AI arms race: China one-ups the US with DeepSeek

28 Jan

DeepSeek, a Chinese tech startup, recently made headlines when it introduced its newest AI model, R1. This model was said to showcase high “reasoning” skills, rivalling American AI models such as OpenAI’s O1.

Nvidia is a widespread provider of graphics processor chips, and due to national security concerns, the American government limited and banned sales of its higher-performing A100 chips to China. However, despite this restriction, the success of DeepSeek in its use of lower-performing chips and its lower cost raises concerns for the AI market, where advanced hardware may not be necessary to innovate and develop AI to a higher level than it already is.

Companies like Google, Amazon, and Microsoft have been large investors in Nvidia’s AI systems. Still, when DeepSeek’s capabilities were revealed, Nvidia lost $600 bn in market value on the same day as Trump’s inauguration, and its stock price dropped 17%.

In line with its cost-effectiveness, DeepSeek is also revolutionary in its ability to perform ‘Test Time Scaling’, where models can think out loud and show their train of thought, using this knowledge to generate new data sources progressively. Although other US competitions could likely share similar capabilities, this marks one of the earliest times that a Chinese company could replicate and match this skill level quickly.

However, like other AI models, DeepSeek has flaws. It is very difficult to obtain opinions from these models, who do their best to be objective, unoffensive, and unpolitical. Additionally, unlike the US models, there are still complications regarding some topics related to China, such as the Tiananmen Square protests, which DeepSeek will refuse to answer.

President Trump describes this incident in the stock market as a “wake-up call for our industry”, where US tech stocks should work harder and more intensely to fight back against external threats like China. Nvidia has also shared an optimistic-seeming attitude, describing DeepSeek’s model as an “excellent AI advancement”, yet the future of the stock market remains uncertain.