Despite American efforts to limit China’s access to advance technology, Deepsek turned the challenge into a chance. The company with the help of many times less resources to Deepsek Developed That it can take a direct competition from AI products of Openai, Google, Microsoft and Meta. The New York Times AccordingEngineers of Deepseek -V3 used much less special computer chips compared to their American competitors – which are essential for AI’s training. These chips have been the main focus of the technical race between the US and China. The US imposed several restrictions on exports to maintain its AI lead, but the New York Times says the success of Deepsek shows that Chinese researchers have become more innovative with their resources.
Deepseek-V3 allegedly performed well in works such as answering questions, solving logic problems and preparing code. There is even more impressive cost, with reports that Deepsek spent about $ 6 million on computing power, while META allegedly used $ 60 million for its latest AI product. While leading AI companies often use supercomputers with 16,000 or more chips, Deepcic reportedly uses only 2,000 Nvidia chips.
Not only this, Deepsek can be more efficient in the coming time, as the developers have adopted the open-source path with it. Dipcik has open-sourced its latest AI system, which shares the code with others. This promotes global cooperation, as noted by Yineng Zhang, a software engineer at Basten.