
Searching for DeepSeek's Liang Wenfeng
TechFlow Selected TechFlow Selected

Searching for DeepSeek's Liang Wenfeng
Investment firms are waiting for the next DeepSeek.
Author: Ren Xueyun

Image source: Generated by Wujie AI
DeepSeek, a large AI model known for being user-friendly, open-source, and free, has gone viral globally, becoming the fastest app in history to surpass 30 million daily active users. Today, DeepSeek is transforming industries across the board.
Yet before its sudden rise, this company was largely unknown to the outside world. It rarely appeared in media reports, headhunter job postings, or industry research, never creating ripples in the market like the "six AI unicorns" did.
Its founder, Liang Wenfeng, is equally low-key. Online information about him is scarce, and there are only a handful of public speeches available.
Even professionals within the large model industry didn't pay much attention to DeepSeek. "I saw it once in a headhunter’s recommendation back in 2023, but didn’t think much of it at the time," said one practitioner. "For years, all the attention focused on the 'six AI unicorns.' Our understanding of DeepSeek was limited to its earlier products," noted an investment manager.
Now, however, everything has changed—rivals and investment firms alike are scrambling to find Liang Wenfeng.
"To be precise, we started paying serious attention to this product after DeepSeek open-sourced its reasoning model, DeepSeek-R1," said the above-mentioned practitioner. After the release of DeepSeek-R1, many in the industry reassessed DeepSeek's technical capabilities.
At the same time, investment institutions sprang into action, going to great lengths to connect with anyone linked to DeepSeek. "Of course we want to meet Liang Wenfeng," said an investment manager at a private equity firm.
But just as during its pre-fame days, DeepSeek and Liang Wenfeng remain hidden amid the current wave of attention—no one can easily track them down.
Nearly 100 visitors flood DeepSeek’s office each day
"Tight security"—this was the most immediate impression of an employee working at Huijin International Tower upon returning from Lunar New Year break. In early February, when she entered the building, she quickly noticed that security had become significantly stricter than before. "Later I learned from the guards that DeepSeek shares the same building as us," she said.
In the following week, as DeepSeek’s popularity surged, Huijin International Tower became a hot "check-in spot." A staff member at the building said that at peak times, nearly 100 visitors came each day.
Yet most of these "visitors" left disappointed. "Although you can find DeepSeek’s exact office address online, the security guards won’t reveal which floor they’re on, let alone allow you to meet their employees," said the employee.
Among those seeking DeepSeek and Liang Wenfeng were investment institutions eager to seize new market opportunities.
An investment manager told TechFlow that getting contact details for DeepSeek staff—or even founder Liang Wenfeng—is not hard, but actually communicating with them is the real challenge. "Through my network in the AI industry, I added several DeepSeek employees, but received little response."
Another investment manager said he spent the entire holiday trying to reach DeepSeek, but without success.
Suddenly, anyone who had ever interacted with DeepSeek—even in the most minor way—became a "gold mine" everyone wanted to tap.
A headhunter who once posted job openings for DeepSeek said, "During DeepSeek’s hottest days, many people asked me about the company—some wanted to know about hiring needs, others sought general information."
One practitioner who had been invited for an interview at DeepSeek found himself suddenly added on WeChat by numerous "strangers" after sharing his experience with friends.
Intense external interest now surrounds DeepSeek, yet the company has never signaled any desire for engagement.
Shortly after resuming work post-holiday, DeepSeek, through the property management of Huijin International Tower, politely declined all media interview requests, stating it would focus on internal research and development. Since its breakout moment, Liang Wenfeng has made only brief appearances in public.
Nonetheless, discussions about DeepSeek and Liang Wenfeng’s past continue to rage on.
"For a month after the holiday, our work revolved entirely around researching DeepSeek," said the investment manager, who claims to have reviewed every piece of publicly available information on Liang Wenfeng and DeepSeek.
On social platforms, conversations about DeepSeek and Liang Wenfeng keep emerging—from tech enthusiasts analyzing architectural optimizations of DeepSeek models, to business analysts debating its potential market valuation. Enthusiasm remains high.
As external scrutiny of DeepSeek and Liang Wenfeng intensifies, this once-obscure company seems to grow even more mysterious. Everyone wants to unravel the many secrets surrounding it.
Investors vs. DeepSeek: A mismatched game amid financing frenzy
Compared to general curiosity, investors in the primary market sense something far more valuable—money. Ultimately, they all want to become investors in DeepSeek.
Even Zhu Xiaohu, managing partner at GSR Ventures, reversed his stance on AI investments because of DeepSeek. Zhu had previously stated he would no longer invest in large model companies, but DeepSeek changed his skeptical view of Artificial General Intelligence (AGI).
In an interview with Tencent News, he said that if DeepSeek opens funding rounds, he would definitely participate, as "witnessing the birth of human AGI" carries immense significance. He even added that price is no longer important—the key is being part of it.
Another investor echoed this sentiment: if given the chance to invest in DeepSeek, who would still care about price or share size?
However, unlike the fervor shown by investors, DeepSeek has no record of external funding in the past, nor does it currently signal any need for fundraising.
According to Qichacha, Hangzhou Deep Exploration—the company behind DeepSeek—shows Liang Wenfeng directly and indirectly holding 84.29% of shares, giving him absolute voting control over DeepSeek.
After its rise to fame, many investors reached out to DeepSeek’s IR (investor relations) team. One PE investor said they’ve been in contact with DeepSeek, hoping to secure investment shares, but DeepSeek made it clear: no financing is currently open.
In Liang Wenfeng’s career path, money has never been a major constraint for his ventures. His quantitative hedge fund,幻方量化 (High-Flyer Quant), has funded DeepSeek’s operations.
Founded in 2015, High-Flyer Quant began as Zhejiang Jiuzhang Asset, followed by Ningbo High-Flyer Quant in 2016. Data from the Asset Management Association of China shows that as of July 31, 2024, Zhejiang Jiuzhang managed assets exceeding 10 billion RMB; as of April 26, 2024, Ningbo High-Flyer also managed over 10 billion RMB.
This strong financial backing provides Liang Wenfeng with a relatively comfortable capital environment for developing DeepSeek.
Yet DeepSeek’s self-funded growth model, combined with its rising market热度 (heat), has deterred many investors.
"On one hand, they’ve clearly stated multiple times they’re not raising funds. On the other, given DeepSeek’s current热度, its valuation likely exceeds what we can afford—we simply can’t invest," said a PE investor.
According to Bloomberg analyst surveys forecasting median valuations, DeepSeek could be valued at up to $30 billion—nearly equivalent to the combined valuations of the "six AI unicorns" today.
Still, Liang Wenfeng hasn’t completely ruled out financing. In a 2023 exclusive interview with "An Yong Waves," he revealed that he had proactively approached multiple potential investors to bring in capital for DeepSeek’s growth. But negotiations exposed a core conflict—VC firms, bound by exit mechanisms, push for rapid commercialization to recoup investments quickly.
Liang Wenfeng’s vision for DeepSeek, however, centers on "research first." This philosophical gap has become a major barrier between DeepSeek and traditional VCs—a divide that remains unresolved even today.
Waiting for the next DeepSeek
On March 1, DeepSeek’s Zhihu account posted its first update, disclosing key technical details about the optimization of its DeepSeek-V3/R1 reasoning system, including cost-profit ratios.
Based on theoretical calculations, DeepSeek claims a 545% "cost-profit margin" for its AI inference services. This means for every $1 spent, DeepSeek earns $5.45 in profit.
However, this 545% figure is purely theoretical. Factors such as "nighttime discounts," "V3’s low-price strategy," and "free web/app access" all affect actual profitability.
This theoretical number mirrors DeepSeek’s current golden moment—no one knows how long it will last, or whether it can be sustained.
A large model industry practitioner told TechFlow that challenges for DeepSeek have already begun. "DeepSeek’s shockwave in the market actually represents a victory for open-source."
In his view, maintaining continuous innovation in open-source models—on par with closed-source competitors—is extremely difficult. Moreover, open-source AI companies cannot avoid the issue of commercialization and monetization—they must find sustainable revenue models to survive.
Not long ago, Wu Bingjian, partner at Xin Capital, shared a reflection on social media: Last year, he asked someone from DeepSeek, "Why is your model so good?" The reply: "Because our boss reads papers, writes code, and handles recruitment himself. Where time is invested, efficiency naturally follows."
Facing upcoming challenges, the DeepSeek team cannot afford any lapse. "They’ll have to stay sharp exploring various technical frontiers in large models, or else risk falling behind," said the practitioner.
For investors who missed out on DeepSeek, this moment calls for reevaluating past decision-making criteria.
One investor reflected: For example, team background—Liang Wenfeng’s team comes from a quant fund, building open-source AI models, which defies conventional wisdom. Or business model—VCs usually favor clear commercial paths, but even today, DeepSeek’s future revenue trajectory remains unclear.
"DeepSeek’s breakout gives us a chance to rethink everything," he said. "DeepSeek breaks traditional investment logic. To not miss the next DeepSeek, we need greater foresight and openness."
Regardless, DeepSeek has disrupted the dominant narrative held by tech giants in AI. Now, investors urgently hope to identify and capture the next DeepSeek before it emerges.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News














