So this proved to be BS. Maybe for NVDA for now, but the entire market is just too unstoppable to care.
DeepSeek is considered a potential "Black Swan" event by some market analysts due to its sudden and disruptive emergence in the AI industry. This unexpected development, characterized by its low cost and high efficiency, is seen as a significant threat to established companies like Nvidia and has caused notable market reactions, including a sell-off in tech stocks and AI-related cryptocurrencies. This market impact underscores the potential of DeepSeek to reshape the AI landscape and influence economic sectors globally beincrypto .
That guy had absolutely no evidence and looks like technically is not even competent to judge DeepSeek. His only argument is, the Chinese lied before. Well, so did Musk or any US start up. Anyhow, even if DeepSeek uses 100 times more chips than they claim, they are still spending at least a magnitude LESS, than their US competitors. So the truth is probably somewhere in between, but in practical tests, DeepSeek was sleek.
Yeah, maybe "this time it's different. I guess if Elon Musk can lie, it's a good idea to base your financial decisions on claims that need approval from the CCP. https://www.promptfoo.dev/blog/deepseek-censorship/
I didn't say they didn't lie. I said: 1. The guy didn't prove it. 2. Even if they spent 600 mill on it, that is still 10% what American companies are spending on it. Or help me out, that 500 Billion Stargate project is how many times more than 600 million?
The video referenced this. https://thatstocksguy.substack.com/p/a-few-thoughts-on-deepseek That doesn't prove anything either, but there is evidence. https://www.ft.com/content/a0dfedd1-5255-4fa9-8ccc-1fe01de87ea6 So, yes, if DeepSeek took shortcuts such as repackaging something else, it would cost a lot less than other AI developments. My conclusion is the hype about DeepSeek might be too good to be true.