Chinese AI lab DeepSeek has made headlines after its chatbot app skyrocketed to the top of the Apple App Store and Google Play charts. The rapid rise of DeepSeek’s AI models, designed with compute-efficient techniques, has sparked debate among Wall Street analysts and technology experts. Many are now questioning the U.S.’s ability to maintain its AI leadership and the long-term demand for AI chips.
The Origins of DeepSeek
DeepSeek is backed by High-Flyer Capital Management, a Chinese quantitative hedge fund leveraging AI for trading strategies. The hedge fund was co-founded by AI enthusiast Liang Wenfeng in 2015. Wenfeng, who started experimenting with trading while studying at Zhejiang University, officially launched High-Flyer Capital Management in 2019 to focus on AI-driven investments.
In 2023, High-Flyer established DeepSeek as a separate AI research lab, spinning it off into its own company. DeepSeek developed its own data center clusters to train AI models, though it has faced challenges due to U.S. export restrictions on high-end chips. As a result, the company was forced to use Nvidia’s H800 chips, a less powerful alternative to the H100 chips available to U.S. firms.
DeepSeek has also built a reputation for aggressively recruiting AI researchers, particularly doctorate holders from top Chinese universities. The company even hires individuals without computer science backgrounds to enhance its AI models’ understanding of various subjects.
DeepSeek’s Powerful AI Models
DeepSeek introduced its first models—DeepSeek Coder, DeepSeek LLM, and DeepSeek Chat—in November 2023. However, the company gained significant attention in early 2024 with the launch of its next-generation DeepSeek-V2 family of models.
DeepSeek-V2, designed for text and image analysis, outperformed many competing AI models while being significantly more cost-efficient. This forced domestic AI rivals like ByteDance and Alibaba to lower their pricing and even offer some services for free.
The launch of DeepSeek-V3 in December 2024 further cemented the company’s reputation. According to internal benchmark tests, DeepSeek-V3 surpassed both open-source models like Meta’s Llama and closed API models such as OpenAI’s GPT-4o.
Another major breakthrough came with DeepSeek R1, a reasoning-focused AI model introduced in January 2025. DeepSeek claims R1 matches OpenAI’s o1 model in performance. Unlike typical AI models, R1 employs self-verification techniques to improve accuracy, making it more reliable for scientific, mathematical, and physics-related tasks. However, this reasoning process slightly increases response times.
Despite its strengths, DeepSeek’s models are subject to Chinese government regulations. For instance, the chatbot app censors responses on politically sensitive topics such as Tiananmen Square and Taiwan’s independence.
A Disruptive Business Model
DeepSeek’s business strategy remains unclear, as the company prices its services significantly below market rates and even offers some for free. Unlike many AI startups, it has also declined venture capital funding despite heavy interest from investors.
DeepSeek attributes its cost efficiency to advanced AI training techniques, but some industry experts question the accuracy of these claims. Regardless, developers have embraced its models, which, while not fully open-source, are available under permissive licenses for commercial use. On Hugging Face, developers have created over 500 derivative models of DeepSeek R1, with a combined 2.5 million downloads.
Impact on the AI Industry
DeepSeek’s rapid success has shaken the AI industry, prompting reactions from major tech players. The company’s disruptive approach contributed to an 18% drop in Nvidia’s stock price in January 2025 and sparked responses from OpenAI CEO Sam Altman. Microsoft, meanwhile, added DeepSeek to its Azure AI Foundry service, integrating the company’s models into its enterprise AI offerings.
During Meta’s Q1 earnings call, CEO Mark Zuckerberg reaffirmed that AI infrastructure spending remains a strategic advantage for the company. Meanwhile, OpenAI labeled DeepSeek as “state-subsidized” and “state-controlled,” advocating for potential U.S. government restrictions on its models.
Nvidia CEO Jensen Huang acknowledged DeepSeek’s technical innovations, noting that its reasoning models require significantly more computing power—potentially benefiting Nvidia in the long run.
Global Bans and Restrictions
Despite its impressive technological advancements, DeepSeek faces growing restrictions from governments and corporations. South Korea has already banned the AI models, while New York state has prohibited DeepSeek from being used on government devices. Reports indicate that the U.S. government is considering a similar ban, citing concerns over foreign influence and security risks.
What’s Next for DeepSeek?
DeepSeek’s future remains uncertain. While it continues to push the boundaries of AI technology, increasing regulatory scrutiny and geopolitical tensions could hinder its global expansion. Nonetheless, DeepSeek is expected to roll out improved AI models, further intensifying competition in the AI sector.
As the U.S. evaluates its stance on Chinese AI developments, the ongoing battle between DeepSeek and Western AI giants will shape the future of artificial intelligence and its regulatory landscape worldwide.
