DeepSeek: the AI Startup Sets the World Talking

Chinese AI startup DeepSeek surprises Silicon Valley and the world by introducing a highly energy-saving model at cheap costs of production. 

DeepSeek challenged experts’ traditional views on AI as a field reliant on the most powerful chips, deemed essential by experts. The sudden rise of the Chinese DeepSeek AI questioned the dominance of the U.S. in manufacturing the world’s finest chips. DeepSeek attempts to question one’s understanding of AI: what if its complex nature could be reduced to the simplest components and a minimum level of resources? 

Similar to ChatGTP, DeepSeek is a generative and free AI-powered chatbot that simulates human reasoning. Not only does DeepSeek have accurate sources of information, it is also extremely successful at reducing the cost of production, placing it as a tough competitor against American AI. Instead of investing money on developing chips, DeepSeek trains AI with techniques that are highly energy-saving and efficient. For example, when given a problem, the chatbot utilizes a specialized sub-model to search for the answer rather than using the entire model. DeepSeek developers claim that “it cost $6 million to train, a fraction of the over $100 million alluded to by OpenAI CEO Sam Altman.” According to BBC, DeepSeek’s founder reportedly stockpiled Nvidia A100 chips, which have been banned from export to China since September 2022. By creatively pairing these chips to less sophisticated ones, Liang, the founder of DeepSeek, built a model even more powerful than ChatGPT.

Using 2048 H800 chips and investing only $500 million in hardware, DeepSeek challenged the position of OpenAI, which has the world’s most esteemed database. Indeed, this situation implies that banning exports failed to limit China's attempts to build AI models. DeepSeek has hit the top of Apple store downloads, significantly sinking U.S. tech stocks. The U.S. immediately sensed the threat. President Trump called DeepSeek’s release as a “wake-up call” for U.S. companies. DeepSeek’s capability to influence the U.S. stock market is stunning. It created the biggest one-day loss as chip-making giant Nvidia shed almost $600 billion of its market value on Monday. In an interview with the company’s founder last year, Liang stated that China’s AI sector “cannot remain a follower forever,” suggesting that rather than hard skills, the true difference between Chinese and American production lies in “originality and imitation.” The prospect of DeepSeek’s staggering success surprised Silicon Valley, as it has always viewed Chinese tech companies as followers rather than innovators. 

DeekSeek makes clear that the real challenge for the AI industry isn’t simply expanding the budget, but finding ways to achieve higher efficiency with limited resources. While the rise of DeepSeek is a serious threat to U.S. tech companies, it could also be an opportunity to test their creative problem-solving skills to counteract competition. For instance, it could ‘democratize’ the field of AI by making the industry more accessible to small firms. The consequent ‘democratization’ would only accelerate the control and dominance of AI, such that perhaps one day it would achieve equality with the human mind.

DeepSeek adds more uncertainty to the future of AI and the optimal way of attaining computing power from computing chips. In the words of BBC, DeepSeek’s achievements undercut the belief that bigger budgets and top-tier chips are the only ways of advancing A.I. A multifaceted problem-solving approach would be a valuable resource to many companies, not just those working on AI. 

Lucy Wang