Key Takeaways from Zhu Xiaohu's Interview:
Investment Approach and Trends: Zhu Xiaohu, a managing partner at GSR Ventures, highlights his adaptability to market trends and opportunities. Despite the challenges posed by the pandemic, he continues to actively invest, as seen in his decision to fund Fancytech after a short video call. Zhu emphasizes the importance of focusing on sectors he excels in, like AIGC (AI-Generated Content) and consumer sectors, adapting his investment pace and decision-making process according to the market conditions.
Perspective on Large Models and AI: Zhu notes the surge in interest in large AI models, like ChatGPT, but cautions against their hype, particularly for startups. He points out the risks and limitations associated with these technologies and advises startups to focus on niche areas and applications where they can excel and differentiate themselves.
Entrepreneurship and Investment in a Tight Market: Zhu observes that the current investment climate is challenging due to reduced liquidity and lower valuations. He advises entrepreneurs to focus on cost optimization and efficiency, citing that companies that adapted to this approach have remained profitable. Zhu advocates for discipline in investment strategies, focusing on profitability and sustainable growth.
Outlook on Consumer Investments: Zhu mentions the changing dynamics in consumer investment, noting a shift from speculative valuations based on future sales to more grounded approaches focusing on current profitability. He stresses the importance of being cautious in this sector, especially under current market conditions.
Global Expansion of Chinese Companies: Zhu highlights the unique strengths of Chinese companies in going global, particularly their strong supply chain advantages and the unparalleled work ethic of Chinese entrepreneurs. He observes the increasing competition among Chinese companies in international markets across various sectors, from e-commerce to fintech. Zhu also discusses the strategic considerations for Chinese companies entering foreign markets, emphasizing the need for language skills and cultural adaptability.
Middle Eastern Investments in China: Addressing the increasing interest of Middle Eastern capital in Chinese ventures, Zhu views this as a strategic move aiming for industry complementarity and reciprocal investment. He advises Chinese companies to align strategically with Middle Eastern interests for successful investment partnerships.
Overall, Zhu Xiaohu's insights reflect a strategic, disciplined approach to investing in challenging market conditions, emphasizing adaptability, focus on profitability, and the importance of understanding global market dynamics for successful entrepreneurship and investment.
(Translated with ChatGPT)
In October, Zhu Xiaohu, a managing partner at GSR Ventures, attended a meeting in the United States, surprising Limited Partners (LPs) with his youthful appearance. "You don't even have white hair, that's not easy," they remarked.
During an interview with "China Entrepreneur" at GSR Ventures' office, Zhu Xiaohu brushed his hand over his hair, feeling it was even darker than before. This, he believed, was partly because the development of his investment projects kept his worries at bay: "As long as the businesses are profitable, I can sleep well."
Zhu remains sensitive to market trends and new opportunities. In April 2022, during Shanghai's lockdown, Fancytech's founder and CEO, Kong Jie (alias), could only communicate with Zhu online for funding. After a 15-minute video call, Zhu decided to invest.
At that time, AIGC (AI-Generated Content) was not yet a hot topic. Fancytech mainly helped e-commerce sites with content aggregation to boost brand conversion rates. After integrating AIGC for a technological upgrade and business growth, Zhu invested in three additional rounds in Fancytech.
However, his sensitivity often manifests in unconventional ways. For instance, in the first half of the year, when many were chasing large models, Zhu, who had been quiet for a while, appeared and poured "cold water" on the hype. In a public speech, he mentioned that ChatGPT is unfriendly to startups and advised against harboring fundraising illusions for the next two to three years. This sparked a public debate with Fu Sheng, Chairman of Cheetah Mobile.
To this day, Zhu maintains that big models are ultimately opportunities for large companies, and startups should focus on verticals like sales and marketing. His investment strategy is similar: "I believe in focusing on what I am good at, as market cycles are always changing."
AIGC and consumer sectors are Zhu's focus areas, but even in familiar fields, his investment decisions have slowed. This year, he spent more time evaluating projects, moving away from quick decisions like the one he made after a half-hour chat with Cheng Wei. "Because no one is rushing (for projects), I can spend three to four months on average per project to ensure it's a 'good card'."
From hasty investments during the influx of "hot money" to a more deliberate pace now, Zhu finds 2023 to be his most comfortable year in investing.
As of November 2023, the number of projects Zhu invested in is only a third of previous years, with several still under his scrutiny. "The sense of achievement for an investor doesn't come from funding, but from the growth of enterprises."
During the interview, Zhu repeatedly mentioned "discipline," which involves controlling the size of investments, pacing, and valuations. Below is an edited version of the interview with "China Entrepreneur."
Discussing Large Models: Those who chased large models in the first half of the year suffered in the latter half
Question: How do you view the impact of large models on the technology and startup cycle after their emergence?
Zhu Xiaohu: I think it's a great start to a new cycle. For entrepreneurs, the key is what you want to do. Like the PC internet and mobile internet each dominated a decade, now we're entering a new AI era, likely lasting 10 to 20 years.
But for entrepreneurs to do something like iOS in the mobile internet is unrealistic. There are many opportunities in applications above large models. We've invested in many companies in the AIGC field, and they're doing well.
Question: Has this change in the entrepreneurial cycle imposed new demands on investors? With fewer entrepreneurs and less need for funding, how do investors find good projects?
Zhu Xiaohu: The core remains the same as before: independent judgment, not following the crowd. In the first half of this year, everyone was chasing large models, pouring in money, now they don't know what to do. The main issue is that investors must have their independent judgment. Where do you see the opportunity? For startup companies, creating a new iOS is undoubtedly very challenging. We made similar mistakes in the mobile internet era and must learn from them.
Many investors might not remember the stories of the PC and mobile internet eras as vividly. We've lost money, so I remember well. I know where the pitfalls are, and we'll be more focused on more creative companies or certain opportunities.
Question: What are these "pitfalls"?
Zhu Xiaohu: It's simple. In the early days of the mobile internet, many companies were initially very popular, like Android optimization masters, with several hundred million users, but who uses them now? A simpler example is a company that made flashlight apps for phones, but now iOS and Android systems come with them.
It's similar now to the early days of operating systems. Large models in the early stage may not cover many small functions, leaving opportunities for startups, but once they get around to it, they can easily fill these gaps, and those startups disappear.
There was an article about GPT-4 launching a suite of tools, destroying a bunch of startups. Entrepreneurs should definitely not run directly under the fire of large models; it would be very difficult.
For example, if large models can now automate coding by 50%, leaving a bunch of bugs, it might help companies reduce costs for junior programmers, but more senior programmers are needed to debug. Many startups aim to use AI to optimize code or debug better, but these are short-term opportunities. If GPT-5 comes out next year and large models can achieve 90% efficiency in code generation, those companies are again at risk.
So I think entrepreneurs should not directly run in the large model track. Keep a distance from it while using it to seize customers and achieve a data loop first; this might be a better defense.
Question: In your previous discussion with Fu Sheng on social media, do you think there's a clear conclusion?
Zhu Xiaohu: Frankly speaking, for startups, I think large models themselves don't offer much opportunity. First, the cost is very high; second, without data and scenarios, it's impossible to optimize.
So I think large models are ultimately opportunities for large companies. Startups should focus on applications in sales, marketing, and other verticals where they excel. There are still plenty of opportunities in various vertical scenarios.
Question: What specifically do you mean by vertical scenarios?
Zhu Xiaohu: First, having vertical industry data is vital, which large model companies find hard to acquire. Second, achieving a data loop to continuously improve your vertical model. Third, maintaining a constant connection with customers is crucial, as large companies can't engage each customer individually. These three points are especially important.
That is, adding AI to existing businesses is the most reliable and easiest; starting with an AI model or product and adding things on top is very difficult.
Question: How do domestic entrepreneurs or big companies consider this matter?
Zhu Xiaohu: We've been discussing it. Big companies are also concerned about their boundaries. Some are better; they don't want to snatch too many opportunities from startups. But whether in China or the US, relying on others' mercy is unrealistic. You must have your value to both customers and big companies to find your footing.
Question: How do you view the current progress of commercialization of large models in China? What stage is everyone at?
Zhu Xiaohu: Everyone is now looking for nails with a hammer. Honestly, by the end of this year, I think the top few large models can basically reach 3.5 level, and there's hope for 4 by next November. But after the launch of OpenAI's GPT-4, the truly commercializable scenarios (in the US) are also very limited. Comparatively, in China, it should be about the same.
Question: Which areas do you think are most promising for commercialization?
Zhu Xiaohu: At this stage, text creation projects are definitely the easiest, like helping to write advertising copy, work summaries, speeches, etc., and they can be perfected quite well. Next is generating photos, such as Midjourney, which is doing very well. In China, we feel generating short videos is even better because China has a lot of short video data, which the US doesn't. Their e-commerce sites, like Amazon, are almost all based on images. In the past three years, China's e-commerce has basically shifted from images to short videos, so today, e-commerce sites prioritize short video traffic, not photos.
China has many short video products and data, which can be used to train their vertical models. So Fancytech's product transformation is going very well.
Question: Is this a sufficiently large investment opportunity and profit opportunity for investors? Image and text generation are still quite niche.
Zhu Xiaohu: We hope that when we start investing, everyone thinks it's a niche market, so big platforms won't pay attention or invest heavily. After a few years, when it grows, there will be more space.
Question: Although AI is hot this year, most people can't get into top projects, and smaller projects are still under observation, so by the end of the year, no one is investing in AI and instead turning to invest in car manufacturing and its upstream and downstream, such as new energy and dual carbon.
Zhu Xiaohu: New energy has also plummeted. I think AI is still non-consensus today, and there are very few investors with real independent thinking ability. In the first half of this year, a bunch of people invested in large models, and the valuations were very expensive. What value does a large model have? Today, every large model is almost the same, each claiming they can score 95.
"Exploring cost-effective projects in non-consensus areas is a very good opportunity right now."
Question: So, are top large model startups still being pursued?
Zhu Xiaohu: They were pursued in the first half of this year, but everyone regretted it in the second half.
Question: For entrepreneurs, is embracing AIGC the only option now?
Zhu Xiaohu: Embracing AIGC is a must. In two years, any company without AIGC will lack competitiveness. This includes consumer and home appliance companies, which need to consider using AIGC to reduce costs and increase efficiency.
Using AIGC for cost reduction and efficiency improvement is very easy now. For example, one of the companies we invested in has more than 1,000 employees. The HR department had four people specifically answering employees' questions about company benefits. Now, by training one robot with AIGC, we can replace the original four people. This kind of robot can be trained internally, which is easy. Just feed the company's benefit policies to the robot for a week of training, and afterwards, it can answer any benefit questions from employees. Other applications, like using AIGC for game image generation, show immediate cost reduction and efficiency improvement.
Talking about Entrepreneurship: "Less Money" is a Double-Edged Sword
Question: The year is coming to an end. How many projects do you anticipate investing in this year?
Zhu Xiaohu: It's about a third of the previous years. It's not that there are fewer projects to invest in; many are still under observation. Currently, I have several good projects that I'm examining to decide whether to invest in them or when to do so.
Before, investors would fund entrepreneurs without even meeting them. That's not happening now. It's like going back to fifteen or sixteen years ago. I can spend three or four months looking at a project. I observe whether the entrepreneur's monthly forecasts match the actual data every month. After observing for three or four months and finding their monthly reports reliable, I am willing to invest.
Question: Have there been any changes or adjustments to GSR Ventures' fundraising and investment pace in the past two years?
Zhu Xiaohu: We completed raising a new fund at the beginning of last year, so it's alright.
In the past two years, our main issue in discussions with LPs is that the secondary market has fallen too much. Many funds have a primary-secondary market mix, and the proportion of the secondary market has fallen a lot, making the proportion of the primary market too high. So there's a serious mismatch between the primary and secondary markets. Under these circumstances, we must wait for the secondary market to recover before rebalancing the primary market.
Secondly, the current risk-free interest rate in the United States is over 5%. Since money in the bank earns an interest of over 5%, why should investors put their money into the illiquid primary market? This affects the whole market, influencing the amount of capital willing to be allocated to the primary market.
The last point is the liquidity issue. The money that prefers investing in the Chinese market is still locked up and not cashed out, so it can't be reinvested.
All these factors limit the circulation of funds. Like many RMB funds, they haven't seen a DPI (Distributed Performance Indicator), so the money can't return, making it difficult to raise new funds. The same goes for the dollar. It needs to see that money can flow back into China, allowing some good companies to exit, and the money can return to China. So I feel that the market won't start to warm up until at least after 2025.
Question: From this perspective, does the current environment pose many challenges for entrepreneurs?
Zhu Xiaohu: Yes, "less money" is a double-edged sword. It's a good thing for genuine entrepreneurs because there's less noise and fewer competitors. However, companies aiming for VC investment definitely can't get funding now.
Question: Are you giving more advice to entrepreneurs this year than in previous years?
Zhu Xiaohu: Companies that listened to our advice last year are doing very well.
Question: What was your advice?
Zhu Xiaohu: Last year it was about optimizing to reduce costs and increase efficiency. Companies that did so are making a lot of money this year, not a single one isn't profitable. Many entrepreneurs listened to the advice. They were very disciplined, so they are profitable this year.
"The coldest time is not winter, but the thawing spring." This year is like the thawing spring. If entrepreneurial companies were prepared last year and reduced their costs, they are doing well this year. Those who didn't lower their costs last year are struggling this year and have to keep reducing costs.
Question: Having invested for so many years, do you feel anything different about this year?
Zhu Xiaohu: This year has been the most comfortable for me in investing. It's really like going back to 15 years ago. I can take my time to see if the cards are really good.
Question: RMB LPs are quite active in the primary market this year. Does GSR Ventures plan to scale up its RMB funds?
Zhu Xiaohu: That's why discipline is very important. Scaling up now is the best way to self-destruct.
I need to reasonably control the scale, the pace of investments, and the valuations. This is especially important.
Question: How do you think institutions performed this year in terms of exits?
Zhu Xiaohu: I think the core is to look at profits. In the coldest times for capital, only when every company is profitable can I sleep at night. Exits are a natural process. As long as every company is profitable, when the U.S. starts to cut interest rates, exit channels will naturally open up.
Question: Your team at GSR Ventures seems a bit unconventional or special in the current primary market, not particularly anxious.
Zhu Xiaohu: Firstly, we basically don't have projects from the peak period. Although I invested during the peak, I still need to reflect on that; but compared to peers, we have very few high-percentage amount investments, so most of our companies are still profitable today. We currently have a company that profits from bank interest. They got too much money during the peak and can earn 5% interest just by keeping it in the bank, so I don't have to worry about whether their company can survive.
At least for me, I'm relatively comfortable now. I didn't enter during the peak, and now in the trough, I can take my time to discover good companies before making a move.
Question: How does GSR Ventures' exit performance this year compare to previous years?
Zhu Xiaohu: Right now, we're looking at one indicator—profit. A few months ago, I reported to RMB investors that in the first half of last year, the top ten companies of the RMB fund lost over a billion, but in the first half of this year, they made a profit of nearly two billion. Everyone is happy to go from a loss of one billion last year to a cumulative profit of two billion this year and can sleep peacefully.
Question: When do you think there might be a peak in terms of exits?
Zhu Xiaohu: The Federal Reserve starting to cut interest rates is a key indicator. Before the dollar starts to cut interest rates, I don't think any large-scale exit is realistic.