Has DeepSeek blazed a new trail in artificial intelligence (AI) model training? Are the start-up’s models superior to those developed by OpenAI or Meta Platforms? And has the Chinese company rewritten the scaling law and proved that Nvidia’s AI chips are unimportant in training AI models?
Advertisement
The answer to these questions is “no”, according to many technology researchers and experts who have sought to demystify the disruptor over the past two weeks. However, it would be a mistake to underestimate the significance of DeepSeek for China, as the implications of its achievements extend far beyond mere technological advancement.
DeepSeek is important because it has boosted China’s confidence and self-perception in a protracted technology rivalry with the US, and it has opened new possibilities for China to empower its already formidable manufacturing apparatus with low-cost AI solutions.
The start-up has dispelled frequent pessimism about China’s chances in overcoming US technology restrictions. Since US start-up OpenAI launched ChatGPT in late 2022, China has been backed into a corner. Washington has deprived China of Nvidia’s advanced chips, while some US tech titans have discriminated against China by willingly cutting its access to cutting-edge technologies.
Last July, for example, OpenAI banned developers in Hong Kong and mainland China from using its service, categorising China as part of a small group of “unsupported countries and territories” along with North Korea and Iran.
The breakthrough of DeepSeek stands in defiance against US hostility and a response to the arrogance of American tech companies, casting doubt on their valuations. As Silicon Valley investor Marc Andreessen described, DeepSeek represents a “Sputnik moment” for the US, reminiscent of when the former Soviet Union launched the first man-made satellite into space in 1957.
Advertisement