Is investing in a big model a one-time solution? More opportunities may lie here, capital | entrepreneurs | models
On the path of artificial intelligence evolution, capital support and companionship are indispensable. On July 7th, at the 2023 World Artificial Intelligence Conference Investment and Financing Theme Forum, a heated discussion on AI investment opportunities took place.
The concept of AI has been very popular in the capital market recently, and Wang Ying, a fund manager of China Europe Fund, believes that this is the result of several factors combined - AI is an industry change that cannot be ignored, especially the rapid changes around it, which has a strong impact on investors. Meanwhile, growth assets are scarce in both A-shares and US stocks, and a major industrial transformation like AI will give rise to many growth assets. In addition, in this year's market environment, the certainty of the AI industry is higher, and investors will have a greater investment risk preference in this field.
Amidst the heat, some calm voices are needed. In the opinion of Zhang Yaqin, a professor of Tsinghua University, president of the Institute of Intelligent Industry, and academician of the CAE Member, AI today is a bit like the Internet that just started in 1998. Opportunities and foam coexist. Just as the Internet foam burst in 2000, AI may also go through this stage. But the Internet itself is not a foam. When the Internet changes the way of life and work, real powerful and valuable companies will emerge. The same goes for AI. Many amazing companies will surely emerge in the wave, but there will also be some companies disappearing in this process.
In view of this, from an investment perspective, Zhu Yunlai, former President and CEO of CICC and visiting professor of management practice at Tsinghua University, also suggests that there should be a systematic focus on the AI industry. For AI, not investing is a failure, but investing is not a worry free situation. Since the entire industry is currently exploring, it cannot be guaranteed that everyone will succeed. "A lot of exploration will come up, and a lot of foam will burst again, but eventually in the interaction, the industry will indeed produce systematic progress." Zhu Yunlai explained, "So investors need to constantly adjust their investment, take a place in the industry, and not feel that it is very solid, but adjust their investment strategies at any time in the process."
So, for investors and entrepreneurs, is the currently popular big model in the AI industry still an opportunity?
"Big models cannot be achieved once and for all, and the implementation of specific scene functions needs to rely on the development of vertical fields." Wang Xiao, a partner of Jiuhe Venture Capital, believes that enterprises do not need to squeeze into the track of big models and compete directly with large factories. "Industrializing the implementation of large models is a very important structural opportunity in the next 3 to 5 years."
Zhang Yaqin also holds a similar view, likening a horizontal large model to an AI operating system. With this operating system, there is a need for a vertical precision model for the industry, on which new applications can be developed. Just like mobile phones, there are now platforms such as Android and iOS, and there are a wealth of apps available on each platform. The big opportunities in the future lie in these apps of the big model.
In fact, besides generous and upward opportunities, investors value the entrepreneurs themselves more. Chen Yu, a partner of Yunqi Capital, believes that entrepreneurs favored by capital must possess three characteristics - the first is to have a new knowledge structure that can adapt to this constantly changing era. The second is to have ambition. Previously, AI only solved small problems, but now it is solving larger problems with unclear paths. It needs to have ambition to achieve them. The third factor is whether the financing ability is strong or not. Large models are very expensive, and it is also very important for entrepreneurs to clearly express their vision and make investors believe and agree with it.
But the entrepreneurs present also bluntly stated that in order for the invested companies to stand out in competition, capital needs to have greater patience. Going out to ask CFO Sun Junbo, the pain point of AI companies is that everyone hopes to produce projects within three to five months, but the reality is that many teams can only produce results within three to five years. He believes that the emergence of chatGPT is not accidental. Silicon Valley's capital tolerance is relatively high, and they can accept that the invested company has no products or customers for several years, so that such capital can support ambitious companies to do it. "We hope that domestic capital can have more tolerance, spend more time with enterprises, and be friends with time."