alt Hacker NewsTSMC can only make about as many Nvidia chips as OpenAI and the other AI guys wants to buy. Nvidia releases gpus made from basically the shaving leftovers from the OpenAI products, which makes them limited in supply and expensive.
So gamers have to pay much more and wait much longer than before, which they resent.
Some youtubers make content that profit from the resentment so they play fast and loose with the fundamental reasons in order to make gamers even more resentful. Nvidia has "crazy prices" they say.
But they're clearly not crazy. 2000 dollar gpus appear in quantities of 50+ from time to time at stores here but they sell out in minutes. Lowering the prices would be crazy.
Replies
Yes. In 2021, Nvidia was actually making more revenue from its home/consumer/gaming chips than from its data center chips. Now 90% of its revenue is from its data center hardware, and less than 10% of its revenue is from home gpus. The home gpus are an afterthought to them. They take up resources that can be devoted to data center.
Also, in some sense there can be some fear 5090s could cannibalize the data center hardware in some aspects - my desktop has a 3060 and I have trained locally, run LLMs locally etc. It doesn't make business sense at this time for Nvidia to meet consumer demand.
This is one reason, and another is that both Dennard scaling has stopped and GPUs hit a memory wall for DRAM. The only reason AI hardware gets the significant improvements is that they are using big matmuls and a lot of research has been in getting lower precision (now 4bit) training working (numerical precision stability was always a huge problem with backprop).