We're used to the forms of hype led, "tech miracle" investor stripper scams now. The resemblance of this cycle to previous cycles is strong; giving a feeling of "surely people won't bit the same bait
again?"
Under all the hype, there are small but real advances. The "VR" wave led to some better sensors, the "self driving cars" wave led to cheaper LIDAR modules, and a lot of experience with approaches that fail.
You'd have never have sold those as the end products for the price that was paid for them, however. The "LLM" wave of AI is as much a product of Nvidia's need for a market as it is any actual advance or novelty in that field. Ask: where's the patents on LLMs?