Be informed about free updates
Simply log in to Artificial intelligence Myft Digest – delivered directly to your arrived mail.
The writer is the founder of the VC partner based in Hong Kong. Capital
Deepseek has forever changed the path of global rivalry in technology. In China, Founder Liang Weenfeng He became a local champion. For a country where foreign degrees-now, those in the United States are further perceived as prestigious than their domestic equivalent, students and parents were astonished to find that the Start-up artificial intelligence research team were educated in the country.
Beijing is more confident than ever in search of technology. The success of Deepseek is undermined by the obstacles created in the war in the USA.
The decision of the company based in Hangzhou to publish a cheap AI model with open sources, with a detailed detection of his training methods, means that everyone, from researchers in São Paul to Start-up in Stockholm and doctors, can access the state -t-t- ART AI by little and at no cost.
A chain reaction takes place within the Chinese Start-up sector. New AI apps are created. Competition will become fierce. The appetite for the risk of investment in the early phase increases. Deepseek’s decision to continue the Ai model of the open code inspires and puts pressure on others to do the same. The first to react was Alibaba Qwen team, which posted Qwen2 last month as an open source regarding Chinese New Year.
This is an extraordinary change. After US start-up openii published its generative AI model Chatgpt at the end of 2022, the global digital economy was surpassed according to the control of a handful of technological giants. These players are persecuting the ladder due to the efficiency-buildings of increasing models that require incredible calculation, energy and capital as they keep their training methods as trade secrets.
Centralized, closed models create a dangerous loop for feedback. The more information they collect, the more powerful they become, the further marginalize anyone outside their doors. For consumers, this means great fees, the data was handed over and the view of Ai’s future is taking place without meaningful participation.
The promise of Deepseek’s model R1 lies in its adaptability. Open source, this can be adapted to local needs. It avoids excess calculations using something called a rare neural network training, which means that its effectiveness reduces calculating and energy needs of size.
This means that advanced AI can benefit from mass, not just a few. This proves that the technology is goods. Billions of dollars do not need to be spent on competition between technological giants with closed models. The value of AI should not lie in ownership models, but in what we are all able to with it.
As an investor, I am worried that Deepseek could state the US to decide to decide even more stricter sanctions. In China, the export limits of graphic units for processing (GPU), such as powerful H100 NVIDIA, interfered with the growth of start-up. Financing foreign investors is limited because of the care of the risk of compliance. However, the actual danger lies in limiting access to global cooperation and research cooperation, which chokes the global flow of knowledge that is crucial to maintaining progress. Talent can bypass the lack of chips, but by setting up obstacles to learning the risk of long -term stagnation.
Still, even additional American limitations, conspiracy theories and smear campaigns target Deepseek cannot change the reality that Chinese start-up has put AI in the hands of humanity.
Against all noise, let’s consider this as a moment in history. In 1440. Johannes Gutenberg brought Europe printing, an invention that broke the monopoly to the knowledge previously held by the elite. Deepseek’s achievement joins this tradition to become more accessible information. His cheap reasoning model proves that AI can belong to everyone, not just those who are storage codes, chips and capital.
