DeepSeek is Making Waves in the AI World
DeepSeek, a Chinese artificial intelligence (AI) lab, has recently gained a lot of attention after its chatbot app climbed to the top of the Apple App Store charts. This sudden popularity has led many financial experts and tech enthusiasts to wonder if the United States can keep its leading position in the AI field and whether there will continue to be a high demand for AI hardware, like computer chips.
But what exactly is DeepSeek, and how did it become so well-known so quickly?
The Origins of DeepSeek
DeepSeek is supported by High-Flyer Capital Management, a Chinese hedge fund that uses AI technology to help make investment decisions. The company was co-founded by Liang Wenfeng, an AI enthusiast who started trading while studying at Zhejiang University. In 2015, he helped establish High-Flyer, which officially became a hedge fund in 2019, focusing on creating and using AI algorithms for trading.
In 2023, High-Flyer launched DeepSeek as a separate lab to explore AI technologies beyond just finance. With backing from High-Flyer, DeepSeek became its own company.
From the beginning, DeepSeek set up its own data centers to train its AI models. However, like many other Chinese tech companies, it faced challenges due to U.S. restrictions on exporting certain hardware. For instance, to develop one of its latest models, DeepSeek had to use Nvidia H800 chips, which are less powerful than the H100 chips available to U.S. companies.
The team at DeepSeek is reportedly quite young and actively recruits talented AI researchers from top universities in China. They even hire individuals without a computer science background to help the AI better understand a wide range of topics.
DeepSeek’s Impressive AI Models
DeepSeek introduced its first AI models—DeepSeek Coder, DeepSeek LLM, and DeepSeek Chat—in November 2023. However, it wasn’t until the release of the next-generation DeepSeek-V2 models in the spring of 2024 that the AI community began to take notice.
DeepSeek-V2 is a versatile system capable of analyzing both text and images. It performed exceptionally well in various AI tests and was much cheaper to operate compared to similar models available at the time. This competitive pricing forced other Chinese tech giants like ByteDance and Alibaba to lower their prices and even offer some of their models for free.
In December 2024, DeepSeek launched DeepSeek-V3, which further boosted the company’s reputation. According to DeepSeek’s own tests, DeepSeek V3 outperformed well-known models like Meta’s Llama and OpenAI’s GPT-4.
Another notable model from DeepSeek is R1, a “reasoning” model released in January 2025. DeepSeek claims that R1 performs as well as OpenAI’s o1 model on important benchmarks. What makes R1 special is its ability to fact-check its own answers, which helps it avoid common mistakes. However, reasoning models like R1 take a bit longer to provide answers—usually a few seconds to minutes longer than typical models. The trade-off is that they tend to be more accurate in subjects like physics and math.
There is a downside, though. Because DeepSeek’s AI is developed in China, it must comply with regulations set by the Chinese government. This means that certain topics, like the Tiananmen Square protests or Taiwan’s political status, won’t be answered by the chatbot.
A Unique Business Approach
DeepSeek’s business model is somewhat unclear. The company prices its products and services much lower than its competitors and even offers some for free. DeepSeek claims that breakthroughs in efficiency allow it to keep costs down, although some experts question the accuracy of these claims.
Regardless, many developers are embracing DeepSeek’s models. While they aren’t completely open-source, they are available under licenses that permit commercial use. For example, the CEO of Hugging Face, a platform that hosts DeepSeek’s models, noted that developers have created over 500 variations of the R1 model, which have collectively been downloaded 2.5 million times.
DeepSeek’s success against larger, well-established companies has been described as revolutionary for the AI industry, leading to significant market shifts. For instance, DeepSeek’s achievements contributed to an 18% drop in Nvidia’s stock price and prompted a public response from Sam Altman, the CEO of OpenAI.
As for DeepSeek’s future, it remains uncertain. While improvements to their models are expected, the U.S. government seems increasingly cautious about what it sees as potentially harmful foreign influences in the tech space.
