The emergence of DeepSeek is definitely a "black swan" event in 2025.
This black swan has directly "blackened" the US stock market and broken the mindset of the subsequent development of artificial intelligence.
Previously, the former CEO of Google had said that the competition in artificial intelligence is a race with a threshold of hundreds of billions of dollars, and if you can't cross this threshold in terms of funding, you can't even get the entry ticket to the race.
Zhu Xiaohu, the founder of Morningside Venture Capital, had once said that he would not invest in any domestic large model companies, because he believed that domestic companies could never catch up with the US in the field of large models.
The reason why these two people made such assertions is that there has always been a basic assumption in the AI competition: to win the competition, especially the competition in the underlying models, computing power and capital are an insurmountable threshold.
The enterprises in our country are stuck on computing power - unable to obtain the best NVIDIA graphics cards.
The most direct shock brought by the emergence of DeepSeek to the whole world is: it has achieved excellent performance at an extremely low cost.
This achievement has directly thrown the biggest "shock wave" into the current global artificial intelligence competition:
If performance improvements and cost reductions can be achieved through algorithm improvements and optimizations, do we still need to make such huge investments in computing power and energy to continue to promote and develop artificial intelligence applications?
Therefore, the founder of "Black Goku" praised DeepSeek as a national-level achievement.
However, no matter how outstanding DeepSeek is, it is still the result of development on the shoulders of giants, as Yang Lekun said:
This is the victory of open-source large models over closed-source large models.
I believe that DeepSeek has three very far-reaching impacts on the subsequent development of artificial intelligence:
First: It has greatly reduced the cost of large models, which means that countless small and medium-sized companies may join the competition of AI large models, and develop specialized large models in various sub-fields and industries.
Second: The wave of artificial intelligence startups in China will explode - far exceeding what we see now.
Third: The artificial intelligence startup teams in China will be looked upon with new respect, and they may once again attract the attention of the global venture capital community.
The success of DeepSeek means that the computing power threshold is not as high as imagined. Just after DeepSeek went online, many laboratories, including universities in Hong Kong, have successfully replicated the results of DeepSeek.
We can fully imagine that a large number of companies will further optimize the model based on the open source of DeepSeek, so that we can train good large models with not so top-of-the-line chips.
The first of these impacts is now interfering with the US stock market.
Due to the amazing results of DeepSeek in terms of cost and performance, the core companies in the artificial intelligence track are now facing a revaluation.
NVIDIA fell by more than 10% on Monday's opening. Other related companies are also facing similar situations.
I immediately thought of Tesla.
Previously, I always thought that Tesla's visual solution was likely to be the ultimate solution for autonomous driving. Tesla's advantage in this solution is large models and data.
Now with the emergence of DeepSeek, Tesla's advantage in large models may not be so obvious anymore, and it only has the advantage of leading in data.
But how long this leading advantage can be maintained depends on the subsequent development of other electric vehicle companies.
Back to the crypto ecosystem.
On the surface, the emergence of DeepSeek has pulled down both the US stock market and the crypto ecosystem, especially the AI agent track. Many people think that the AI agent track has collapsed. But I think this actually shows from another angle that the competition in the AI agent track is far from over.
Because if large models can iterate so quickly, it means that the leading advantages currently held by existing AI agents are only temporary, and if they cannot keep up with the trend and the iteration of technology in the future, they may be overthrown by the newcomers at any time with the latest technology and the latest models.
At the same time, this also means that the R&D threshold of AI agents has been further lowered, and more participants can participate in the development of AI agents, which also means that domestic Chinese teams have a chance to shine on this competitive field.
Today is the last article of our Dragon Year. After the Spring Festival of the Year of the Snake, we will meet with you again. I hope that in the new year, we can continue to maintain enthusiasm and continue to find our own love and harvest our own gains in this industry.
