Deepseek, the Chinese -based artificial intelligence attempt, shook financial markets and technology giants by making a quick entry into the artificial intelligence sector. The new Chatbot, developed by the company, was the most downloaded free application at the US Apple App Store this week, and OpenAI’s ChatGPT was on the throne. This artificial intelligence assistant, which Deepseek strengthened with its open -source models, draws attention with its allegations that it is educated at a much lower cost. This development led to a decrease in Nvidia shares exceeding 12 percent.
Behind Deepseek’s success is the new generation R1 model, which was available on January 20th. The company says that R1 is designed to solve complex problems and performs equivalent to OpenAI’s O1 in certain comparison tests. This new model attracts the attention of the sector, especially with the innovations it brings in cost and resource usage.
Deepseek’s R1 model was built on the V3 Language Model (LLM) announced in December 2024. The company claims that the total cost of V3’s development is only $ 6 million. In contrast, Sam Altman, OpenAI ‘CEO, stated that the GPT-4 trainee required a budget of more than $ 100 million.
Moreover, it is stated that only 2000 nvidia chip to train Deepseek’s model. This figure is quite low compared to 16000 chips or more used by other major artificial intelligence companies. This led to the questioning of high processing power and cost -requiring approaches in artificial intelligence development processes. Despite the US trade restrictions implemented by the company by developing creative solutions in this field, the Company offers an effective model.
Companies such as Nvidia, Microsoft, OpenAI and Meta, known for its major investments in the artificial intelligence sector, have allocated hundreds of billions of dollars of budget for data centers in recent years. For example, it is stated that 100 billion dollars of investments, known as the Stargate project and whose total budget is estimated to be $ 500 billion, is only divided into Nvidia. However, Deepseek’s low -cost model launched a discussion that questioned the efficiency of these huge budgets. The shares of Nvidia, Microsoft and other major players decreased after these discussions.
Deepseek’s innovative approach, cost efficiency and resource use offers a different perspective in the sector. However, the company’s claims have not yet been confirmed by independent authorities. Nevertheless, the success of the company reveals that competition in the field of artificial intelligence is not only shaped by technology, but also with strategic cost management and innovative solutions.
It can ensure the redistribution of deepseek cards
Deepseek’s low -cost structure emphasizes efficiency in artificial intelligence development processes. Nvidia and other major players may have to review their current strategies. Deepseek, which develops creative solutions especially in the face of the US trade restrictions on Chinese -based companies, can create a new competition dynamics in this field.