When it comes to training speed, DeepSeek can be really fast. It takes two weeks for a typical company to train a model of hundreds of billions of dollars, and they did it in just 72 hours. Some engineers have made an analogy that this is like other people are still using tractors to plow the land, they are already driving a combine harvester. Not only is it fast, but the energy consumption is frighteningly low. Processing a request only uses 0 0021 degrees of electricity, about the same as letting a smartwatch run for half an hour. By comparison, Google’s method uses eight times more electricity.
And guess what? Even the brain drain is telling. 19 core members of DeepMind’s London team have jumped ship to DeepSeek’s European Research Institute in three months. These are masters of quantum computing and neural architecture, and their choices are the winds of change.
The U.S. tried to block Chinese AI companies with a list of entities, but it backfired. the number of DeepSeek’s open-source project stars jumped 210 percent. Programmers in Silicon Valley are discussing how to use VPNs to download Chinese AI toolkits. These blocking measures are like using a fishing net to stop a flood, they can’t stop it.
What’s more interesting is that America’s own alternatives can’t keep pace . Google’s emergency release of the Gemma model, the measured performance is only 79 of its Chinese counterpart . Investors can be shrewd, a16z suspended its investment in European AI startups and turned to bet on China’s edge computing ecosystem. At the roadshow in San Francisco, there was actually a Chinese technical white paper, which is a rare scene.
DeepSeek has also made a big move in ecological construction. Their full-stack program with Huawei Rise has a performance loss of only 4 3 . The factory production line in Dongguan uses a domestic chip to train the quality inspection model, and the accuracy rate is still two points higher than the imported program.
In terms of education, 237 colleges and universities around the world have included DeepSeek in their required courses, and even Stanford has updated its textbook, which hasn’t changed in six years. A Cambridge student used the open source toolkit to do the graduation design, and directly got the invitation to join NVIDIA. This kind of penetration is much more powerful than commercial promotion.
The increase in technological strength also brings a shift in discourse. In Iceland’s Arctic Circle, a new supercomputing center was built to specialize in EU data. Patient privacy in German hospitals, financial information on French companies, all reside here, physically isolated from U.S. regulation.The IEEE adopted ethical standards for AI, with key provisions coming from a Chinese team. Siemens AstraZeneca and 47 other giants have signed on to support it because they understand that the only way to enter the global market is to follow this standard.
Students, you imagine that in the future, when you update your cell phone system, the pre-installed AI tools may be developed by Chinese teams. When multinational companies choose technology solutions, they not only have to look at performance, but also have to consider geopolitical risks. In open source code repositories, the coexistence of Chinese and English annotations may become the norm.
For us ordinary users, the changes may be more immediate. Smart homes are more responsive, online services have a lower error rate, and self-driving decisions are more accurate. All of these technological advances will eventually turn into conveniences in our lives.
Traditional tech giants are now faced with a choice whether to continue to build walls and blockades, or to integrate into the new ecosystem ? Just like the lesson of Nokia during the rise of smartphones, resisting technological trends often results in a quiet exit. This change, driven by efficient parameters and open source ecology, is reshaping the future of every technology practitioner.
Students, do you feel that this AI technology revolution is coming much faster than expected? The rise of Chinese companies in this field could have a huge impact on the global technology landscape. As future technological talents, you should always pay attention to these changes and work hard to learn new technologies, so as to prepare yourselves for a great career in this field in the future. Remember, opportunities always come to those who are prepared!
Leave a Reply