nVidia has some SERIOUS R&D in the use of AI for the past 10 years.
But using AI in the graphic space... upscaling, downscaling, faking lighting, faking physics..
This is all very useful in making videogames.
Then there was a leap in the way AI Image generation was done with the above hardware. And that opened up a whole new growing field.
It's just some people took basic language models that have been around for 30 years and scaled them up with their hardware.
And it was neat, and surprising some of the stuff a LLM would output. But not reliable.
And then suddenly a lot of layman's got their hand on the LLM's and thought it was the 2nd coming of Jesus, and started throwing big money at it.. it will be surprise to no one who knows how these AI's work that that big money isn't going anywhere.
But those first two, is no hype. It's a real viable use case for the AI, and money will be made there.
Even with Trump, the US military complex will just NOT let TSMC to be taken away from them.
But as soon as foundries are done and producing on US soil... then ya.. pray for Taiwan, because US won't have a good reason to risk direct conflict with China over that island.
I'm not defending nVidia's business practices at all.
My point is the 'AI' hype isn't hype.
There's real value added AI work being done outside of the ChatGPT LLM thing going on.