Chinese startup DeepSeek has made waves in the global AI sector with its innovative models, DeepSeek-V3 and DeepSeek-R1, which it claims rival or surpass the performance of leading U.S. models, such as OpenAI's ChatGPT, at a fraction of the cost. These models have already gained significant traction, with DeepSeek-V3 powering the company's AI Assistant, which recently overtook ChatGPT as the top-rated free app on Apple’s App Store in the U.S.

The models have drawn praise for their performance and cost-efficiency. DeepSeek claims that its DeepSeek-R1 model is 20 to 50 times cheaper to use than OpenAI’s GPT-4, depending on the specific task. This has raised questions about the AI industry's future, particularly regarding the massive investments made by U.S. tech giants in AI development, including those in Nvidia’s chips, which are central to AI model training. DeepSeek's cost advantages could upend the current technological balance, especially as Silicon Valley engineers and executives have taken notice of the startup’s progress.

However, some skepticism surrounds DeepSeek’s achievements. Some industry experts, such as Scale AI CEO Alexandr Wang, have raised questions about DeepSeek's claim of using a large number of Nvidia H100 chips, which might violate export control regulations that restrict the sale of these advanced chips to Chinese companies. DeepSeek has yet to respond to these allegations, adding fuel to the debate about its rapid rise.

Company Background: DeepSeek was founded in 2023 by Liang Wenfeng, the co-founder of a Chinese hedge fund called High-Flyer. High-Flyer, which focuses on quantitative trading, has shifted its resources toward artificial general intelligence (AGI) development, with DeepSeek emerging as part of this broader strategy. It is unclear how much investment High-Flyer has put into DeepSeek, but the hedge fund reportedly owns patents related to AI chip clusters, which may play a role in the training of DeepSeek’s models.

Political Significance in China: DeepSeek’s success has not gone unnoticed in China’s political circles. On January 20, 2023, Liang attended a symposium hosted by Chinese Premier Li Qiang, signaling that DeepSeek’s advancements may align with Beijing's broader goals of achieving technological self-sufficiency and overcoming U.S. export restrictions. This suggests that DeepSeek’s innovations could play a crucial role in China’s push to compete globally in AI and other strategic industries.