Two years after Chinese technologist and AI researcher Lieng Wen Feng founded DeepSeek as an AI research company — with backing from High-Flyer Capital Management a hedge fund he co-owns — DeepSeek has grown to be the most discussed tech force this week.

DeepSeek first entered the AI scene in November 2023 with a series of models (DeepSeek Coder, DeepSeek LLM, and DeepSeek Chat), but it wasn’t until December 2024 that it truly captured global attention with the release of DeepSeek-V3. Then, on January 20, the conversation hit a fever pitch with the release of DeepSeek R1 “Reasoning,” a model claimed to exceed OpenAI’s o1 on key metrics.

The rollout of new AI models is almost always an incremental improvement over existing ones. But DeepSeek’s sudden dominance isn’t just about abilities, it’s largely about what went into achieving them: significant lower capital (and time) investment compared to industry-leading models and the implications of this: China potentially displacing the United States as the number one force in the artificial general intelligence (AGI) race.

According to Aljazeera, DeepSeek’s new R1 model was developed with less than $6 million, in contrast to OpenAI CEO Sam Altman’s revelation that GPT-4 cost over $100 million to train. Even more disruptive is DeepSeek’s minimal reliance on Nvidia chips — a vital AI-developing component of AI inventions. DeepSeek claims to have used just 2,000 specialised Nvidia chips to train V3, compared to the 16,000 or more that leading models reportedly require, according to the New York Times.

These developments led Nvidia’s shares to plunge by 17%, resulting in the world’s most valuable company losing $589 billion in market capitalisation. This is the largest single-day value loss by any US company in history.

This is DeepSeek’s doing, but perhaps the biggest contributor to its status as the topic du jour over the past week has been the reaction within the American AI community, which has long led the AGI race, as it unseats OpenAI from the top of the Apple App Store rankings.

President Donald Trump has issued a call to action to American technologists, saying, “The release of DeepSeek AI from a Chinese company should be a wake-up call for our industries that we need to be laser-focused on competing.”  

Similarly, OpenAI CEO Sam Altman has acknowledged the impressiveness of DeepSeek’s latest model given its limited resources but has promised to deliver much better models than its new competitor.  

Meanwhile, Meta CEO Mark Zuckerberg appears unfazed by DeepSeek’s rise. The Meta boss says the company is planning to invest “heavily” in AI, with billions set aside for future models, particularly its anticipated Llama 4 model.

No country has challenged the US this aggressively in the AI sector since the industry’s boom. DeepSeek is expected to maintain its momentum, even as reports that it had secretly leveraged OpenAI’s data to enhance its models continue to gain ground.