Investment Institutions Await the Next DeepSeek

Author: Ren Xueyun

Image Source: Generated by Wujie AI

The AI large model DeepSeek has taken the world by storm with its three main features: user-friendly, open-source, and free. It has become the fastest app in history to surpass 30 million daily active users, and now various industries are being "DeepSeeked."

Before its explosive growth, however, very little was known about the company. It rarely appeared in media reports, headhunter recruitments, or industry research, and it never stirred the market like the "AI Six Dragons."

Its founder, Liang Wenfeng, is also low-profile, with only a few words available online about him, and the number of public speeches he has given is extremely limited.

Even among practitioners in the large model industry, DeepSeek was not particularly noted. "I saw it in headhunter recommendations in 2023, but I didn't pay much attention at the time," said one large model practitioner. "In the past few years, the focus has been on the 'Six Dragons' of large models, and research on DeepSeek was limited to past products," said an investment manager.

But now, things are completely different; peers and investment institutions are all looking for Liang Wenfeng.

"To be precise, we started to pay special attention to this product after DeepSeek open-sourced its inference model DeepSeek-R1," said the aforementioned large model practitioner. After the open-sourcing of DeepSeek-R1, many industry insiders reassessed DeepSeek's technical strength.

At the same time, investment institutions have begun to take action, trying every means to contact more people related to DeepSeek. "Of course, we want to meet Liang Wenfeng," said an investment manager from a private equity firm.

However, just like before its explosive growth, DeepSeek and Liang Wenfeng remain elusive in this wave of enthusiasm, and no one can easily find their traces.

Nearly a Hundred Visitors to DeepSeek's Office in One Day

"Highly guarded," is the most intuitive feeling of an employee working in a company in the Huijin International Building on the first day back to work after the Spring Festival. In early February, when she entered the building, she keenly noticed that the security situation was completely different, with a level of strictness far exceeding the past. "Later, I heard from the security personnel that DeepSeek is in the same building as us."

In the following week, as DeepSeek's popularity soared, the Huijin International Building became a hot "check-in point." A building staff member said that at its peak, nearly a hundred visitors came in one day.

However, most of these "visitors" left disappointed. "Although you can find DeepSeek's specific office address online, the security personnel won't disclose the specific floor it's on, let alone expect to see their employees," said the aforementioned employee.

Among those searching for DeepSeek and Liang Wenfeng, investment institutions that have long sensed market opportunities are also making efforts.

An investment manager told Tech Planet that finding contact information for DeepSeek staff, even founder Liang Wenfeng, is not difficult, but the challenge lies in communication. "Through connections I built in the AI industry, I added a few DeepSeek employees, but didn't get much feedback."

Another investment manager also stated that they were trying to contact DeepSeek during the Spring Festival, but there were no results.

In a short time, those who had intersected with DeepSeek in various ways, no matter how trivial the connection, became "gold mines" that everyone was eager to dig into.

A headhunter who had previously posted recruitment needs for DeepSeek said, "During the hottest days for DeepSeek, many people asked me questions related to DeepSeek, some inquiring about recruitment needs, others about the company's situation."

And a practitioner who was invited by DeepSeek for an interview found that after sharing their experience with friends, many "strangers" added them on WeChat.

The almost fanatical attention from the outside world surrounded DeepSeek, but it has never signaled a desire for communication.

Shortly after returning to work after the Spring Festival, DeepSeek, through the property management of Huijin International Building, declined all interview requests, stating that it would focus on its own research and development. Since its explosive growth, Liang Wenfeng has only briefly appeared in the public eye.

Nevertheless, discussions surrounding DeepSeek and Liang Wenfeng's past experiences remain heated.

"During the month after the holiday, our work has still revolved around researching DeepSeek," said the aforementioned investment manager, who has nearly reviewed all publicly available materials on Liang Wenfeng and DeepSeek.

On social media, discussions about DeepSeek and Liang Wenfeng continue to emerge, from tech enthusiasts discussing the optimization of DeepSeek's model architecture to business people analyzing its potential market valuation, with enthusiasm unabated.

As the external search for DeepSeek and Liang Wenfeng deepens, this previously obscure company seems to become even more mysterious, with everyone wanting to unravel the many mysteries surrounding it.

Investment Institutions and DeepSeek: A Misaligned Game Amidst a Financing Frenzy

Compared to the curious eyes from the outside, the investment institutions in the primary market are sensing the scent of money; they ultimately want to squeeze into the ranks of DeepSeek's investors.

Even Zhu Xiaohu, the managing partner of Jinsha River Ventures, has reversed his attitude towards AI investments because of DeepSeek. Zhu had previously stated that he would no longer invest in large model companies, but DeepSeek changed his skeptical view of General Artificial Intelligence (AGI).

In an interview with Tencent News, he stated that if DeepSeek opened for financing, he would definitely participate because "witnessing the birth of human AGI" is of great significance. He even mentioned that the price is no longer very important; the key is to be involved.

Another investor also stated that if they could really participate in the investment in DeepSeek, who would still worry about price and share?

However, unlike the eagerness shown by primary market investors, DeepSeek has never had any record of external financing, and currently, it has not signaled a need for financing.

According to Qichacha, the company behind DeepSeek, Hangzhou Deep Seeking Company, is 84.29% directly and indirectly owned by Liang Wenfeng, giving him absolute voting rights over DeepSeek.

After its explosive growth, many investors approached DeepSeek's IR (Investor Relations) team. One PE (Private Equity) investor stated that they have been in contact with DeepSeek, hoping to obtain investment shares, but DeepSeek clearly stated: currently, it is not open for financing.

In Liang Wenfeng's personal career history, money does not seem to be the core factor hindering his entrepreneurship. The profits from his quantitative hedge fund, Huanfang Quantitative, supported his investment in DeepSeek.

Huanfang Quantitative was established in 2015, initially as Zhejiang Jiuzhang Asset, and in 2016, it established Ningbo Huanfang Quantitative. According to data from the Asset Management Association of China, as of July 31, 2024, Zhejiang Jiuzhang Asset's management scale exceeds 10 billion yuan; as of April 26, 2024, Ningbo Huanfang Quantitative's management scale also reached over 10 billion yuan.

This strong financial backing allows Liang Wenfeng to have a relatively relaxed funding environment when investing in the development of DeepSeek.

However, DeepSeek's model of relying on its own funds for development, along with its rising popularity in the market, has also deterred many investors.

"On one hand, they have clearly stated multiple times that they are not seeking financing. On the other hand, given DeepSeek's current popularity, its valuation is certainly beyond what we can afford," said a PE investor.

According to Bloomberg's research analysts' predictions for DeepSeek's median valuation, it could reach $30 billion. This price is nearly the total valuation of the current "AI Six Dragons."

But Liang Wenfeng has not ruled out the possibility of financing. In an exclusive interview with "Dark Current Waves" in 2023, he revealed that he had actively communicated with multiple investors, trying to bring in funds to support DeepSeek's development. However, the negotiations highlighted the contradictions—venture capital (VC) firms, constrained by exit mechanisms, are eager to commercialize projects to quickly recoup funds and generate returns.

In contrast, Liang Wenfeng's development blueprint for DeepSeek centers on "research first." This ideological difference has become a significant barrier between DeepSeek and VCs. Even today, this contradiction cannot be easily resolved.

Waiting for the Next DeepSeek

On March 1, DeepSeek's team published their first update on their Zhihu account, publicly revealing key information about the optimization technology details and cost-profit ratio of the DeepSeek-V3/R1 inference system for the first time.

Based on theoretical calculations, DeepSeek claims that its AI model's inference service has a 545% "cost-profit ratio." This means that for every $1 spent, DeepSeek can earn $5.45 in profit.

However, 545% is merely a theoretical figure; in reality, "night discounts," "V3 low-price strategies," and "free access via web/App" are all factors that influence this data.

This theoretical figure of 545% is akin to the best circumstances DeepSeek currently finds itself in; no one knows if it can be sustained or for how long.

A practitioner in the large model industry told Tech Planet that the challenges facing DeepSeek have already arrived, "DeepSeek's shock to the market actually represents a victory for open-source."

In his view, it is very difficult for open-source models to continuously launch products that can compete with closed-source models, and on the other hand, open-source model companies cannot avoid the topic of commercialization and monetization; to continue, they must find ways to monetize.

Not long ago, Wu Bingjian, a partner at Xincapital, shared a sentiment on social media: last year, he asked someone from DeepSeek, "Why is your model so good?" The response was that their boss reads papers, writes code, and does recruitment; where time is spent, efficiency is easily achieved.

Facing the next round of challenges, the DeepSeek team cannot afford to be complacent. "Their team must definitely be energized to explore various technical issues of large models; otherwise, they may fall behind," said the aforementioned practitioner.

For investors who missed out on DeepSeek, now is the time for them to reassess the key factors that influenced their investment decisions in the past.

One investor reflected that, for example, looking at the team, Liang Wenfeng's team has a background in quantitative funds, working on AI open-source models, which does not align with past experiences. Another example is the business model; venture capital tends to favor clear commercial routes, but even today, it is difficult to see DeepSeek's future commercial direction.

"The explosive growth of DeepSeek has given us a chance to rethink," in his view, DeepSeek has broken the conventional understanding of past investments, cultivating stronger foresight and inclusivity, which is the capability needed to not miss the next DeepSeek.

But in any case, DeepSeek has disrupted the narrative of giants in the AI field, and now investors are eagerly hoping to seize it before the next DeepSeek emerges.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。