#DeepSeek冲击全球算力 #DeepSeek冲击全球算力 really makes you guys go crazy, the entire Binance Square is filled with similar press releases: From a 90% reduction in training costs, we've won again, the US stock market is doomed, and ultimately it concludes that the cryptocurrency circle is also being crushed, which is absurd beyond absurdity. Today, let's objectively critique Deepseek:

First, talking about Deepseek itself, Deepseek explores the possibility of optimizing training frameworks and model designs under the extreme shortage of computing power to complete expected tasks. Most importantly, it opens up the final results back to the community, which is much better than many closed-source companies that create technical barriers. It also punctures the price perception shaped by companies like OpenAI (gpt-o1 requires a $20/month membership to use, while pro requires $200). Due to insufficient model data and the loss of certain accuracy from extremely optimized training algorithms, Deepseek's overall user experience still has some shortcomings, but its inference capability has already matched that of o1.

Next, let's talk about financial issues. In the short term, it is definitely negative for NVIDIA because AI companies will pause their aggressive purchasing strategy for computing power and instead return to model tuning. Investors will also reassess the investment return ratio for AI. However, in the long term, it is still positive. If a company develops cheaper training cards, then it makes sense to say that's negative, but saying this is also negative makes no logical sense and just focuses on winning. Here are two small stories: