As the "exit master" of the mobile internet era, Zhu Xiaohu's investment philosophy has always centered around "doing the math."

From ofo to Didi, each of his exits has validated his strategy of "buying in on disagreements, selling out on consensus." Now in the realm of embodied intelligence, he continues this approach: when the valuations of humanoid robot companies skyrocket due to policy catalysts, he publicly questions "who would spend tens of thousands to buy a robot to do work" and, after exiting the projects of Xinghaitu and Songyan Power, classifies them as "tracks that early-stage VCs want to avoid."

At a recent Zhongguancun conference, Zhu Xiaohu bluntly stated, "Today's AI application entrepreneurs are akin to early internet entrepreneurs; they must bravely admit there are no barriers." However, his intention was not to be bearish but to advise practitioners to focus on user experience and product design - "building barriers beyond AI capabilities, with all capabilities stemming from beyond AI."

Those who favor his style view this rhythm of "laying out in controversy, exiting during the frenzy" as rational pragmatism. When the consensus is high but the commercialization path is unclear, the rational choice for early-stage VCs is to cash in; critics may dismiss him as a "standardized speculator" who cashes out on trends or even "stabs the industry in the back."

Returning to the initial question, does China need more "Zhu Xiaohu"s?

The crux of this question is whether, in an era where capital frenzy and innovation confusion intertwine, China's venture capital industry needs more rational gatekeepers who focus on the essence of business. Alternatively, does it need a Zhu Xiaohu who occasionally throws cold water on every overheated industry?

01

"When consensus is highly concentrated, it's a signal for us to exit"

In the investment circle, Zhu Xiaohu is known for his early-stage investments and rapid exits, emphasizing commercialization and short-term returns. His most notable investments, from Ele.me, Didi, to Xiaohongshu and Inke, are primarily concentrated after 2010, the era when the mobile internet just exploded.

In the case of ofo, in 2016, GSR Ventures led the A round with US$10 million. At that time, the shared bike market was in its infancy, and skepticism abounded. However, Zhu Xiaohu identified the "last mile" pain point and capital arbitrage opportunity. In less than two years, as ofo completed its E round of funding, its valuation soared to US$2 billion.

When rumors of Mobike and ofo merging were rampant and the market consensus expected the industry to soon consolidate, he quietly cashed out US$300 million through equity transfers, successfully avoiding the peak. Subsequently, ofo fell into a deposit refund crisis due to a broken capital chain, validating the prediction of the "pseudo-proposition of the sharing economy" - and Zhu Xiaohu became the only investor to profit from the sharing economy.

Didi followed a similar trajectory. As the mobile internet boomed, ride-hailing was still novel, with the market riddled with doubts. The traditional taxi industry dominated, people's acceptance of online hailing was low, and the business model was highly questioned. However, Zhu Xiaohu was optimistic about urban transportation demand and the pain points of traditional taxis. Despite early challenges in technology and user growth, and the immense pressure of burning money to subsidize, Zhu Xiaohu not only invested US$3 million in Didi's angel round but also helped Didi cooperate with Tencent, driving a surge in user volume.

With Didi defeating Uber China in 2016, its business model became clear, and its valuation soared. Simultaneously, Zhu Xiaohu perceived risks and valuation bubbles, gradually reducing his stake in Didi when the trend was at its peak. From 2012 to his exit in 2017-2018, Didi generated nearly a thousand times the investment return, becoming a classic case in the investment world.

To some extent, even today, the phrase "when consensus is highly concentrated, it's a signal to exit" remains the principle of GSR Ventures. Times and trends may change, but his underlying logic remains steadfast:

Exit during the trend's peak, deploy high-cost-effectiveness projects in low-consensus areas, and be afraid when others are greedy. Zhu Xiaohu believes that the essence of entrepreneurship is still doing business, doing the math, prioritizing "whether it can be commercialized," focusing on hard indicators such as customer willingness to pay, sales cycle, sustainability, and retention rate, trusting in logic and data, and refusing emotional decisions.

From a professional perspective, GSR Ventures' capital stems from various global funds, strategic investors, and private equity. As an institution's responsibility, it is to generate returns for investors, with investor returns as the core consideration, paying for the future but not for stories laden with excessive metaphysical factors, distinguishing it from many USD fund investors.

"Zhu Xiaohu's mindset is sound. As an LP, if I saw my money being invested daily into projects that only tell stories, I would go crazy." Facing online criticism of Zhu Xiaohu, some investors have spontaneously expressed their opinions on social media, "Money spent is to make money, but surely there are also people willing to spend money to gamble on the long term and take risks, so just choose such institutions. This is just two investment styles in itself."

Zhu Xiaohu's investment preferences remain focused on verifiable rigid demand areas.

He avoids investing in large models due to the unclear commercialization path for basic models. He favors enterprise-level tools and consumer-grade applications that can quickly land and solve practical needs, hence his bet on AI applications. Another major category is the consumer track, catering to the upgrading demand of the domestic market. He has invested in numerous new consumer brands and internet products in the past and prefers projects with stable demand and a clear commercialization path.

However, for technological conceptual types such as humanoid robots, purely research-oriented embodied intelligence, or "long slopes with thick snow" that rely on capital subsidies but cannot prove a unit economic model, he privately does not deny that these "consensus tracks" will produce good companies in the future and even deployed in related fields years ago - but at this point in time and valuation, he does not expect to make long-term money.

02

The "Prisoner's Dilemma" of China's Venture Capital

Zhu Xiaohu feels like the sentiment from ten years ago has returned, but he cannot say if it is joy or fear.

Ten years ago, the mobile internet was in full swing, from O2O local life, the sharing economy, to internet finance. Trends emerged frequently, and a vast amount of capital flowed into the investment circle. During this period, the investment logic centered on "DAU" and "MAU" as core valuation indicators, neglecting unit economic models such as customer transaction value and retention rate; the market catered to "winner-takes-all," with capital betting on top players in the track. Meanwhile, under the FOMO sentiment, fearing missing the next BAT, investors rushed into popular tracks, pushing up the valuations of startups.

This situation eerily mirrors the AI trend that has emerged in the past three years. The controversy sparked by Zhu Xiaohu has amplified certain details once again.

An industry analyst noted, "Interestingly, friends in the robotics industry say that finally someone who understands the situation has come out to warn about bubbles and risks; friends in VC are all scolding him, saying that his position determines his perspective and he's sour grapes because he can't eat them."

A person who has been engaged in robot motion control for many years candidly stated, "It will be difficult to have a real market within ten years" because the robot industry currently lacks the nature of a mature product. The uniformity of each robot is too poor, and most of the videos released are after lengthy debugging on a brand-new prototype. Most of the work in motion control involves not writing algorithms but repeated debugging. There are too many issues to solve, such as product battery life, post-maintenance, and operating heat.

"The large-scale influx of funds in the past two years has accelerated progress, but it is not good competition. Everyone is eager to release videos with display effects, and there has been no significant progress in addressing real issues."

In the eyes of many, Zhu Xiaohu's words primarily sting many investors who tacitly understand each other. Humanoid robots are a subfield within embodied intelligence. From the perspective of commercialization progress, compared to industrial robots and hotel delivery robots, humanoid robots such as Tesla's Optimus and UBTECH's WalkerS, although entering factory training, still face issues of high costs, technological bottlenecks, safety ethics, and insufficient market demand for commercialization.

Under this circumstance, the market indeed needs to calmly consider "who will pay," "how to profit," and "when to land." Zhu Xiaohu's "unorthodox behavior" is, in a sense, an inevitable product of the imbalance in China's venture capital ecology.

Over the past decade, the market has experienced countless frenzies and collapses. Shared bikes burned through financing and piled up in ruins, the metaverse concept soared and then collapsed, and autonomous driving consumed capital under the slogan of "infinitely close to mass production." Too many investors were obsessed with "technological sexiness" rather than real demand, replacing business validation with PPT fundraising, ultimately leaving behind a mess.

These lessons prove that innovation divorced from the essence of business will eventually be liquidated by the market.

However, the nature of capital is to chase trends, especially in economic downturns, where LPs' anxiety about short-term returns is amplified, forcing investors to swing between "chasing the wind" and "hedging risks." The popularity of the humanoid robot track is a microcosm of this contradiction: primary market valuations soar, but real demand only comes from narrow scenarios such as procurement displays by state-owned enterprises and university research; entrepreneurs weave a beautiful vision of companion robots but are helpless in cost control and mass production.

At this juncture, actively refusing to pass the hot potato and based on past experience, when consensus is highly concentrated, valuation bubbles are often close to the breaking point. As the industry undergoes reshuffling, for all investors, the companies that remain after the washout also satisfy the purpose of reverse project screening.

03

Does China Need More "Zhu Xiaohu"s?

Zhu Xiaohu's voice can curb industry overheating, but the "sole commercialization theory" may also stifle innovation. This is another consensus from the perspective of industry development, setting aside capital interest relationships.

"If all capital followed Zhu Xiaohu's lead, sectors with long cycles such as chips and biopharmaceuticals would have long been neglected." An investor drew a comparison between the Chinese and American markets and concluded, "A few American institutions can invest in a company for over a decade or two, which is hard to imagine in China. Most people think that a company must be profitable within three years and preferably packaged for an IPO exit within five years. The redemption pressure comes as soon as there is any sign of trouble."

In fact, the difference between the mindset of "profitability within three years, IPO within five" and "a decade of companionship" is essentially the result of the combined effects of capital attributes, market environment, and exit mechanisms.

To illustrate, the survival cycle of most domestic startups does not exceed three years. Coupled with a more competitive market and severe homogenization under trends, these real dilemmas are also forcing a shift in investment logic. Moreover, compared to the long-term capital of American VC/PE, the LPs of domestic GPs are mostly high-net-worth individuals and corporate funds, with generally higher expectations for annualized returns, forcing investors to compress cycles. A partner at a RMB fund bluntly said, "LPs ask me every year how much dividends I have distributed. How can I invest in a project that doesn't exit for ten years?"

This leads to a situation where discussions about "long-termism" have always been a hot topic in the domestic investment circle, but not many can truly achieve long-termism.

From this perspective, compared to insisting on "speaking with data" and requiring companies to prove their commercialization capabilities in the short term, investors represented by Zhang Ying advocate trading time for space, willing to tolerate the cycle of "bubble-death-rebirth" for humanoid robots, which is particularly important for an industry that has not been fully validated.

The divergence between the two approaches lies in the anxiety of financial capital on the one hand. The duration of RMB funds is generally 7-10 years, while the commercialization of humanoid robots requires at least a 10-year cycle. This mismatch leads most VCs to choose "invest early, exit early," and there is even a phenomenon of "planning an exit at the Pre-A round funding stage."

On the other hand, it reveals the patience of industrial capital. For example, technology giants such as Tencent and Alibaba strategically invest in embodied intelligence, valuing technology positioning rather than short-term returns. Ant Group invested in Xinghaitu after GSR Ventures exited, valuing its synergy with its AI layout, which is also a typical example.

China needs both Zhu Xiaohu-style "rational gatekeepers" and Hillhouse Capital's "long-termism," but what it lacks most is game-changers who can bridge technological ideals with business realities. In this debate about humanoid robots, there is no absolute right or wrong. The future venture capital ecosystem should find a dynamic balance between "accountants" and "idealists," "short-termism" and "long-termism."

As one investor aptly said, "We must revere the market, but we must revere innovation itself."