The clear paths, hidden paths, and bumpy paths of AI financing

Text | Shi Jiaxiang

Editor | Liu Jing

Before DeepSeek and Manus became popular, the domestic AI industry almost endured a harsh 2024.

A report released by data analysis firm Dealroom shows that in the past year, the venture capital investment in the AI field in the United States was $80.8 billion, more than ten times that of China. Qu Kai, the founder of 42 Chapters, described the primary market before September 2024 as “basically stagnant water.” 

Meanwhile, top - tier US dollar funds are only willing to use the club deal approach to allocate funds to a handful of high - potential entrepreneurs. 

As a result of the chain reaction, the financing cycle for startups is shortening, while the time to complete a single round of financing is lengthening. The founder of an AI company once said that it is now more suitable to raise funds in a “quick, simple, and efficient” way. 

This has created a strange financing phenomenon for AI in the primary market: “The first round of financing comes from friends or the most active US dollar funds, followed by relevant industrial players and CVCs. Since there are no other obvious sources of incremental funds, state - owned capital has entered the market more rapidly than before.” A state - owned capital insider told us bluntly that state - owned capital may be the most active investor in the market now. 

Besides the price, another reason why investors are reluctant to invest is “When will it be profitable?” When an AI application becomes popular, entrepreneurs or investors always question the final commercialization effect of the project, saying “the token cost doesn't add up.” An investor even believes that “if we exclude the investment in human resources and only rely on the product, it will be difficult for these projects to raise the second or third round of financing.” 

Never before has the commercialization requirement for startups been so strict (of course, that was in an era when capital was abundant), so much so that when being interviewed by us, entrepreneurs would reflexively mention “ARR” and “achieving PMF.” 

The rise of international barriers and the squeezing out of capital bubbles have also forced these AI entrepreneurs to act cautiously. “Every round of financing is a choice,” an AI entrepreneur who has raised four rounds of financing told us. The nature of the previous round of investors may affect the success of the next round of financing.

The density and quantity of talent, combined with the endless computing power costs, have made this entrepreneurship particularly difficult. “If continuous financing cannot be obtained, the previous accumulation will become a burden,” an AI industry insider analyzed. 

This industry, which has the potential to rewrite the future destiny of humanity, is full of thorns in terms of financing - and financing is precisely the battlefield where startups need to act quickly and decisively.

In an era when AGI is being loudly advocated and a large amount of capital is needed, it has encountered a cycle with the least amount of money and the highest cost.

So, for AI industry entrepreneurs, where can the money come from? 

The Clear Path: Embrace State - Owned Capital

In early November last year, an announcement by Visual China revealed the equity structure of Zhipu AI to the public. Among the “Six Little Tigers,” Zhipu AI's investors include a large number of Tsinghua - affiliated entities and state - owned capital. Although Zhipu AI had a reputation of “only raising RMB” in the market before, such a “mosaic” structure with up to 40 participating entities was still astonishing. 

It is also the first company invested in by the Beijing Artificial Intelligence Industry Investment Fund since its establishment. 

Other large - model companies are also generally following a similar path. In the Series B financing of Jieyue Xingchen, Shanghai state - owned capital is involved; MiniMax has a close relationship with Shanghai State - owned Investment Corporation; Baichuan has tied up with the governments of Beijing, Shanghai, and Shenzhen at the same time. 

To some extent, the entry of state - owned capital into large - model companies is like a status symbol. Currently, except for Kimi, all of the “Six Little Tigers” have received state - owned capital investment to varying degrees. 

In the AI wave in Silicon Valley, it is the technology giants that have squeezed out venture capital institutions. In China, besides the big tech companies, it is the state - owned capital. 

If it is understandable for large - model companies, which serve as infrastructure, to accept state - owned capital, then a group of AI application companies that were originally supposed to target consumers and enter the market are also frequently receiving support from state - owned capital. 

Beijing established eight highly - regarded government industrial investment funds last year, with a total scale of hundreds of billions. Guo Chuan, the deputy secretary of the Party Committee, director, and general manager of the State - owned Assets Supervision and Administration Commission, revealed at the end of last year that the eight funds had made investment decisions for 167 projects, with a total investment decision amount of approximately 17 billion yuan. Among them, since its establishment, the Beijing Artificial Intelligence Industry Investment Fund has invested in more than 30 AI companies. 

According to “AnYong Waves,” an AI generative application company is close to completing a new round of financing, with the investors being the Beijing Artificial Intelligence Industry Fund and a Beijing state - owned enterprise. However, six months ago, they were still targeting the overseas market for commercialization. Two well - known video - generation companies that have received investment from the Beijing Artificial Intelligence Industry Fund are Aishi Technology and Shengshu Technology. 

An FA who is familiar with state - owned capital told us: “The market conditions in the primary market in 2024 were more than halved, and Beijing state - owned capital has a good reputation and endorsement.” In short, there is no other choice. 

“Doing business entirely for the government may be a dead end, but 'to Gvc' is a way out,” a US dollar equity lawyer commented. Many governments regard AI as a new target for investment promotion, just as they pursued cloud services in the past. 

It is worth noting that the state - owned capital willing to invest in AI in advance is almost limited to Beijing, Shanghai, and Shenzhen. “Even Guangzhou is not included,” data provided by IT Juzi also proves this: In 2024, the number of AI financing events in Beijing, Shenzhen, and Shanghai were 213, 123, and 110 respectively, far exceeding those of other cities. 

Moreover, due to the difference in currency, it is difficult for state - owned capital to make large - scale early - stage investments like US dollar funds. The angel projects they support are generally centered around local universities and research institutions, with an orientation towards talent, technology, and scientific research, and follow a standardized investment process. 

“Now, getting money from state - owned capital is similar to getting it from BAT in the past. What's important is not just the money itself. Entrepreneurs should not only focus on the money but also consider policies and money together,” a partner of an FA firm familiar with state - owned capital told us. In the era of US dollar investment, the most important factors were money, growth, products, and the commercialization system. But in an industry era like large - model development that requires infrastructure, more important are the upstream and downstream industrial support, mergers and acquisitions, and policy support - which the government can provide. 

Of course, every gift comes with a price. Different state - owned capitals have different demands. If a company accepts investment from multiple state - owned capitals, it needs to balance the interests of all parties. “Often, the location of the company's headquarters depends on which place offers the most support.” Companies should not act until they see clear benefits - start implementing part of the business after receiving part of the funds. 

Almost all the AI companies that have received investment from the Beijing Artificial Intelligence Industry Fund have offices in Beijing. A relevant person told “AnYong Waves” that an AI company that received financing from the Beijing Economic - Technological Development Area will move its office to Yizhuang. Hangzhou Urban Construction Investment Industry Fund participated in the investment of Zhipu AI, and Zhipu has also established a branch in Hangzhou. 

The aforementioned FA told us that some state - owned capitals value the actual output value on the ground more, pay more attention to downside risks - financial security and compliance, and care less about valuation. Therefore, companies should focus more on the absolute amount of funds received. 

The Hidden Path: International Funds and “Going to Silicon Valley”

Where can the money come from? This is the question that Austin, the co - founder of Sapient, has been thinking about from the beginning. Sapient is a large - model company that attempts to challenge the Transformer architecture.

When Austin started raising funds for his startup last year, the pattern of large - model startups in China had already been established, and the business they were exploring required hundreds of millions of US dollars. At an inappropriate time, facing venture capital institutions that had already made their investments, it was obviously extremely difficult. 

He remembers that when he first met with investors, the most frequently asked question was: “Why should I invest in you instead of Yang Zhilin?” “Chinese VCs have no preference for technological innovations that have not been widely verified in the United States. They require PMF and commercialization,” Austin said. They also sought funds from the Middle East and Europe. 

The answer to the question lies in Singapore and Japan. In December last year, Sapient received tens of millions of US dollars in financing from well - known overseas investment institutions such as Singapore's Vertex, Japan's largest private equity investment group JAFCO, and Sumitomo Corporation at a valuation of over two hundred million US dollars. Austin summarized that there are always opportunities in the market and there are always investors who are willing to believe. 

Sapient's financing story is a typical case of a Chinese - founded company raising funds globally. 

And Sapient's path is also the common portrait of most entrepreneurs today: Chinese founders, targeting the global market from day one, most employees based in China, registering a company in Singapore or other countries, paying salaries through labor outsourcing agreements, and leveraging the cost difference between Chinese engineers and those in Silicon Valley to gain an advantage at the starting line.

In the past, this was within the scope of typical US dollar funds. But today, the implementation of venture capital laws and regulations has hindered this investment chain. So, many people are also looking towards a deeper pocket: Silicon Valley. 

However, it is very difficult to actually obtain investment from Silicon Valley. A lawyer familiar with overseas financing told us that currently, three conditions - the company's structure, the founder's nationality, and physical presence - need to be met simultaneously to get money from Silicon Valley. “Only a few founders can succeed.” 

It is generally believed that such founders are either serial entrepreneurs in the United States or have worked in well - known large US companies and maintain close contact with investors. “Otherwise, it is very difficult to get the first - round US investment,” a partner of a US dollar fund told us. 

Successful examples include AI video - editing tools such as OpusClip and Pika. However, Zhao Yang, the founder of OpusClip, started his business in the United States as early as 2013, and only three of Pika's founders are Chinese, with most of the other employees being Americans. 

A more typical example is the AI video - generation company HeyGen, which increased its ARR from $1 million to over $35 million in just over a year. This company, founded before the GPT era, received investments from Sequoia China Seed Fund, ZhenFund, and IDG Capital in 2021 and completed a $60 million Series A financing led by Benchmark at a valuation of $500 million in June last year. 

When talking about HeyGen, most entrepreneurs believe that they gave up their domestic popularity and “some valuation advantages” and “made a timely decision.” The founder of an AI application company said to us enviously that his company's revenue is similar to that of HeyGen, but HeyGen's valuation is ten times that of his company. 

Following entrepreneurs to Silicon Valley are also Chinese US dollar funds, but they also face new barriers. 

According to the technology media The Information, at the beginning of last year, both HeyGen and OpusClip relocated their Chinese engineers to Canada and Singapore. Lightspeed China, an early investor in OpusClip, and Sequoia China, an early investor in HeyGen, also withdrew from the boards of directors one after another. There were even rumors that they would be forced to sell their old shares. 

A lawyer familiar with the Silicon Valley ecosystem explained to us that Silicon Valley funds are forced to do so. “Otherwise, they will be frequently disturbed by the US government.” “Chinese entrepreneurs in Silicon Valley are also very concerned about the source of funds and the registration location,” an investor told “AnYong Waves.” 

However, this does not seem to be the biggest obstacle. An investor once told us that a US dollar fund invested in an AI company, which was also favored by a Silicon Valley fund. “The partner was happy to withdraw.” “What's important is to invest in good assets. If you are asked to leave after an early - stage investment, negotiate a reasonable price. I think it's not a bad deal,” a partner of a US dollar fund told “AnYong Waves.” 

But for pure Silicon Valley founders, the advantages of US dollar funds are not obvious. Wang Huai, the founding partner of Linear Capital, believes that “a core entity without the support of Chinese capabilities has no special competitiveness overseas except for having some money.” 

So, the final reality may be that Chinese entrepreneurs in the United States will take the money from these overseas - going US dollar funds - a compromise for both sides.

Some people have taken a more radical approach. An AI education company was once regarded as a model for going global. But the founder told us that they are not a company going global; they only target the US domestic market and are a pure US company. 

This founder was a successful serial entrepreneur in China, but in the United States, he is an unknown figure at an older age. “It takes a great deal of determination to start over in Silicon Valley and abandon everything in China.” He also successfully obtained investment from Silicon Valley. 

This path is also full of difficulties. 

The Rocky Path: CVC, Mergers and Acquisitions, or “Drip - Irrigation Capital”

CVC may be an incremental source of funds.

For example, behind the “Six Little Tigers” of large - model companies, there are Internet giants to varying degrees - especially Alibaba, which has heavily invested in Yuezhianmian. 

The same is true for AI applications. For example, Jingying Technology, which launched the world's first AI short drama, was invested in by Baidu. AI video - generation companies Aishi and Shengshu Technology, as well as Mitata Technology, which focuses on search, all have Ant Group as their backer. We also learned that an AI game company valued at hundreds of millions has received financing from Ant Group. The latest news is that Tencent led the investment in the Manus development team, Monica. 

The advantages of CVC are obvious: sufficient funds and no policy dilemmas. But the disadvantages are also clear: as they are in the same business upstream or downstream, there is direct competition with startup companies. The founder of an AI application company based in Shenzhen told “AnYong Waves” that he had rejected several visits from Tencent - related personnel. 

The rapid entry of large companies has also increased the possibility of startup companies being acquired. Two AI acquisition cases have attracted attention: OPPO's acquisition of Waveform Intelligence and Ant Group's acquisition of Biansai Technology. The common point of the two cases is that the early - stage investors completed their exits and the companies continued to operate independently. “It was difficult for Biansai to achieve commercialization, but from a pure technical perspective, Wu Yi has his own unique insights,” an Ant Group strategic investment insider explained. 

A person close to the transaction analyzed that Ant Group's logic in acquiring Biansai Technology was to gather technical talents, while OPPO's purpose in acquiring Waveform was for internal business exploration. Such acquisitions usually do not offer a price based on valuation but rather return the principal + interest to the investors, trying to let them break even. At the same time, the founders and technical personnel are given stock incentives and are required to stay for at least four years - just like the case of Character.AI being acquired. 

There is also the case of Lingyi Wanwu, which was rumored to be acquired by Alibaba. Li Kaifu refuted the rumor, saying that Lingyi Wanwu and Alibaba established an industrial large - model joint laboratory. “Those who want to work on large - scale cluster infrastructure and training will join the joint laboratory and become Alibaba employees.” 

In the unpredictable AI application market, Wang Huai once pointed out to “AnYong Waves” that the AI application market will not be a winner - takes - all situation, and burning money to gain market share may not work. 

Wang Bolong, the founder of Huidu, quit his job at a large company to start a business due to the AI wave. After selling his company, he joined a traditional financial company with his team. He