I can’t wait for the spectacular implosion
Yes, my point is that the compute from those chips can still be used. Maybe on actually useful machine learning tools that will be developed latter, or some other technology which might make use of parallel computing like this.
I know of at least one company that uses cuda for ray-tracing for I believe ground research, so there is definitely already some usefull things happening.
I mean there are a lot of applications for linear algebra, although I admit I don’t fully know in what way “AI” uses linear algebra and what other uses overlap with it.