When Liang Wenfeng returned to his rural Guangdong village for the 2025 Lunar New Year holiday, neighbors hung a red banner praising him as the community’s pride, while bodyguards stood watch, a first for that quiet farming town. Weeks earlier, an app from his little-known company had dethroned ChatGPT atop Apple’s US App Store, triggering a single-day loss of roughly $600 billion in Nvidia’s market value.
The DeepSeek founder has since become the subject of global business profiles portraying him as a reclusive architect of China’s most stunning AI breakthrough. His story defies the conventional Silicon Valley founder narrative. Liang never chased headlines. Instead, he spent nearly a decade accumulating computing infrastructure as a quantitative investor before the Western world ever learned his name.
Born in 1985 to schoolteacher parents in Guangdong province, Liang excelled academically and showed an early aptitude for mathematics. He attended Zhejiang University, earning both bachelor’s and master’s degrees in information and communication engineering. His postgraduate work focused on object-tracking algorithms using low-cost cameras, far removed from the AI systems that would later define his career.
The Liang Wenfeng career timeline started not with artificial intelligence but with financial markets. In 2013, he launched an investment firm using algorithmic trading methods. Two years later, he co-founded High-Flyer with university classmates, building one of China’s largest quantitative hedge funds with reported assets around $14 billion at its peak.
The decisive pivot came between 2019 and 2022. Anticipating US export restrictions on advanced chips, Liang directed approximately $139 million of High-Flyer’s trading profits into a supercomputing platform, amassing more than 10,000 Nvidia A100 GPUs while they remained available. He intended to build faster trading models. That decision instead positioned him with one of China’s largest privately held GPU clusters just as generative AI exploded.
In 2023, Liang spun off the AI research division into DeepSeek in Hangzhou, reportedly self-funding initial work with about $1.4 million. The company built its own large language models rather than licensing existing ones. DeepSeek’s V2 model launched in 2024, sparking a price war in Chinese tech. Then came R1 in January 2025, claiming performance comparable to leading Western systems despite restricted chip access and at a fraction of typical training costs.
Liang’s leadership style remains unconventional. He rarely gives interviews, avoids conference circuits, and hires mostly fresh graduates and PhD candidates rather than industry veterans. DeepSeek released several models with open weights, bucking the closed approach of most frontier labs. He funded early development from High-Flyer’s profits, retaining roughly 78 percent ownership even after later funding rounds.
DeepSeek closed a funding round of about $7.4 billion in June 2026, valuing the company near $50 to $52 billion. Bloomberg’s Billionaires Index now estimates Liang’s personal net worth at roughly $36 billion, while reports point toward a mainland IPO filing in 2026 and potential listing in 2027. A planned public offering would expose both DeepSeek and Liang to unprecedented disclosure requirements and scrutiny as competition inside China intensifies.
His trajectory, engineering student to quant fund manager to AI founder, reversed the standard tech disruptor path. Liang quietly built infrastructure for one purpose, then redeployed it for another, forcing the world’s biggest AI labs to recalculate what frontier models should cost.
