According to DeepSeek it can train a programme at a cost of $5.6 million compared to the hundreds of millions cited by top US AI developers and, if true, could be a game-changing breakthrough that threatens the dominance of US Big Tech.
The company’s programme called R1 is intended for complex problem solving, and was trained using just 2000 Nvidia GPUs compared to the many thousands typically used by AI programme developers
DeepSeek also has a programme called V3 and both were rated in the top ten AI models on the University of California at Berkeley’s AI rating service, Chatbot Arena.
Some analysts are already suggesting that Deepseek is going to challenge Silicon Valley’s leadership, disrupt the global tech landscape and reshape the direction of the AI arms race.
It is certainly true that US Big Tech, long considered untouchable, is facing credible and intensifying competition from China but more worryingly for the US, China’s technological advances are fast eroding the US’s ability to use tariffs as a tool to maintain its global technological supremacy.
The rise of DeepSeek highlights the accelerating pace of the global ‘AI arms race’, with China now firmly establishing itself as a serious competitor.
For companies like Alphabet, Meta, and Apple, the emergence of formidable competitors such as DeepSeek, will also raise some serious questions about their ability to sustain the valuations and dominance they’ve enjoyed for so long.
