Who Is Liang Wenfeng, the Founder of DeepSeek?
The High-Flyer quantitative fund manager who built DeepSeek into a global artificial intelligence lab.
Liang Wenfeng is the founder and chief executive officer (CEO) of DeepSeek and the co-founder of the quantitative hedge fund Ningbo High-Flyer Quantitative Investment Management Partnership.
At Layer3Labs, we build custom artificial intelligence (AI) agents for business workflows, and DeepSeek's models are among those a client team can pick to run an agent on.
Liang co-founded High-Flyer in February 2016 with two engineering classmates, then announced in May 2023 that the fund would pursue artificial general intelligence (AGI) and launched DeepSeek.
Who Liang Wenfeng Is
As of October 2026, Liang runs DeepSeek as CEO. He is in the news for DeepSeek's V4 model releases and for a 2026 funding round that The Information put at more than $7.4 billion. In September 2026, Anthropic said DeepSeek, Moonshot AI and Xiaomi carried out unauthorized distillation of its Claude models.
According to Wikipedia, Liang personally holds only 1% of DeepSeek directly, while the quantitative fund entity Ningbo High-Flyer holds the remaining 99%. In July 2026, Bloomberg and the Financial Times reported that DeepSeek began preparatory work for an initial public offering (IPO) that could occur as early as 2027.
- Current role: Chief Executive Officer of DeepSeek and co-founder of Ningbo High-Flyer Quantitative Investment Management Partnership.
- Organizations: DeepSeek and Ningbo High-Flyer Quantitative Investment Management Partnership.
- Known for: Founding DeepSeek, whose R1 model topped US iOS App Store free downloads in January 2025, and co-founding the quant fund High-Flyer.
- Education: Zhejiang University, Bachelor of Engineering in Electronic Information Engineering (2007) and Master of Engineering in Information and Communication Engineering (2010).
- In the news for: DeepSeek's V4 model releases, a 2026 funding round of more than $7.4 billion (per The Information), and Anthropic's September 2026 distillation accusation, to which DeepSeek did not immediately respond.
Education and Quantitative Trading at High-Flyer
Liang earned a Bachelor of Engineering in Electronic Information Engineering from Zhejiang University in 2007. He continued his graduate studies at the same institution, receiving a Master of Engineering in Information and Communication Engineering in 2010.
In February 2016, Liang co-founded Ningbo High-Flyer Quantitative Investment Management Partnership with two engineering classmates.
According to Wikipedia, citing reporting by Chinese technology outlet 36Kr, High-Flyer acquired 10,000 Nvidia A100 graphics processing units (GPUs) before the United States government established semiconductor export controls.
Founding and Corporate Structure of DeepSeek
In May 2023, Liang announced that High-Flyer would pursue artificial general intelligence and initiated the DeepSeek project. Two months later, on 17 July 2023, DeepSeek was formally spun off as an independent company based in Hangzhou, per Wikipedia. High-Flyer was its principal backer.
Per Wikipedia, Liang told 36Kr that the AI price war in China set off by the DeepSeek-V2 release came as a huge surprise to him.
The DeepSeek-R1 Launch and 2025 Symposiums
DeepSeek launched the DeepSeek-R1 reasoning model on 20 January 2025. By 27 January 2025, DeepSeek topped the free downloads chart on the US iOS App Store.
On the day R1 launched, 20 January 2025, Liang attended a state symposium with technical experts hosted by Chinese Premier Li Qiang, as noted on Wikipedia.
Four weeks later, on 17 February 2025, Liang participated in a symposium with private-sector enterprises hosted by General Secretary Xi Jinping. In August 2025, Time included Liang on its list of the 100 Most Influential People in AI, while Fortune named him to its roster of the 100 Most Powerful People in Business.
Funding Rounds, Valuation, and Personal Net Worth
The Information reported, via coverage by Silicon Republic on 16 June 2026, that DeepSeek raised more than $7.4 billion at a post-money valuation exceeding $50 billion. According to The Information, Liang personally invested about $2.94 billion in that round. External participants included Tencent, which contributed around $1.5 billion, Alibaba, CATL, and the state-backed National Artificial Intelligence Industry Investment Fund.
Wikipedia gives different figures for the 2026 round: a May 2026 Series A of $7 billion at a $52 billion post-money valuation. On 6 August 2026, PYMNTS, citing Bloomberg, reported that DeepSeek sought to raise close to $8 billion at a valuation near $74 billion. That financing effort had been paused in late July 2026 after viral comments on American-Chinese AI competition were widely attributed to Liang, according to PYMNTS. As of October 2026, no outlet had reported that the roughly $8 billion round closed.
On 14 July 2026, Bloomberg published an assessment estimating Liang's personal net worth at $36 billion. Bloomberg's estimate made Liang the richest founder of an AI-model company in the world.
The V4 Flagship Series and Distillation Allegations
On 10 April 2026, AIbase reported that Liang had stated in an internal communication that V4 would be released in late April 2026. DeepSeek published technical previews of the 284-billion-parameter V4-Flash and 1.6-trillion-parameter V4-Pro on 24 April 2026.
StartupHub.ai reported that DeepSeek-V4-Pro has 1.6 trillion total parameters, 49 billion active parameters per token, and a standard one-million-token context window. According to Wikipedia, V4-Flash launched officially on 31 July 2026, followed by V4-Pro on 13 August 2026 and V4.1-Flash on 10 September 2026. For a technical analysis of model routing and inference latency, see our review of DeepSeek V4.1-Flash explained.
On 10 September 2026, Anthropic claimed that DeepSeek, Moonshot AI, and Xiaomi carried out unauthorized distillation of its Claude model family, as reported by Bloomberg on Yahoo Finance. DeepSeek representatives did not immediately comment on the allegations. Organizations assessing data governance should consult our overview of DeepSeek data privacy and security risks.
Research Philosophy and Public Commentary
In a 2024 interview with Chinese publication Waves (36Kr), translated on Substack, Liang outlined his research thesis: "The real gap lies between originality and imitation."
In the same interview, Liang described open-sourcing this way: "Open source is more of a cultural act than a commercial one."
Related Founders and DeepSeek Alternatives
Teams evaluating these models against Western alternatives must weigh licensing terms, hosting costs, and hosting locations, as analyzed in our guide to DeepSeek alternatives.
Other Chinese AI founders we profile include Yang Zhilin at Moonshot AI and Junyang Lin, the former Qwen technical lead at Alibaba who founded Pragmatik Labs in 2026. All three appear on our list of AI people to watch.
Frequently Asked Questions
- Liang Wenfeng is famous for founding DeepSeek, the company behind the DeepSeek-R1 model and the V4 series. He also co-founded the quant fund Ningbo High-Flyer, which, per Wikipedia, had acquired 10,000 Nvidia A100 GPUs before US chip export restrictions and holds 99% of DeepSeek.
- Bloomberg estimated Liang Wenfeng's net worth at $36 billion on 14 July 2026, ranking him as the world's richest AI-model founder.
- Liang Wenfeng studied engineering at Zhejiang University. He earned a Bachelor of Engineering in Electronic Information Engineering in 2007 and completed a Master of Engineering in Information and Communication Engineering in 2010.
- Yes. As of October 2026, Liang Wenfeng is the founder and CEO of DeepSeek. In April 2026, AIbase reported his internal note that V4 would be released in late April, and The Information reported he personally invested about $2.94 billion in DeepSeek's 2026 funding round.
- Liang Wenfeng has not published a specific timeline or prediction for when artificial intelligence might take over human tasks or achieve artificial general intelligence. He announced in May 2023 that High-Flyer would pursue artificial general intelligence, and in his 2024 Waves interview he spoke about originality, imitation and open source rather than about dates.
Evaluate Foundation Models for Enterprise Workflows
Speak with our technical team to assess model reliability, hosting architectures, and operational security before deploying autonomous agents.
Book a Consultation