
The competition between China and the United States is increasingly prominent, particularly in the realm of technological innovation. Recently, a significant development has emerged from China that has markedly disrupted the technology sector, specifically Silicon Valley. The Chinese enterprise DeepSeek Lab has unveiled its R1 AI model, which has garnered considerable attention worldwide and has the potential to surpass established artificial intelligence systems such as ChatGPT and Google Gemini.
Overview of DeepSeek AI
The DeepSeek AI tool operates on an advanced linguistic framework and employs a hybrid architectural model. Notably, the R1 model represents the third iteration in the company's series of AI models, designated as V3. Founded in 2023 by Liang Wenfeng, DeepSeek is headquartered in Hangzhou, China, with a mission focused on the development of Artificial General Intelligence (AGI).
Who is Liang Wenfeng
Liang Wenfeng grew up in the 1980s in a less-known city in Guangdong, China. Coming from a family of educators, he excelled in his studies and went on to attend Zhejiang University, one of the top universities in China, where he earned both his undergraduate and graduate degrees.
Career Shift from Finance to AI
In 2015, Wenfeng teamed up with two of his college friends to launch a hedge fund called High-Flyer. This fund used advanced mathematical techniques and artificial intelligence to make smart investment choices. By 2019, they had successfully managed over USD 10 billion in assets. During his time at High-Flyer, Wenfeng developed a strong interest in artificial intelligence, which led him to buy a large number of Nvidia GPU chips to build a powerful system for training AI models.
Despite being described as a bit quirky with an unusual hairstyle, Wenfeng’s ambitious ideas paved the way for a new project: DeepSeek.
Launching DeepSeek and its impact
In May 2023, Wenfeng officially introduced DeepSeek as an offshoot of High-Flyer, which continues to support the new AI lab financially. DeepSeek quickly gained attention after unveiling its V3 model in late 2024. They published a significant research paper in December, revealing that they trained this AI model using 2,000 Nvidia chips at a cost of less than $6 million—much cheaper compared to what competitors typically spend.
The R1 Model and its global significance
On January 20, 2025, DeepSeek made headlines again by launching its R1 model. This innovative model showcased impressive abilities and efficiency, sparking conversations about the growing influence of Chinese technology companies in the world of artificial intelligence.
ALSO READ: Google rolls out new theft protection feature, initially available on Pixel, Samsung smartphones