Liang Wenfeng, the founder of the Chinese AI lab DeepSeek, has rapidly emerged as a pivotal figure in the global AI landscape. Transitioning from a highly successful career in quantitative finance as the co-founder of the hedge fund High-Flyer, Liang has steered DeepSeek to challenge established AI giants by producing powerful, open-source models with remarkable efficiency. His philosophy is rooted in a commitment to fundamental research, fostering young talent, and a belief that China must become a contributor to, not just a beneficiary of, global technological innovation.
Vision for AI and China's Role
Learnings:
- The Ultimate Goal is AGI: Liang's primary focus is on solving the hardest problems in AI, with the ultimate destination being Artificial General Intelligence (AGI).[1][2][3] He views Large Language Models (LLMs) as a critical and necessary step on this path.[2][3]
- China Must Innovate, Not Imitate: He strongly believes that China needs to shift from being a "beneficiary" or "free rider" on the innovations of others to becoming a primary "contributor" to global technology.[4][5][6]
- Innovation is Driven by Curiosity, Not Just Profit: True breakthroughs come from a genuine desire to create and explore the boundaries of technology, not solely from commercial incentives.[3][4][7]
- AI is in a Phase of Technological Explosion: Liang believes the current era is defined by rapid technological innovation in AI, not yet by a boom in applications. The priority should be on advancing the core technology.[3][7]
- Solving the Hardest Problems is the Core Driver: The fundamental motivation behind DeepSeek is to tackle the most difficult scientific questions in the world.[1]
Quotes:
- "Solving the hardest questions in the world—that's what drives us... And we're just getting started.”[1]
- "We believe that with economic development, China must gradually transition from being a beneficiary to a contributor, rather than continuing to ride on the coattails of others.”[5]
- “Over the past 30 years, we have been solely focused on making money, largely ignoring innovation. But innovation isn't purely driven by business; it also requires curiosity and a desire to create.”[3][7]
- "AGI is probably one of the hardest things we can do next, so for us it was a question of how, not why."[2]
- “We're done following. It's time to lead.”[8]
On Open Source and Competition
Learnings:
- Closed-Source is Not a Sustainable Moat: In an era of disruptive technology, keeping models proprietary offers only a temporary advantage.[4][9] The real competitive moat is a culture of continuous innovation.[9]
- Open Source Builds Ecosystems: Liang's strategy is to focus on foundational models and allow an ecosystem of B2B and B2C companies to build applications on top of DeepSeek's open technology.[3][7][9]
- Openness Fosters Achievement: Sharing research and models through open-sourcing doesn't mean losing a competitive edge; for technical talent, being followed and built upon provides a powerful sense of accomplishment.[9]
- Efficiency Can Outmaneuver Brute Force: DeepSeek has proven that clever engineering and architectural innovations can achieve top-tier performance without the massive budgets and computing resources of Big Tech.[1][10][11]
- Pricing Should Be Based on Cost, Not Hype: When DeepSeek drastically lowered its API prices, it wasn't intended to start a price war. The price was simply set slightly above their actual, much lower, operational costs.[3][4]
Quotes:
- "In the face of disruptive technology, closed-source moats are temporary."[4][9]
- "We embed value in our team. As colleagues grow and accumulate know-how, forming an innovative organization and culture becomes our moat."[9]
- "Giving is an extra honor. A company doing this has cultural attractiveness."[9]
- "We didn't intentionally become a disruptive force; we accidentally became one."[3][9]
- On pricing: “We didn't expect pricing to be such a sensitive issue. We simply followed our own pace, calculated costs, and set a fair price.”[3]
- "Even if OpenAI is closed-source, it can't stop others from catching up."[6]
On Company Culture and Talent
Learnings:
- Maintain a "Bottom-Up" Structure: DeepSeek operates without pre-defined roles. The division of labor and specialization emerge organically from the team's interests and collaborations.[8][9][11]
- Provide Unlimited Access to Resources: Researchers have unrestricted access to GPU clusters and other computing resources, allowing them to test ideas immediately without needing approval.[4][7][12]
- Trust in Young, Local Talent: DeepSeek's breakthroughs have come from young, passionate graduates from top Chinese universities, prioritizing their potential and confidence over extensive industry experience.[1][4][8]
- Innovation Requires Confidence: Liang believes a key gap between China and Silicon Valley was a lack of confidence in pursuing frontier research. He fosters an environment where young people are empowered to try ambitious ideas.[8]
- Stay Lean and Focused on Research: The company deliberately avoids the bureaucratic bloat of larger corporations by remaining steadfastly focused on R&D rather than building out large commercial or application teams.[3][12]
Quotes:
- "DeepSeek, like Phantom Quant, is bottom-up. We don't pre-assign roles; division of labor happens naturally."[9]
- "There are no limits on accessing computing resources. If someone has an idea, they can immediately access GPU clusters without needing approval."[7]
- “Why is Silicon Valley so innovative? Because they dare to try. When ChatGPT debuted, China lacked confidence in frontier research. But innovation requires confidence, and young people tend to have more of it.”[8]
- "Each person has their own unique learning journey and brings their own ideas — we don't need to push anyone."[7]
- “We want to create an ecosystem where industry uses our technology. We focus on basic models and frontier innovation, while others build B2B and B2C businesses on our foundation.”[9]
From Quantitative Finance to AI
Learnings:
- Technology-Driven Innovation is a Universal Principle: The success of his quantitative fund, High-Flyer, strengthened Liang's confidence that technology and data-driven approaches could be applied to solve complex problems in any field, including AGI.[9]
- The Transition Was an Unconventional Move: Liang acknowledges the unusual nature of a hedge fund effectively pivoting to burn its resources on fundamental scientific research.[1]
- Early GPU Investment Was Strategic: His decision to begin stockpiling Nvidia GPUs in 2021, before US export restrictions, was driven by a long-term curiosity about AI's potential and laid the foundation for DeepSeek's computational power.[9][13][14]
- Finance Was a Stepping Stone to a Harder Problem: While finance was a complex field to master, Liang viewed the pursuit of AGI as one of the most challenging and meaningful things his team could tackle next.[2]
Quotes:
- "It was as if Jane Street decided to become an AI startup and burn its cash on scientific research."[1]
- "Many assumed there was hidden business logic behind hoarding GPUs, but it was simply driven by curiosity about AI's capability boundaries."[9]
- “From a young age, I was fascinated by the intersection of mathematics and technology. Quant trading was a natural fit, but I always knew there was more to explore.”[8]
- Comparing the acquisition of GPUs to buying a piano: You do it “because you can afford it” and “because you have a group of people who are eager to play music on it.”[15]
- "Whenever I face difficulties at work, I remember [Jim] Simons' words: ‘There must be a way to model prices.’”[9]
On Business and Strategy
Learnings:
- Money is Not the Main Bottleneck: Liang has stated that DeepSeek's primary challenge is not a lack of funding but the US export restrictions on high-end chips.[4][7]
- More Investment Doesn't Always Equal More Innovation: He cautions against the idea that simply pouring more money into a problem will lead to better results.[7]
- Focus is Key: DeepSeek will not prematurely focus on building applications on top of its models; the core mission is to perfect the foundational models themselves.[2]
- Constraints Drive Innovation: Operating under US chip restrictions has forced DeepSeek to focus on efficiency and algorithmic advancements rather than relying solely on scaling up hardware.[16][17]
- Avoid Overthinking the Business Model: When the end goal is improving societal efficiency through technology, getting bogged down in transitional business models can lead to confusion. The focus should be on the core mission.[9]
Quotes:
- "We have no short-term fundraising plans. Our main problem has never been money — it's the export restrictions on high-end chips."[7]
- "More investment doesn't always lead to more innovation."[7]
- “The market is changing.”[18]
- “If the end goal is improving social efficiency, it's valid. Everything else is transitional - overthinking leads to confusion.”[9]
- “An exciting thing cannot be measured purely by how much its worth.”[15]
Sources
- [1] Fortune India: "How Liang Wenfeng built DeepSeek into a global AI powerhouse"
- [9] Substack: "30 Essential Quotes from DeepSeek's Founder: A Vision for AI, Innovation, and China's Tech Future"
- [4] Reddit: "Inside DeepSeek's Bold Mission (CEO Liang Wenfeng Interview)"
- [7] Substack: "Inside DeepSeek — An Interview with Founder Liang Wenfeng on AI Disruption and AGI"
- [2] LessWrong: "Two interviews with the founder of DeepSeek"
- [12] Hacker News: "Interview with DeepSeek Founder: We're Done Following. It's Time to Lead"
- [3] Medium: "Interview with DeepSeek Founder Liang Wenfeng: 'Our Destination is AGI, Not Just Making a Quick Buck'"
- [19] SEO.AI: "Who is the Founder of DeepSeek AI, Liang Wenfeng?"
- [13] Wikipedia: "Liang Wenfeng"
- [5] CNA: "China's new face of AI: Who is DeepSeek founder Liang Wenfeng?"
- [10] Medium: "The Man Behind DeepSeek — How Liang Wenfeng is Reshaping AI"
- [6] AGI: "Liang Wenfeng's Bold AI Vision: Open, Smart, Original"
- [18] Entrepreneur: "'Pride of His Hometown': Who Is DeepSeek Founder Liang Wenfeng?"
- [11] Smart Living: "What's the story of DeepSeek and its founder Liang Wenfeng?"
- [14] AIDigitalX: "Liang Wenfeng Biography, DeepSeek's Founder"
- [15] MoneyWeek: "Deepseek's Liang Wenfeng: the maths whizz who shook Big Tech"
- [8] Frederick AI: "Founder Story: Liang Wenfeng of DeepSeek"
- [16] GOLDSEA: "Liang Wenfeng Develops Low-Cost AI Models that Outperform US Giants"
Sources
