Liang Wenfeng, co-founder of High Flyer hedge fund, boldly shifted focus to developing artificial general intelligence (AGI) despite skepticism and significant challenges, including hardware restrictions imposed by geopolitical tensions. Through heavy investment in GPU resources, innovative training methods, and unwavering commitment, Liang transformed his AI research lab into DeepSeek, positioning it as a formidable player in the AI industry.
The story begins with the co-founder Liang Wenfeng stepping away from his previous ventures, surprising many who thought his interest in AI was merely a hobby. In 2023, Liang announced a bold new direction for his hedge fund, High Flyer, shifting its focus toward artificial general intelligence (AGI). AGI aims to achieve human-level intelligence capable of outperforming humans across various domains, not just financial trading. This ambition was considered audacious, as AGI development was largely the domain of major tech giants like Google, OpenAI, Meta, Baidu, and Alibaba.
Liang faced skepticism, especially regarding the feasibility of startups competing in the foundational AI model space. Foundational models require immense resources, and experts often discouraged smaller teams from attempting to build them. Despite this, Liang believed in pushing boundaries and encouraged others to try, even while acknowledging the difficulty. AGI, being an even more complex challenge beyond foundational models, had yet to be successfully developed by anyone, and American companies were still far ahead in this race.
Recognizing the critical role of hardware in AI progress, Liang invested heavily in acquiring thousands of Nvidia A100 GPU cards to support his AI research. Many viewed this as a reckless move, but Liang was undeterred. He established a dedicated research lab to pursue his AI ambitions, treating what was once a hobby with newfound seriousness and commitment.
When former President Biden imposed restrictions on China’s access to American GPUs, many thought Liang’s efforts would be halted. However, Liang turned this obstacle into an opportunity by innovating his training processes to reduce dependency on GPUs. This adaptability allowed him to continue advancing his AI research despite significant geopolitical and technological challenges.
Ultimately, Liang’s perseverance paid off as he spun off his research lab into an independent company named DeepSeek. From humble beginnings and against considerable odds, DeepSeek emerged as a serious contender in the AI landscape, driven by Liang’s vision to push the boundaries of AGI and foundational AI models beyond what was previously thought possible.