means that Jenson Huang's projection of a 1 billion X in something like inference is one tap.
Spot on. With GPUs scaling, algorithms optimizing, and synthetic data exploding, we could see inference capabilities hit a billion X sooner than we think. The AI flywheel is spinning faster than anyone can predict