DeepSeek Overthrows American AI Giant?---Who Would Have Thought the Boss of DeepSeek is A Post-85 Prodigy from China
深度求索迎战美国AI巨头?
Source: North America International Student Observer
【Editor’s Note: Liang Wenfeng（梁文锋）, a visionary Chinese entrepreneur born in 1985, has disrupted global AI and quantitative investing with groundbreaking innovations. From humble beginnings in Zhanjiang, Liang’s academic excellence led him to Zhejiang University, where he developed a passion for AI and machine learning. His career took off during the 2008 financial crisis when he co-founded High-Flyer, a quantitative investment firm that leveraged AI to revolutionize trading strategies. At its peak, High-Flyer became one of China’s top firms, managing over 100 billion yuan.
In 2023, Liang founded DeepSeek, aiming to break global monopolies in AI with cost-effective, high-performance large language models. By 2024, DeepSeek-V3 stunned the industry by matching OpenAI’s GPT-4o at one-tenth the cost, earning acclaim as a “mysterious force from the East.” Liang’s commitment to open-source innovation and long-term progress underscores his vision of making AI accessible, positioning him as a symbol of China’s technological ascendancy.】
A Post-85 Prodigy from China Overthrows an American AI Giant! The Boss of DeepSeek Is…
Source: BackChina.com
2025 marks a striking beginning, as the Prime Minister’s first New Year symposium with experts and entrepreneurs welcomed the presence of an extraordinary young man—Liang Wenfeng, born in 1985, founder of DeepSeek (深度求索) and High-Flyer (幻方量化). Not only is he a “king” in the realm of quantitative investing, but also a pioneer in artificial intelligence. Under his leadership, a team of developers took merely 2,048 GPUs and 53 days to create the world-leading large-scale model “DeepSeek-V3,” undercutting America’s OpenAI in a surprising “technical upset” at a fraction of the cost.
At the symposium, Liang presented his unique perspective on developing domestically produced large-scale AI models. Quiet yet potent, he symbolizes China’s emerging technological force. From an ordinary teenager in Zhanjiang to an AI innovator who shook Silicon Valley, his story is about technology, boldness, and tenacity.
01 From a Gifted Youth: Zhanjiang to Zhejiang University
Born in 1985 in a modest household in Zhanjiang, Guangdong—a third-tier city with slow economic growth—Liang Wenfeng was raised in an environment shaped by both wisdom and resilience, largely thanks to his father, an elementary school teacher. From a young age, Liang showed an exceptional knack for mathematics and science, with sensitivity to numbers and logic far surpassing his peers, quickly earning the label of “prodigy” among teachers and classmates.
In 2002, at the age of 17, Liang’s outstanding test scores propelled him into the electronic information engineering program at Zhejiang University—making him the first in his family to attend a top-tier institution. At the time, Zhejiang University was renowned for its strength in science and engineering, and electronic information engineering was one of its signature disciplines. This setting offered Liang a fresh platform to expand his horizons. Over four undergraduate years, he not only studied diligently but also actively competed in various math modeling contests, gradually making a name for himself in academic circles.
After receiving his bachelor’s degree, Liang decided not to seek immediate employment but continued at Zhejiang University to pursue a master’s, shifting his focus from electronic information to the more cutting-edge realm of machine vision. Under the mentorship of Professor Xiang Zhiyu, an authority in the field, Liang considerably elevated his academic prowess, gaining deeper exposure to AI technologies.
Yet unlike many classmates seeking what might be deemed “high-paying, comfortable jobs,” Liang refused to confine himself to the role of an ordinary engineer. Alongside academic research, he monitored larger industrial transformations, particularly AI’s real-world applications. He believed technology would serve as the bedrock of future societies, and he aimed to seize this timely opportunity.
His determination to “stay ahead of others” led him, while still at university, onto a path filled with challenges. Liang and a team of kindred spirits explored using machine learning to analyze financial market data, venturing even under the looming shadow of the global financial crisis to test fully automated quantitative trading. These bold, pioneering pursuits would later form the basis of Liang’s future endeavors and pave the way for his emergence on the global tech stage.
02 The Rise of Quantitative Investing: Legendary Beginnings at High-Flyer
The year 2008 brought a global financial crisis, shaking traditional financial markets and opening doors for technology-driven quantitative trading. While still a graduate student at Zhejiang University, Liang recognized how data and technology could reshape financial markets. Teaming up with like-minded classmates, he harnessed machine learning to analyze market data, experimenting with an automated quantitative trading system. They collected massive troves of market figures and macroeconomic information, applying mathematical models to track price fluctuation patterns. These experiments not only furnished him with practical know-how but also laid the groundwork for future start-ups.
In 2013, Liang and a fellow Zhejiang alumnus, Xu Jin, co-founded Hangzhou Jacobi Investment Management Co., Ltd., marking Liang’s formal entry into the world of quantitative investing. At the time, quantitative investing was still in its infancy in China, yet Liang’s steadfast trust in technology and his deep market understanding led him to believe firmly it was the finance sector’s path forward.
By 2015, China’s quantitative investing industry reached a turning point with the launch of CSI 500 Index Futures, which gave private quantitative funds richer hedging tools and ushered in a “2.0 era” for the sector. That same year, Liang, 30, teamed up with Xu Jin to officially found High-Flyer (幻方量化). Their vision was to harness mathematics and AI to become one of the world’s top hedge funds.
In its earliest phase, armed with just 10 GPUs, High-Flyer completed numerous product designs and launches. During the 2015 “stock market crash,” Liang’s high-frequency strategies delivered noteworthy returns and drew in considerable capital and clientele. Within a year, High-Flyer deployed 20 products, swiftly expanding its AUM. This technology-driven efficiency allowed High-Flyer to stand out in quantitative investing.
In 2016, Liang led the creation of a pioneering AI model powered by deep learning while integrating GPU technology into trading calculations—a first in China’s quantitative field. By the end of the year, High-Flyer’s AUM had passed 1 billion yuan. Over subsequent years, Liang continuously ramped up AI research, weaving AI into the heart of the firm’s strategies and gradually phasing out conventional models.
In 2018, High-Flyer obtained China’s top private securities accolade—the Private Fund Golden Bull Award—highlighting its industry leadership. Yet the team encountered computational bottlenecks. Liang’s devotion to technology spurred further investment in Firefly One, a deep-learning platform that furnished substantial computing resources for High-Flyer’s AI.
By 2019, High-Flyer oversaw more than 10 billion yuan in assets, fully entering the ranks of China’s elite quantitative funds. In August 2021, the firm’s AUM soared beyond 100 billion yuan, making it one of the nation’s “Big Four Quantitative Kings.” That year, speaking at the Golden Bull Award ceremony, Liang declared, “The future of quantitative investing lies in using technology to improve market efficiency.”
But behind such glory lurked challenges. At the end of 2021, market volatility and strategic miscalculations caused a drawdown for High-Flyer, prompting the team to scale back its AUM and halt new product launches. Though initially controversial, the move showcased Liang’s acute focus on risk control.
That same year, Liang’s team secured access to NVIDIA A100 GPUs ahead of many cloud service firms. Its AI cluster featured as many as “ten thousand” GPUs. According to reports, fewer than five entities in China possessed over 10,000 GPUs at that time, and aside from High-Flyer, the other four were internet giants.
Though High-Flyer’s scale dipped slightly, Liang’s faith in technology never wavered. Through the practical lens of quantitative investing, he proved the far-reaching impact of technology on finance. High-Flyer’s accomplishments also gave him the capital and resources to pursue new opportunities in AI.
03 Bridging Into AI: The Emergence of DeepSeek
In 2023, then 38-year-old Liang once again seized the crest of technological waves, this time focusing on the promising yet demanding field of artificial intelligence. He founded DeepSeek (深度求索), a firm committed to developing large language models. His objective was to dismantle the global AI monopoly with cutting-edge methods at minimal cost.
Liang knew AI well. In High-Flyer’s rise, AI was at the heart of its trading strategies. Now, however, he sought to step beyond finance and channel that same passion into AI’s large-scale model research. His approach was clear—innovate to keep costs low, thereby broadening AI’s reach. “We’re not looking to follow,” he emphasized, “but to genuinely surpass in core technology.”
In 2024, DeepSeek made a resounding debut. Within a year of its founding, the company’s innovative efforts made it a star in AI. In May 2024, DeepSeek introduced its first “mixture-of-experts” model, DeepSeek-V2, stunning the market with its ultra-low pricing: 1 yuan per million tokens for inputs, 2 yuan for outputs—just one percent of GPT-4 Turbo’s price. Its potent performance-to-cost ratio ignited a worldwide “price war” among large-model producers, forcing tech behemoths at home and abroad to slash their fees.
Yet the real breakthrough arrived in December 2024, when Liang and his engineers unveiled DeepSeek-V3. Besides stellar performance in math calculations, code generation, and Chinese Q&A tasks, the model’s training cost was a mere $5.576 million—compared to the $78 million necessary for OpenAI’s GPT-4o. Matching GPT-4o at merely one-tenth the price, DeepSeek-V3 was hailed by American tech circles as “the mysterious force from the East.”
DeepSeek’s triumph was no fluke. It sprang from Liang’s tailored strategies for people management and R&D. He built an all-local team of young coders, eschewing reliance on overseas returnees or senior experts. Most hires were recent grads or held less than five years of experience. Liang contended, “Innovation requires breaking away from inertia—experience can be a burden.”
Additionally, DeepSeek’s open-source stance—freely sharing key technologies with a global developer base—has earned it international accolades and accelerated the industry’s technical ecosystem. Andrej Karpathy, an original founding member at OpenAI, observed, “DeepSeek shows astounding engineering prowess under limited resources. It could redefine how large models are developed.”
These accomplishments reverberated in Silicon Valley, where many insiders hailed DeepSeek-V3 for breaching the “cost barrier” of top-tier global AI. Observers remarked that DeepSeek’s feats in reducing computing demands and boosting training efficiency may fundamentally rewrite AI’s landscape.
Yet Liang has not hurried to commercialize DeepSeek, insisting that AI success is determined by long-term development, not short-term profit. “DeepSeek’s mission is to make AI accessible more efficiently and at lower cost,” he says, “always placing technology first.”
04 Technology Driving Tomorrow: The Emerging Power of Chinese Innovation
All along, Liang has stood at the forefront of technical progress. Whether in quantitative finance or AI, he has demonstrated time and again that Chinese tech can do more than join the global race—it can take the lead. He maintains that China’s tech firms must not simply trail behind but confidently occupy the vanguard.
DeepSeek’s success vindicates his boldness and perseverance. In a world of intensifying AI rivalry, Liang’s unwavering commitment to technological breakthroughs has secured a prominent place for China’s AI on the global stage. As he often says, “We’re not just catching up; we’re determined to truly exceed in core technologies.”
Thus emerges the story of Liang Wenfeng, a post-1985 Chinese visionary who, with quiet resolve and formidable skill, is composing new chapters in which Chinese technology stuns the world.
This translation is an independent yet well-intentioned effort by the China Thought Express editorial team to bridge ideas between the Chinese and English-speaking worlds. The original text is available here:
https://www.backchina.com/news/2025/01/27/955254.html
Kindly attribute the translation if referenced.