.png){kind=logo}
Chinese AI startup DeepSeek is climbing as a game-changer in the artificial intelligence industry, challenging the monopoly control of US tech giants including OpenAI and Nvidia. DeepSeek stands apart for its cutting edge AI models, cost effective training methods, and rapid growing global adoption transforming how artificial intelligence operates.
DeepSeek is an AI platform specialising in building advanced models of natural language processing, data analysis and machine learning. DeepSeek-V3 has attracted significant attention to its flagship model that provides performance close to those established US giants–OpenAI and Meta–at only a fraction of the cost.
The DeepSeek-V3 took less than US$6 million to train using Nvidia H800 chips, highlighting its efficiency. DeepSeek's DeepSeek-V3 powered AI Assistant has soared to the top of Apple's App Store in the US ahead of OpenAI's ChatGPT app.
The DeepSeek-R1 model, the latest release by DeepSeek has cemented its place in the industry too. The company said that DeepSeek-R1 is 20 – 50 times cheaper per use than OpenAI's GPT-3 model, depending on what the task is. For that reason, this cost effectiveness is redefining the economics of AI adoption and making DeepSeek a significant player in this field.
Liang Wenfeng, co-founder of the quantitative hedge fund High-Flyer, founded DeepSeek in Hangzhou. High-flyer has disclosed a strategic change from trading to Artificial General Intelligence (AGI) in March 2023. As part of this new vision, DeepSeek was developed.
DeepSeek’s success could be drawn to High-Flyer, a company holding patent chip clusters used to train AI models. The fund's AI division operates a cluster of 10,000 A100 chips — underscoring the scalability of technological capability of the firm.
DeepSeek’s achievements have garnered attention form China’s political leadership as well. The alignment of DeepSeek's success with Beijing's ambition for self-sufficiency in AI was signaled when founder Liang Wenfeng participated in a Chinese premier Li Qiang's symposium on January 20.
Innovations from DeepSeek are critically leveraged to China’s larger ambition to remove US export restrictions in critical technology sectors, and to gain competitive advantage in those areas.
DeepSeek is recasting the face of AI. DeepSeek, DeepSeek V3 and DeepSeek R1 is challenging the American technological superiority narrative and questioning the cost efficiency of US AI projects. DeepSeek is a long way behind, but it is poised to be a major force in the global AI arms race.
