DeepSeek's Liang Wenfeng: The Quant Trader Who Disrupted AI
From a Rural Village to the Cutting Edge of AI
Liang Wenfeng, the mastermind behind DeepSeek, is a name that is rapidly gaining traction in global AI circles. Born in 1985 in a rural village in China’s Guangdong province, Liang’s journey from an academically gifted student to a dominant force in quantitative trading and artificial intelligence is a case study in persistence, strategic thinking, and the convergence of finance and technology.
With humble beginnings, Liang’s early education revolved around mathematics. By the time he entered high school, he had already self-studied advanced mathematical concepts, setting the stage for an academic trajectory that led him to Zhejiang University, one of China’s top institutions, where he studied electronic information engineering. This background laid the foundation for his entrance into quantitative finance—a field where computational mathematics meets capital markets.
The Rise of a Quant Trading Maverick
By 2008, while pursuing his master’s degree, Liang became deeply involved in algorithmic trading. With an initial capital of just 80,000 RMB (about $12,000 at the time), he experimented with automated trading strategies in China’s A-share market. Over the next seven years, he reportedly turned this into a fortune, accumulating over one billion RMB ($140 million), a feat that positioned him as one of China’s most successful independent traders.
However, trading alone was not enough. In 2013, Liang and his partner, Xu Jin, institutionalized their strategies by founding Jacobi Investment in Hangzhou. Two years later, they rebranded as Phantom Technology , a quantitative hedge fund that rapidly scaled its assets under management . By 2017, Phantom had fully integrated AI into its trading models, an early adoption of machine learning in high-frequency trading that paid off handsomely. By 2019, Phantom’s AUM had exceeded 100 billion RMB ($14 billion), making it one of China’s largest quantitative hedge funds.
Between 2015 and 2017, Phantom refined its AI-driven high-frequency trading strategies, which became exceptionally well-suited to China’s stock market. By 2018, Phantom won the prestigious Private Equity Golden Bull Award, further solidifying its dominance. However, rapid success brought increased regulatory scrutiny, forcing Phantom to adjust its strategies.
Pivot to AI: From Trading to Deep Learning
Liang’s ambitions did not stop at finance. Recognizing the power of AI beyond trading, he embarked on an ambitious project to build one of China’s most powerful AI computing infrastructures. In 2019, Phantom invested 2 billion RMB ($280 million) in building "Firefly One" , a high-performance AI computing cluster equipped with 1,100 GPUs. This initiative expanded in 2021 with "Firefly Two," featuring 10,000 Nvidia A100 GPUs, positioning Phantom as a major player in AI infrastructure.
However, the regulatory landscape in China was shifting. By 2023, increased scrutiny on high-frequency trading forced Phantom to scale down its AUM from a peak of over 1 trillion RMB ($140 billion) to 450 billion RMB ($63 billion) by 2024. With its AI-driven trading strategies facing headwinds, Liang made a strategic pivot: leveraging Phantom’s compute power for general AI research.
DeepSeek: China's Open-Source Answer to OpenAI
In July 2023, Liang founded DeepSeek, a company dedicated to the development of large language models . The venture quickly made waves, and within a year, it released DeepSeek-V2, a mixed-expert LLM designed to rival Western AI models. On December 26, 2024, DeepSeek-V3 was released, gaining widespread attention due to its efficiency and cost-effectiveness—trained with just 2.8 million GPU hours at a cost of approximately 40 million RMB ($5.6 million).
But the real breakthrough came in early 2025 with the launch of DeepSeek-R1. Designed to compete directly with OpenAI's GPT-4 and Gemini models, DeepSeek-R1 demonstrated comparable performance in mathematics, coding, and natural language reasoning. Its most disruptive aspect? It was fully open-source.
This move sent shockwaves across the AI industry. While companies like OpenAI and Anthropic continued their walled-garden approach, charging premium prices for API access, DeepSeek took the opposite stance. By releasing model weights and training methodologies for free, it challenged the monopolization of AI by Western tech giants.
Beyond the technological and business implications, Liang made it clear that DeepSeek's approach was rooted in ideology. In a rare interview, he emphasized that China needed to stand at the frontier of technology, not just follow Western innovations. He described open-source AI as a cultural movement, arguing that proprietary AI models create temporary moats but ultimately hinder progress.
Public Perception and Controversy
Liang Wenfeng’s meteoric rise has not come without controversy. Public opinion about him remains sharply divided. While some hail him as a visionary disruptor, others view his journey with skepticism. Many critics argue that his success in quantitative trading was more about exploiting market inefficiencies rather than true innovation, labeling him as an opportunist rather than a pioneer. Some investors and analysts remain wary of DeepSeek’s aggressive open-source strategy, questioning whether it is a genuine push for democratizing AI or a calculated move to undermine Western tech dominance. Additionally, Liang's introverted and reclusive personality has fueled speculation about his motivations—some admire his dedication to technology, while others question his lack of engagement with public discourse. His critics also point to his high-frequency trading background as a red flag, suggesting that his AI ambitions could similarly prioritize financial gains over ethical considerations.
Implications for Global AI Development
Liang’s decision to open-source DeepSeek-R1 is not just a technological breakthrough—it’s a geopolitical statement. The AI race has long been dominated by American firms like OpenAI, Google DeepMind, and Meta. With China’s strict regulatory environment, many assumed that domestic firms would lag in foundational AI research. DeepSeek has proven otherwise.
For investors, this signals a major shift. China is no longer just following Western AI trends—it is setting them. The open-source approach could accelerate AI adoption across industries, from fintech to healthcare, creating investment opportunities in AI-powered applications, semiconductor manufacturing, and cloud infrastructure.
What’s Next for Liang Wenfeng?
Liang remains an enigmatic figure. Despite his success, he shies away from media attention, preferring to focus on research and development. His recent return to his hometown for Chinese New Year was met with grand celebrations, underscoring his rising status in China’s tech landscape. But beyond the accolades, his core motivation remains unchanged: pushing the boundaries of AI and technology.
The next frontier for DeepSeek likely involves scaling its AI capabilities beyond language models. With one of China’s most advanced AI computing infrastructures at its disposal, DeepSeek could venture into AI-powered robotics, advanced neural networks, or even AI-driven drug discovery.
Final Thoughts
Liang Wenfeng’s journey encapsulates the convergence of finance, technology, and artificial intelligence. From a self-taught quant trader to the founder of one of China’s most ambitious AI initiatives, his story is a testament to strategic foresight and relentless execution.
As AI continues to reshape global industries, DeepSeek’s rise serves as a reminder: the future of AI is not confined to Silicon Valley. The next major breakthrough could just as likely emerge from Hangzhou.