Cai Wensheng: I firmly believe in the explosion of Web3; investors are investing in their own dreams

Respondent: Cai Wensheng, Angel Investor

Organizer: Wanwu Island

On June 17, at the third module of the Wanwu Creation Camp in Hong Kong, renowned investor Cai Wensheng held a 2-hour discussion and sharing session with 65 entrepreneurs from Wanwu Island. As an entrepreneur and investor who has traversed the three generations of the internet—Web1, Web2, and Web3—how has he captured the benefits of these three waves, and how does he view the transformation of the internet? What insights can this provide to Chinese entrepreneurs? Below are some highlights from the Q&A transcript:

Wanwu Island: Why choose Web3 and AI?

Cai Wensheng: In Web2, users continuously click to agree while using a product, and all this consent belongs to the platform. Even if one day you decide to leave the platform, all the data you generated there still belongs to the platform because you agreed to it long ago.

Web3 aims to solve this problem; any open Web3 platform cannot prevent you from registering, and all your accounts and assets belong to your own wallet address.

The concept of AI was popular a few years ago, and everyone thought AI would change the world. However, before ChatGPT emerged, people were quite confused because the applications of AI had reached a dead end. Why did ChatGPT excite everyone? We can compare AlphaGo and ChatGPT. Lee Sedol had an excellent memory from a young age and learned hundreds of thousands of ancient game strategies, so he immediately knew how to respond to others' moves. However, AlphaGo could learn more game strategies in a short time. This was the problem with earlier AI; it first taught, then learned, and played to continuously train a model. If you wanted it to do something else, you had to teach, learn, and start all over again.

Now, ChatGPT starts off knowing nothing and does not specifically teach it content from a particular field. It continuously exercises its logical thinking, similar to teaching a child. It first builds a model and develops logical abilities before teaching content, based on this underlying logical model, thus achieving general applicability.

AI is more about the development of productivity, while Web3 is a redefinition of production relationships. The reason Web3 has been controversial from the start is due to the change in production relationships, which aims to alter existing company structures and financial systems. Web3 and AI may complement each other; OpenAI is a great example. In 1600, the East India Company was established in the Netherlands, greatly propelling the world, making the Netherlands a world power. Much of the Netherlands is below sea level, unable to engage in agriculture or manufacturing, so it relied on trade. A large number of merchant ships went east to earn huge profits, and just three years later, in 1603, the world's first stock exchange was established to allow more people who couldn't board ships to share in trade profits, while also using capital leverage to expand merchant shipping.

The corporate system that has existed for hundreds of years is no longer suitable. For example, the share structure of some large internet companies has already been fully allocated. If these companies continue to survive without selling stocks to others, does that mean we will work for these companies our entire lives? This is unreasonable.

So how to break this? We spend so much time on social media, yet the platforms sell us ads, and the money goes to a few major shareholders, with no relation to the users who contribute this data. Therefore, in the era of Web3, users who contribute data should be rewarded. The challenge lies in how to reasonably quantify the corresponding contribution behaviors; the reward system needs to be a key focus. Thus, whoever can establish a more reasonable and sustainable distribution reward rule for Web3 will bring significant innovation.

The improvement in productivity brought by AI is already visible. Many designers have been replaced, and programmers may be next. AI is reforming various industries, but some worry that with so many people unemployed, the world will become worse. I believe this is an overreaction. Just like the shift of the manufacturing industry in the United States in the 1970s, many people lost their jobs, but they naturally transitioned to the tertiary sector, information, and service industries. The AI revolution will eliminate many existing jobs, but I believe new jobs will emerge. Perhaps these jobs will involve watching movies, playing games, or hiking—entertainment that we previously thought required spending money may become work in the future.

In traditional games, perhaps only a few dozen players contribute 80% of the revenue, while the rest are just there for fun. This is unreasonable; these players, while enjoying themselves, do not earn money. They should be compensated because they are also important contributors to the game ecosystem.

Parami Protocol: Will the platform incentivizing users through tokens be driven by mature large companies using part of their revenue and shares to change the status quo, or will it be the native Web3 startups that disrupt these platforms?

Cai Wensheng: Historically, the birth of a new model often comes from innovative companies; it is difficult to emerge from giants because they have too many existing structures and vested interests. For example, if a CEO of a giant wants to do something, other shareholders and the board may not agree. The three major portals in China did not succeed in creating search engines, even though their traffic far exceeded that of early search startups. Why did ChatGPT emerge from OpenAI? Google has been working on it for so long, and even giants like Apple have not reached the level of ChatGPT. History has proven that it is very difficult to create something new on an old structure.

Saame Protocol: As a Web3 entrepreneur, many traditional investors often do not understand what you are doing. How do you solve this problem?

Cai Wensheng: This issue existed in Web2 and Web1 as well. Entrepreneurs like Jack Ma, Pony Ma, and Zhang Yiming faced many doubts from investors. If they truly understood, it would not be reasonable; they would quickly tell previous projects to do the same. So, it is normal for investors not to understand a new startup project. However, if you still want to raise funds from them, you should try to speak in terms they can understand. Investors are not investing in the entrepreneur's dreams but in their own dreams. They will invest only if they believe what you are doing is changing the world. Therefore, you need to let them know that what you are doing can help them achieve their grand dreams, rather than trying to force them to pay for your dreams.

When meeting investors, you generally have 45 minutes, so you must clearly state what you want to do, what you have already accomplished, and you must not lie about the data of completed projects. For future endeavors, you can be as logically coherent and ambitious as possible to spark the investor's imagination. The best way to converse is not to keep talking but to let the other party ask questions.

CHE DAO: There aren't many Web3 users right now, and a large number of users are still in Web2. This may rely on large Web2 platforms to guide users into Web3. Do you think the explosion of Web3 will come from entrepreneurs or be guided by large companies?

Cai Wensheng: I firmly believe in the explosion of Web3; it is just a matter of timing, which is hard to predict. As mentioned earlier, AI also went through many years of dormancy before the explosion of ChatGPT. Regarding the issue of Web2 users entering Web3, the centralized benefits of large platforms are that they retain users, and the cost for users to leave is high due to the accumulation of a lot of user information.

However, if we look at Web3 products like MetaMask, users do not have data accumulated within it but still exhibit high stickiness. This is because it does not retain users through data; users can transfer their assets to another wallet at any time. It retains users through usage habits and better experiences from internal third-party collaborations. The question of how to transition Web2 users to Web3 is fundamentally not an issue. Just like when the first generation of the internet emerged, there was no effort to pull traditional people in. As long as something new emerges, there will naturally be users who accept it.

Cell Protocol: One potential issue with tokens is that they may attract a large number of low-quality users who exploit the system, even driving out good users and affecting normal usage. How do you view this problem?

Cai Wensheng: This problem also existed in Web1 and Web2. Just like many people used to run QQ bots, we can use basic product functionality to limit such behavior. Of course, Web3 can use rewards to restrict it. Most projects make the mistake of rewarding users too much at the beginning, leading to unsustainable follow-up. Therefore, the release of rewards and the mechanism model must be well designed to avoid this issue to some extent. However, it is also important to note that too clean a mechanism may not necessarily be beneficial to your ecosystem; it is about finding the right balance.

Createra: What advice do you have for teams that have made good accumulations in Web2? What model of Web3 startup teams do you find promising?

Cai Wensheng: In Web1 and Web2, games were the first to rise, followed by social media, e-commerce, and other applications. The reason Web3 games have not developed on a large scale is that Web3 is a financial system, which is also its biggest challenge. How to ensure that games remain stable and sustainable within this financial system is a significant problem. Once this financial system stabilizes, I believe the next step will be for games to rise first, followed by social media and other fields.

Returning to AI, ChatGPT now is like iOS and Android; we missed the operating system of Web2, but we developed many applications on top of it. Our strength as Chinese people lies in creating applications; building the underlying systems is really challenging. Therefore, my suggestion is to develop upper-layer applications based on large models like ChatGPT, which may provide an opportunity for a leapfrog development.

Wanwu Island: What is your dream?

Cai Wensheng: Dreams are constantly changing. I posted a Weibo in 2010, and out of the three dreams, the first two have been realized. The first dream was to invest in a company worth 10 billion USD, which was achieved by 58 and Meitu. The second dream was to travel to 100 countries, which I have basically completed. The third dream is to write a book because my experiences are quite unique; I only studied until the first year of high school but then started working in the technology industry. So I hope to write a book to summarize my experiences.

Recently, I added a new dream related to entrepreneurs. Garage Cafe in Beijing has greatly promoted the development of blockchain in China. I found that the cost of starting a business in Hong Kong is quite high, so I am considering finding a venue in Hong Kong to provide free space for early-stage entrepreneurs to communicate and support young people in starting their businesses.