I think the Tensor decision was very impactful to both the direction of AI tech and Nvidia as a company, though. I can't judge how much of the decision to invest in that technology was his but assuming he had a significant proportion of influence in that decision, he deserves a significant portion of the credit. A different person in the same position might not have seen the potential of machine learning at scale and not been willing to point their GPUs at the problem.

They basically invested in pickaxes a decade before the gold rush, and people only found out about the gold because pickaxes were pretty convenient

Not really. Nvidias gpu architecture was tensor.. It is useful for other things like AI. That wasn't a decision to create AI other than it was what they had. They landed on their feet. They didn't invest in pickaxes, they had the right pickaxe at the right time.