Liang's career initially took root in the finance sector. In 2013, he co-founded Hangzhou Yakebi Investment Management Co., Ltd., which marked his entry into quantitative finance. This venture integrated AI with quantitative trading strategies, leveraging machine learning to predict market trends and make investment decisions. His success in this field led to the establishment of Zhejiang Jiuzhang Asset Management Co., Ltd. (formerly Hangzhou Huanfang Technology Co., Ltd.) in 2015, where he further solidified his reputation as a leader in China’s financial sector.In 2016, Liang co-founded Ningbo High Flyer Quantitative Investment Management Partnership. This firm quickly became a significant player in the finance industry, managing assets worth 10 billion yuan by 2019. High-Flyer was known for its innovative approach to quantitative trading, emphasizing the use of algorithms and data-driven methods over traditional human portfolio management.
Transition to AI
Liang's interest in AI began to take shape around 2019 when he founded High-Flyer AI, a venture focused on researching AI algorithms and their practical applications. This move was a natural extension of his work in quantitative finance, where AI played a crucial role in analyzing and predicting market behaviors.
Preview
In 2021, Liang started acquiring thousands of Nvidia GPUs, a move that initially puzzled his peers but later revealed his grand vision of creating transformative AI models. This led to the founding of DeepSeek in 2023, a company dedicated to developing artificial general intelligence (AGI). DeepSeek's mission was to challenge the dominance of Western tech giants by creating cost-effective and efficient AI models.
Preview
DeepSeek's Rise to Prominence
DeepSeek quickly gained momentum, releasing its first AI model, DeepSeek Coder, in November 2023. This was followed by the release of DeepSeek LLM, a 67B parameter model aimed at competing with other large language models. The company's V3 model, launched in December 2024, ran on 2,000 Nvidia H800 chips at a cost of less than $6 million, showcasing remarkable efficiency compared to industry standards.
Preview
Preview
Preview
Preview
The real breakthrough came in January 2025 with the launch of DeepSeek's R1 model. This model demonstrated advanced reasoning capabilities and was significantly cheaper to develop compared to similar models from US tech giants like OpenAI. The R1 model's success catapulted DeepSeek into the global spotlight, challenging Silicon Valley's dominance and reshaping the AI landscape.
Leadership and Vision
Liang's leadership style is characterized by a hands-on approach and a focus on long-term innovation over short-term profitability. He has been described as a visionary who emphasizes strategic vision, efficiency, and innovation. His decision to make DeepSeek's models open-source has been seen as a significant move to democratize access to advanced AI capabilities and foster global collaboration.Liang's influence extends beyond his companies. He has been invited to high-profile symposiums and has provided insights on China's innovation ecosystem, stressing the need for China to transition from being a technology consumer to a global contributor.
Personal Life
Liang is married to Zhang Mei and has two children. Despite his busy professional life, he remains deeply involved in his family and personal interests. His journey from a modest upbringing to becoming a leading figure in the AI industry is a testament to his dedication and visionary approach.
