Post

Not sure I understand what I am looking at, but I do agree with your sentiments.

' A 'business model' based on a product that generates "plausible looking" code, code that _might work_, if someone knew how to fix it, and financed by selling dollars for a nickel apiece doesn't work? Who knew?

bu-bu-but The influx of "investor" cash into the US companies will go on forever. And they will be profitable some day. So there! 😁

How would AI be dumped by China? What would that look like?

Just for the fun of it: #predictions

The chip makers are doing great (see Micron’s stock, as an example), but core servers, storage, and networking costs have gone up literally 3x in price over the past 6 months. That is absolutely hammering IT budgets (combined with Broadcom’s VMware graft, it’s really bad).

Please provide more details/information to this post

The question is timing. I know this is the path ... the question is the speed of the crash.

I don’t know that this chart supports what you say it supports. China doesn’t seem to be dumping. US model consumption seems pretty stable and not necessarily growing very much for a while now. So why is a crash coming?

Started using mostly Chinese open source models as well since the start of this month but all of them hosted on European providers because of EU GDPR privacy rules. New Claude is amazing but only rarely needed.

Either that or by switching to smaller models, the built out infrastructure becomes sufficient for actual demand based on higher usage pricing. And all those data center projects become redundant. Which will devastate the book value of loans held be PE firms financing data centers.